$AEHR

@jasonl_capitalMay 24, 2026 · 11:32 PM

9. $AEHR - Aehr Test Systems Every chip in this thread has to be tested before deployment. Aehr builds the burn-in and reliability test systems for AI processors, silicon photonics, and power semiconductors. Stock up over 200% in 2026. Expecting follow-on production orders from its lead hyperscale customer for custom AI processor testing.

@thevalueistMay 24, 2026 · 1:43 PM

$AEHR $SNDK $MU $000660 $005930 EXECUTIVE SUMMARY AEHR’s HBF narrative has moved through 3 distinct phases. In Q1 FY26, HBF was introduced as an emerging AI memory architecture that could structurally increase the need for high-power wafer-level test and burn-in. In Q2 FY26, the discussion became customer-specific: a global NAND leader had completed an initial wafer-level benchmark, HBF had become a major reason the customer’s existing infrastructure was inadequate, and AEHR had proposed a new HBF-capable solution based on FOX-XP, WaferPaks, and automation. By Q3 FY26, the story had advanced technically but remained pre-commercial: AEHR said it had achieved the correlation the customer wanted, was now discussing test-system specifications for next-generation flash and HBF, and hoped to close a development agreement within the next few months, followed by a 12-18 month development program for memory-optimized blades for FOX-XP and FOX-NP. The most important investment conclusion is that HBF has become a credible strategic option for AEHR, but it is not yet a revenue ramp. The Q3 language was more technically de-risked than Q2, but also more explicit that orders would likely be FY27 and ramps FY28, making the HBF opportunity a medium-duration call option rather than a near-term bookings driver. The commercial signal strengthened QoQ because AEHR moved from “benchmark completed and proposed solution” in Q2 to “correlation achieved and specification discussions under way” in Q3. However, the timing signal became more sober. In Q2, management framed customer feedback and a potential order decision as something that could emerge over the next few months; in Q3, the expected next step became a development agreement that would then require 12-18 months of product development before system and WaferPak supply. That is not a negative read-through if the investment debate is 2028-2030 earnings power, but it is negative for any thesis requiring HBF to contribute materially to FY26 or early FY27 revenue. The opportunity is no longer just “can AEHR test HBF?” It has become “can AEHR define a memory-specific FOX architecture that scales across next-generation flash, HBF, DRAM, and eventually HBM?” The HBF market implication is material. HBF is not simply a faster SSD; it is being positioned publicly as a new memory tier between HBM and SSD storage, targeted primarily at inference workloads that require high capacity, high bandwidth, and better cost-per-capacity than DRAM-based HBM. SanDisk and SK hynix announced an MOU in August 2025 to establish an HBF specification and stated that HBF targets comparable bandwidth to HBM while delivering 8-16x HBM capacity at a similar cost; they also targeted 1st HBF memory samples in 2H26 and 1st AI-inference devices using HBF in early 2027. In February 2026, the 2 companies formalized HBF standardization work under the Open Compute Project, describing HBF as purpose-built for inference at scale. Kioxia separately demonstrated a 5TB, 64GB/s flash-memory module using PCIe 6.0 and <40W, showing that the broader NAND industry is exploring multiple routes to move flash closer to the AI memory hierarchy. SOURCE BASE AND TIMELINE The uploaded source set contains AEHR’s Q2 FY26 earnings call, the January 2026 Needham Growth Conference presentation transcript, and AEHR’s Q3 FY26 earnings call. External transcript and industry sources were used only to establish Q1 FY26 baseline commentary and broader HBF market context. The sequential picture is important because HBF was not a static talking point across the calls. It entered the AEHR discussion as an adjacent NAND and AI-memory concept, then became the central reason a memory customer’s existing test infrastructure was insufficient, and then became part of a broader memory-platform roadmap that includes standard flash, HBF, DRAM, and HBM. Q4 FY25 is useful as a pre-HBF baseline. The company discussed a flash-memory proof-of-concept with a global flash leader, a MEMS-based fine-pitch WaferPak, and a next-generation system opportunity for high-volume production wafer-level test and burn-in of flash wafers. The issue was framed as NAND technology driving new wafer-level burn-in requirements and as a benchmark delayed by component shipments, not yet as an HBF-specific customer-pull opportunity. HBF itself was not found as an explicit term in the Q4 FY25 transcript reviewed, while the Q4 commentary focused on NAND flash, high-density/high-power testing, and the need to avoid the manufacturing and yield penalties of later-stage test. Q1 FY26 marked the 1st explicit HBF framing. AEHR described NAND flash as being in “flux,” with earlier hybrid-bonding requirements for higher-density NAND on 300mm wafers now shifting toward HBF, creating very different test-system requirements. Management said both hybrid-bonded flash and HBF were driving power requirements up substantially, which it characterized as directly aligned with AEHR’s strength. AEHR also described HBF as an emerging technology from 2 flash leaders that aims to create a massive-capacity AI memory tier by combining HBM-like packaging with 3D NAND flash, potentially offering 8-16x HBM capacity at similar cost with comparable bandwidth. This was a major shift from generic flash WLTBI toward an AI-specific memory-wall solution. The Q1 Q&A added a critical reliability angle. AEHR stated that HBM and eventually HBF need burn-in and stress cycling; otherwise, latent failures would appear inside AI stacks. Management framed the test industry as “scrambling” for approaches to these new power and reliability challenges. AEHR also emphasized the magnitude of its power-density differentiation: management referenced 3.5kW per wafer, contrasted conventional wafer probers at roughly 300W, described specialized single-wafer probers at 1,500-2,000W, and said AEHR systems could deliver 3,500W on each of 9 wafers in 1 machine. This established the core HBF equipment thesis: HBF’s value proposition depends on massive parallelism and high wafer power, and conventional probe infrastructure may be structurally underpowered. Q2 FY26 converted the HBF story from concept into a named development vector. AEHR said it had completed a wafer-level benchmark with a global NAND leader just before the holidays and that the customer had taken the wafers back for further processing to validate correlation with its internal process. Management claimed the benchmark demonstrated the ability to test flash-memory wafers with significantly higher parallelism and power than traditional probers and group probers from TEL or ACCRETECH. AEHR also said it had proposed a next-generation solution to test a new emerging flash memory device called HBF, designed for AI workloads. That proposed solution would use FOX-XP, WaferPaks, and auto-aligner technology to support 1-touchdown, high-power test on 300mm wafers, but development would take >1 year after customer commitment. The Q2 Q&A was the most revealing HBF discussion in the entire source set because it explained why the memory project had slipped and why HBF matters. AEHR said the original customer engagement was not aimed at HBF because HBF did not exist in the customer’s planning when the benchmark began; instead, the starting point was commodity/data-center SSDs and then hybrid-bonded flash. According to management, HBF “broke” the customer’s existing infrastructure across power supplies, I/O pins, and parallelism, creating a new power problem. AEHR then distinguished hybrid-bonded flash from high-bandwidth flash: hybrid-bonded flash places logic on a logic process and stack memory on a memory process, then bonds them to create a taller memory stack with a smaller footprint and more die per wafer, while HBF is architecturally similar but has even higher power because speed adds additional power supplies and because the device is taller. Q2 also disclosed that HBF forced a tester redesign. AEHR said HBF was “even more of a problem” than the prior target device and required the company to redesign the tester because the original design was aimed at the earlier device. Management also stated that the customer had shifted from an enterprise-flash focus to HBF, which slowed both the benchmark and the tester review process. This point is crucial for interpreting the timeline: the delay was not presented as pure execution failure; it was presented as a customer roadmap change that increased the technical and commercial relevance of AEHR’s platform. However, it also showed that HBF requirements were still evolving and that AEHR was not yet at a production-ready memory blade architecture. The January Needham presentation sharpened the message further, though it should be treated as investor-presentation language rather than a signed commercial commitment. AEHR said it had engaged with multiple flash-memory companies, recently completed the 1st phase showing full-wafer contact and test/burn-in of a 300mm NAND flash wafer, and used that result to evaluate FOX-XP and proprietary WaferPak contactors for new HBF flash devices. The slide stated that the application was for 100% test and burn-in of HBF memory devices for mission-critical applications including AI LLM inference, automotive, and industrial robotics. This was stronger language than the Q2 call because it explicitly linked HBF to 100% production screening and mission-critical applications, but the formal call language still made clear that a new solution would require development after customer commitment. The Needham presentation also framed HBF as part of a broader memory TAM expansion. Management said new technologies like hybrid-bonded flash and HBF were doubling parallelism and quadrupling power per wafer, and it emphasized that AEHR could handle more power per wafer and more than 1 wafer at a time. It also stated that device-market annual revenue can translate into 2%-5% capital spend on overall test, and that memory and AI processor markets dwarf AEHR’s legacy silicon photonics, silicon carbide, and gallium nitride device-market targets. This was the most explicit attempt to link HBF to AEHR’s long-term TAM expansion, although no HBF order was announced. Q3 FY26 provided the clearest evidence of technical progress. AEHR stated that its engagement with a key memory supplier continued to progress, that additional wafer testing had occurred just the prior week, and that it had achieved the correlation the customer was asking for. It then said discussions had moved to test-system specifications for next-generation flash memories and specifically HBF devices. Management hoped to close the next step within a few months, which would lead to a development agreement to supply systems and WaferPaks after a 12-18 month development effort for new memory-optimized blades for FOX-XP and FOX-NP. This is a technically positive but commercially staged update: correlation was a meaningful milestone, but AEHR still needed a development agreement, then product development, then production orders. Q3 also broadened the memory roadmap materially beyond HBF. AEHR said it was now in discussions with other key memory suppliers that produce HBM, standard DRAM, and flash. Management referenced NVIDIA’s HBM roadmap from HBM3E to HBM4, HBM4E, and HBM5, and said AEHR’s FOX multi-wafer test and burn-in roadmap extends to flash, HBF, DRAM, and HBM memories. The Q3 framing was therefore no longer “HBF as the next NAND device.” It became “HBF as 1 insertion point in a memory-optimized FOX platform that could eventually address multiple memory categories.” The company said this was a key focus for the year and could drive orders in FY27 with ramps in FY28. The Q3 Q&A made the strongest statement that HBM had become parallel to flash rather than a later-stage adjacency. Management said it had identified interesting HBM opportunities, likely around HBM4E, where customers would like to perform wafer-level burn-in, and that memory extensions to FOX would add “channel modules” to make the platform memory-focused. Management then said HBM had effectively moved in parallel with flash. This matters because it raises the strategic value of the memory blade program: the same development architecture could support HBF, standard flash, and potentially HBM/DRAM, making the project less dependent on 1 HBF product schedule. The Q3 Q&A also changed the risk framing. When asked whether the flash engagement could bear fruit on enterprise products before HBF gets under way, management said the timing was up to the customer and that the system definition was intended to be a “superset” capable of doing both. Management said that if HBF were delayed, AEHR might intercept standard products. Importantly, it also said HBF is in some ways easier than broad legacy flash because a broad flash system can become burdened by requirements to test legacy voltage and speed interfaces that customers may no longer economically need. That implies AEHR is trying to avoid building an overly broad, backward-compatible memory tester and instead wants customers to define a forward-looking specification around where capital will actually be spent. QOQ CHANGE IN AEHR’S HBF MESSAGE The Q1-to-Q2 change was from conceptual market education to concrete customer evidence. In Q1, HBF was presented as a technology trend: 2 flash leaders were developing it, it promised 8-16x HBM capacity at similar cost, and it would stress conventional wafer test infrastructure through higher power and parallelism. In Q2, AEHR moved to customer-specific execution: a global NAND leader had completed a wafer-level benchmark, wafers were being returned for correlation, and AEHR had proposed a 300mm, 1-touchdown, high-power HBF solution. This was a meaningful upgrade in evidence quality, but it also introduced a more explicit product-development dependency because AEHR said the system would take >1 year to develop after commitment. The Q2-to-Q3 change was from “benchmark and proposed solution” to “correlation achieved and specification discussions.” This was the most important technical de-risking step. In semiconductor test, correlation is not a trivial milestone; it determines whether customer-side internal data and AEHR-side wafer-level results are sufficiently aligned to justify development of a production-worthy flow. By saying correlation was achieved, AEHR reduced the risk that the benchmark was merely a lab demonstration. However, the Q3 update also showed that the commercial process had not yet reached a PO or even a signed development agreement. The project progressed technically but remained pre-order. The Q2-to-Q3 commercial timing moved out rather than in. In Q2, Q&A language suggested customer feedback and potentially an order could come in “the next couple of months.” In Q3, the update became “hope to close in the next few months,” followed by a 12-18 month development program and possible FY27 orders/FY28 ramps. The distinction matters. Q2 sounded like a potential near-term order catalyst; Q3 reframed the opportunity as a structured development program. This is more credible but less immediate. It lowers the probability of a surprise near-term HBF revenue contribution and increases the importance of development-agreement economics, engineering milestones, and FY28 capacity planning. The technical scope expanded materially QoQ. In Q2, AEHR was focused on HBF as a specific next-generation flash device. In Q3, the roadmap explicitly included flash, HBF, DRAM, and HBM, with memory-optimized blades and channel modules becoming the core development architecture. This makes the opportunity strategically larger but technically more complex. The upside is platform leverage across multiple memory markets. The downside is specification creep, longer validation, higher R&D intensity, and the possibility that no 1 memory supplier wants to fund a generic solution that serves other suppliers unless the system is highly customized. The customer set broadened QoQ. Q2 centered on a global NAND leader and referenced the proposed HBF solution. The Needham presentation added that AEHR had been engaged with multiple flash-memory companies. Q3 went further by saying AEHR was in discussions with other key memory suppliers that also produce HBM in addition to standard DRAM and flash. This broadening is positive because it reduces dependence on 1 unnamed NAND customer, but it also suggests that AEHR is still in market-development mode rather than order-execution mode. Broad discussions can increase strategic value, but they are not equivalent to production commitments. The language on HBF difficulty became more nuanced QoQ. In Q2, HBF was described as a device that broke the customer’s infrastructure due to power supplies, I/O pins, and parallelism; it was “more of a problem” than prior hybrid-bonded flash. In Q3, management said HBF could be easier in some ways than broad flash because the interface target is forward-looking and does not necessarily require legacy voltage and speed support. This is not contradictory. It means HBF is harder electrically and thermally, but potentially easier from a product-definition perspective if the customer avoids forcing AEHR to support legacy devices that will not justify new capex. The investment signal improved on strategic relevance but weakened on near-term measurability. Q2 sounded like a specific HBF opportunity could emerge from the completed benchmark. Q3 made the project larger and more credible technically, but also made it clear that the revenue path is contingent on development closure, 12-18 months of blade development, and subsequent production ramp. The best interpretation is that HBF became a higher-quality long-term thesis and a lower-quality near-term catalyst. For an investment committee, the correct underwriting approach is to treat HBF as upside optionality until a development agreement is announced, then begin assigning probability-weighted FY28 revenue and gross-margin contribution. TECHNICAL READ-THROUGH HBF changes the flash test problem from storage qualification to high-power memory-class screening. Traditional NAND test is already difficult at high layer counts, especially as hybrid bonding, more die per wafer, higher current, and higher density drive power and parallelism requirements. HBF adds the AI-memory angle: the device is not just storing data behind a conventional SSD controller; it is intended to serve bandwidth-intensive inference use cases, likely with far more parallel read paths, additional power rails, and a packaging architecture closer to HBM than to conventional SSD NAND. AEHR’s Q2 comments that HBF stressed power supplies, I/O pins, and parallelism are consistent with the public HBF architecture thesis, which depends on creating massive parallelism out of NAND density. The key AEHR technical advantage is not generic burn-in; it is full-wafer, high-power, high-parallelism burn-in with thermal control and automation. The company repeatedly emphasized that conventional probers and group probers are not designed for the same power density or parallel wafer count. AEHR’s FOX-XP can test multiple wafers in parallel, its WaferPak architecture enables full-wafer contacting, and its auto-aligner enables automated handling of 300mm wafers. For HBF, these features matter because the cost of serializing high-power test through conventional probe infrastructure could be prohibitive in floor space, power delivery, contactor cost, and cycle time. The memory-optimized blade is the gating product. AEHR’s existing FOX architecture provides the mechanical, thermal, and automation foundation, but Q3 language shows that HBF requires a new memory-specific blade with channel modules. That implies the current platform is not yet a fully qualified HBF production tester. It can demonstrate contact, power, thermal, and benchmark capability, but a production solution will need memory-specific resources, likely including appropriate channel count, protocol support, timing, voltage domains, data path architecture, per-die monitoring, stress modes, and production software integration. The 12-18 month timeline should be interpreted as real engineering work, not a minor configuration exercise. The HBF test problem is likely more similar to AI processor wafer-level burn-in than to commodity NAND test in 1 key respect: early-life failures must be removed before expensive downstream integration. HBF devices are being positioned for AI inference, automotive, and robotics applications where reliability failures can be costly, disruptive, or safety-relevant. The Needham slide’s “100% test and burn-in” language indicates that AEHR sees HBF as a production screening application, not just qualification. If HBF is packaged in advanced stacks or integrated near accelerators, the economic logic mirrors AI accelerator wafer-level burn-in: identifying bad die at wafer level avoids scrapping more expensive downstream assemblies. However, HBF will not automatically require AEHR’s exact solution. Existing memory manufacturers already have deep internal test flows, incumbent tester relationships, and significant process-control capability. The question is whether HBF’s power and parallelism requirements exceed what internal memory test architectures and incumbent prober/test suppliers can economically address. AEHR’s commentary suggests at least 1 global NAND leader concluded its existing infrastructure was inadequate, but this is still an unnamed-customer claim rather than a public industry standard. The competitive response from TEL, ACCRETECH, Advantest, Teradyne, memory-internal test teams, or custom captive tooling remains an important risk.

@thevalueistMay 24, 2026 · 9:15 AM

$NVDA $MU $SNDK $LITE If you listened to the last $AEHR conference call, you’d know HBF is much closer to commercialization than the market comprehends. https://t.co/EfbH8OiGSk EXECUTIVE INVESTMENT VIEW The TrendForce report is strategically important because it reframes NVIDIA’s Vera Rubin cycle as more than a GPU, HBM4, and NVLink transition. The report points to a possible architectural shift in which the GPU becomes a more direct orchestrator of storage and near-memory resources, potentially allowing NAND-based high-bandwidth flash to move from peripheral storage into the AI memory hierarchy. The highest-conviction conclusion is not that HBF replaces HBM, but that the AI server memory stack is broadening into a tiered hierarchy: HBM remains the latency-critical, bandwidth-dense working memory; HBF or HBF-like NAND becomes a high-capacity read-mostly tier for model weights, cold experts, long-context state, and potentially selected KV-cache use cases; SSDs and networked storage remain lower-cost capacity tiers. This is directionally positive for NVIDIA’s platform control, positive for NAND vendors with credible HBF roadmaps, neutral-to-positive for HBM leaders over the medium term, and incrementally negative for CPU-centric data movement architectures and undifferentiated storage vendors. The key caveat is that the specific GPU-Initiated Direct Storage Access architecture described by TrendForce remains reported, not formally disclosed by NVIDIA as a named Vera Rubin feature. Official NVIDIA materials already confirm the broader strategic direction: Vera Rubin NVL72 integrates 72 Rubin GPUs, 36 Vera CPUs, ConnectX-9 SuperNICs, BlueField-4 DPUs, NVLink 6, and storage-oriented BlueField-4 STX/CMX infrastructure, with NVIDIA describing accelerated storage and context memory as part of the Vera Rubin platform. The distinction matters: official NVIDIA disclosures validate that storage is becoming a 1st-class AI-factory subsystem, while the TrendForce report extends that thesis into a more aggressive version in which the GPU initiates and controls storage access more directly. That incremental command-path shift would be technically material if confirmed. The investment significance is concentrated in 3 areas. 1st, NVIDIA’s moat would expand from accelerator and interconnect leadership into memory-tier orchestration, making the CUDA/runtime/software layer even more central to AI cost-per-token economics. 2nd, NAND may gain an AI-specific premium growth vector after years of cyclicality, particularly for SanDisk/Kioxia, SK hynix, Samsung, and Micron if their roadmaps can support high-bandwidth, high-endurance, thermally stable products. 3rd, HBM demand is not impaired in the investable 2026-2027 window; rather, HBF is more likely to reduce the severity of model-capacity bottlenecks and expand the use cases that justify high-end GPU clusters. The HBM cannibalization debate becomes more relevant in 2028-2030, when software, standards, packaging, and qualification can mature enough for HBF to become a mainstream architectural tier rather than a prototype or niche inference accelerator. WHAT THE TREND FORCE REPORT ACTUALLY SAYS TrendForce reports that NVIDIA and Amazon are advancing storage architectures that allow GPUs to directly control storage devices such as SSDs, and states that NVIDIA is said to plan GPU-Initiated Direct Storage Access, or GIDS, beginning with Vera Rubin. The article contrasts GIDS with existing GPUDirect Storage, under which CPUs still issue data requests before data is transferred to GPUs. Under the reported GIDS model, GPUs would access storage directly, bypassing CPUs and DRAM. The report also cites Yonsei University professor Song Ki-hwan to argue that CPU thread limits are increasingly mismatched with GPU-scale parallelism, and that GPU-HBM data transfer may account for roughly half of system power, supporting interest in NAND-based high-bandwidth flash placed closer to GPUs. The article’s most important claim is the capacity math around HBF. TrendForce states that NAND has roughly 30x higher bit density than DRAM and that replacing a conventional all-HBM package with a hybrid configuration of 6 HBF units and 2 HBM units could raise memory capacity from 192GB to 3,120GB, or 16.25x. That math is broadly consistent with SanDisk’s public HBF fact sheet, which describes a 1st-generation 16-die HBF stack with 512GB capacity and 1.6 TB/s read bandwidth. However, the comparison is not a full system-performance comparison. It compares capacity, not latency, endurance, random-access behavior, software scheduling complexity, or total cost of ownership. It also uses a 192GB baseline that maps to an 8-stack, 24GB-per-stack HBM configuration, whereas official NVIDIA Vera Rubin material points to 20.7 TB of HBM4 per 72-GPU NVL72 rack, or roughly 288GB per Rubin GPU. The 16x headline is therefore directionally useful but should not be applied mechanically to Vera Rubin economics without adjusting for HBM4 stack density and system configuration. The article also correctly highlights the central limitation of NAND-based memory: endurance. TrendForce cites around 100,000 program/erase cycles for NAND versus DRAM’s effectively far higher write tolerance, and therefore frames HBF as better suited to AI model parameters that are largely read-only during inference. That framing is technically important. HBF is most compelling when data is large, reused, bandwidth-sensitive, and not frequently rewritten. HBF is less compelling for optimizer states, activation scratchpads, high-frequency KV-cache writes, and latency-critical random access unless software can amortize NAND latency through prefetching, batching, and locality-aware scheduling. TECHNICAL INTERPRETATION Existing GPUDirect Storage already reduces the traditional CPU-memory bottleneck by enabling direct DMA transfers between storage and GPU memory, reducing CPU overhead and avoiding unnecessary CPU copies. NVIDIA’s documentation frames GDS as a way to move large amounts of data efficiently between storage and GPUs with lower latency, higher throughput, and fewer CPU resources. The reported GIDS architecture would be a deeper change: not merely a faster data path, but a more GPU-native control path. In practical terms, the difference is that GDS can still rely on CPU-side orchestration and file-system mediation, while GIDS implies that GPUs can issue or schedule storage requests directly enough to remove the CPU and system DRAM from a larger part of the I/O loop. That distinction matters because AI inference is increasingly bottlenecked by memory movement rather than peak FLOPS. Large language model decode, particularly for large dense models and trillion-parameter MoE systems, repeatedly streams weights and reads KV-cache state while generating relatively small amounts of compute per token. In agentic systems, the problem worsens because multi-step tool use, long context, and multi-agent workflows create growing state footprints and unpredictable memory reuse. NVIDIA’s own Vera Rubin materials emphasize trillion-parameter MoE models, long-context windows, accumulated KV cache, and high-concurrency serving as core platform targets, with Vera Rubin NVL72 delivering 20.7 TB of HBM4 and 1.6 PB/s of memory bandwidth per rack. This makes the storage-memory interface strategically relevant rather than peripheral. The most plausible implementation path is not literal GPU random access to commodity SSDs at HBM-like latency. The more realistic architecture is a software-managed hierarchy in which HBM is the hot working set, HBF is a near-memory read-mostly extension, BlueField or equivalent infrastructure manages security, virtualization, and data services, and the GPU runtime/compiler schedules prefetches based on model execution. The important product question is whether NVIDIA can hide NAND latency behind predictable model execution and large-scale parallelism. If model weights, MoE experts, or context blocks can be staged ahead of use, HBF can behave like a capacity-rich bandwidth tier. If access is fine-grained, random, and data-dependent, NAND latency will show through and HBF will degrade utilization. This also explains why HBF is more likely to matter first in inference than in training. Training has heavier write traffic, optimizer updates, activation checkpointing, gradient synchronization, and more stringent memory-consistency requirements. Inference has more read-dominant model-weight traffic and can tolerate more explicit placement if the serving stack is optimized. SanDisk explicitly positions HBF for AI inferencing and states that HBF can deliver 8-16x HBM capacity at similar bandwidth and similar cost, with simulated performance within 2.2% of unlimited-capacity HBM for reading 8-bit pretrained weights on Llama 3.1 405B. That benchmark is favorable but narrow: it is an internal simulation focused on read-only pretrained weights, and it does not prove general-purpose DRAM substitution. HBF VERSUS HBM HBF should be viewed as a complement to HBM rather than a replacement through at least 2027. HBM remains indispensable for low-latency, high-bandwidth, high-write-endurance operations. It holds the hottest model shards, activations, attention state, and latency-sensitive KV-cache segments. HBF would instead expand the addressable memory footprint for read-heavy data that cannot economically fit in HBM. The best analogy is not “NAND replaces DRAM,” but “NAND becomes a new tier between HBM and SSD.” SK hynix and SanDisk explicitly describe HBF as a new memory layer between ultra-fast HBM and high-capacity SSDs, designed to bridge HBM performance and SSD capacity for AI inference while improving scalability and power efficiency. The relative economics are potentially attractive because HBM scaling is constrained by DRAM die supply, TSV capacity, advanced packaging, yield, power, and customer qualification. HBF leverages NAND density and could create much higher capacity per package footprint. SanDisk states that 1st-generation HBF reaches 512GB per 16-die stack at 1.6 TB/s read bandwidth, while projected Gen 2 and Gen 3 products exceed 2 TB/s and 3.2 TB/s read bandwidth, respectively, with capacities up to 1 TB and 1.5 TB per stack and lower power consumption versus Gen 1. These projections, if achieved, would create a credible high-capacity memory tier for inference, but still not erase the latency and endurance gap versus HBM. The potential bear case for HBM is therefore long-dated and conditional. If HBF becomes standardized, production-qualified, and broadly supported by NVIDIA/AMD runtimes, future systems may require less HBM per parameter served, especially for sparse MoE inference where cold experts can reside off-HBM. However, larger models and longer contexts usually consume any memory efficiency dividend quickly. Historically, memory relief in AI tends to enable larger workloads rather than reduce total high-end memory spend. In the base case, HBF reduces the binding constraint that caps model scale and improves GPU utilization, thereby increasing total AI infrastructure return on invested capital and preserving HBM attach as the hot tier. The more immediate risk is not HBM unit displacement; it is HBM bargaining power. If NVIDIA can credibly supplement HBM capacity with HBF and storage-class context memory, NVIDIA’s dependence on any 1 HBM supplier is reduced at the margin. That would not remove HBM scarcity, but it could slightly weaken the long-term strategic leverage of HBM vendors if HBF becomes a standardized alternative for capacity expansion. Near term, this is outweighed by continued HBM4 demand for Vera Rubin and competing platforms. NVIDIA’s official Q1 FY27 release states that Data Center revenue reached $75.2 billion, up 92% YoY, and the company guided to $91.0 billion in Q2 FY27 revenue while not assuming any Data Center compute revenue from China; these figures indicate that near-term demand remains constrained by high-end AI platform supply rather than by insufficient end demand. WORKLOAD FIT The cleanest HBF use case is storing model parameters for inference, especially for large MoE models. MoE architectures activate only a subset of experts per token, creating a large inactive parameter pool that does not need to reside entirely in HBM if the active experts can be prefetched and staged quickly. HBF could materially reduce the number of GPUs required to host a frontier model or allow a larger model to run within a given rack footprint. The benefit is less clear for dense models, where the full weight set is read repeatedly and HBF bandwidth must be high enough to avoid throttling every token. Dense models can still benefit from larger memory capacity, but they are less forgiving if HBF becomes part of the critical decode path without excellent prefetching. Long-context inference is the 2nd major use case. Multi-100K and million-token contexts create large KV-cache footprints. NVIDIA’s CMX context memory platform is explicitly designed to hold latency-sensitive, reusable inference context and prestage it to increase GPU utilization, with NVIDIA claiming up to 5x higher tokens per second and 5x better power efficiency than traditional storage. This is highly aligned with the TrendForce thesis even if the exact GIDS mechanism is not confirmed. CMX/STX demonstrates that NVIDIA is already productizing context memory as a separate tier in Vera Rubin-era AI factories. RAG, vector search, recommender systems, and data-intensive training-adjacent workflows are additional beneficiaries. These workloads often involve large external corpora, embedding tables, sparse feature lookups, or retrieval steps that do not map neatly into HBM. GPU-directed storage could reduce CPU overhead, reduce data-copy latency, and make GPU clusters more efficient at mixed inference plus retrieval pipelines. However, latency variance and tail behavior are critical. An architecture that improves average bandwidth but worsens p99 latency would be less attractive for premium agentic services. NVIDIA’s broader Vera Rubin messaging is heavily focused on low-latency, long-context, high-throughput agentic inference, suggesting that any storage-tier innovation must be evaluated on end-to-end token latency and utilization, not raw bandwidth alone. CREDIBILITY AND TIMING The report has medium credibility as a directional technology signal and lower credibility as a fully specified NVIDIA product disclosure. The directional credibility is supported by 4 independent data points: NVIDIA’s existing GDS documentation, NVIDIA’s official BlueField-4 STX/CMX announcements, SanDisk/SK hynix HBF standardization activity, and multiple NAND vendors’ work on high-bandwidth or low-latency flash for AI. The lower-confidence element is the precise claim that Vera Rubin introduces GIDS in the form described by TrendForce and The Elec, because official NVIDIA materials reviewed do not use that specific GIDS nomenclature in the same way. Commercial timing is unlikely to be binary. The 1st monetization layer is already visible in high-performance SSDs, DPUs, NICs, and AI storage reference architectures such as BlueField-4 STX. The 2nd layer is prototype and sample-stage HBF in 2026-2027. TrendForce reported that SanDisk is moving to establish an HBF prototype production line, with prototypes targeted for 2H26, pilot operation around year-end, and commercialization targeted for 2027. SK hynix and SanDisk have launched an OCP workstream for HBF standardization, but SK hynix also states that demand for complex memory solutions including HBF is expected to pick up around 2030. This points to a staged adoption curve: early samples and hyperscaler qualification in 2026-2027, specialized deployments in 2027-2028, and broader standardization later if software support and production economics validate. The path is technically non-trivial. GPU-initiated storage must address command submission, memory protection, virtual addressing, multi-tenant isolation, file-system/object semantics, error handling, wear leveling, encryption, telemetry, and orchestration across thousands of GPUs. NAND page sizes and SSD optimal access sizes are not naturally aligned with GPU warp-level fine-grained memory operations. HBF can overcome some of this through massive parallelism, TSV-style stacking, controller logic, prefetching, and software-managed placement, but the runtime stack must be tier-aware. This is exactly the type of co-design problem NVIDIA is structurally advantaged in, but it also means adoption will likely be limited to NVIDIA-optimized serving stacks before becoming broadly portable.

@aleabitoredditMay 24, 2026 · 2:28 AM

Photonics is nuanced and using ChatGPT/Gemini makes you miss all of it: 1. $SIVE is actually a chokepoint and partially a bottleneck. The reason it's a chokepoint is leading CPO/optical hyperscaler players go through Sivers, likely: Ayar. Celestial. Lightmatter. Lightelligence. Poet. If you take out Sivers, you literally can't make some of their products + delay their roadmap by years. As many are sole/primary source but are heading the direction on multi-source. As for the bottleneck argument: Win Semi is the bottleneck for scaling laser production. But... the nuance is when you have capacity allocated for the next few years. You become part of the bottleneck itself if players fight you for allocation of finished lasers. That's the nuance people miss with capacity allocation dynamics. It's like saying $SNDK is not part of the NAND bottleneck when Kioxia makes all of it. But when Sandisk has the ultimate control of output supply, they become the bottleneck + have all the pricing power. Sivers controls output supply of CW lasers given allocations, and as seen with $LITE earnings, CW laser is currently bottlenecked as everyone seems to be stuck producing EMLs. 2. Like how LLMs always uses em-dashes. You can tell when people use AI when they always use the same "CW is a dumb interchangeable laser" argument or compare "power" specs after conflating different architectures. That's why your "analysts" using AI will get this wrong over and over. There's CW lasers... and then there's a specific architectural design that Sivers achieves with DFB lasers. If you compare power specs with $LITE vs. Sivers, Lumentum wins in isolation. But they're completely different laser architectures. All the leading CPO players like Ayar, chose $SIVE for an architectural reason for high power, low thermal, laser arrays. $JBL 1.6T LRO also made one of the most dramatic moats cited by their fireside chat, using Sivers lasers. If you think CW lasers are interchangeable with Sumitomo/Furukawa, and others. And can be plug-and-play... i don't know what to tell you? Again: $SIVE makes architecturally unique CW lasers for leading CPO players. 3. I'm not sure how many times I need to say this: $SIVE for 2024-2025 has been going through development contracts. People using TTM revenue or former P/S metrics are using completely the wrong metrics, when there's volume ramp in 2027. It's the same with $AAOI which volume ramps in H1 2027. $AEHR which volume ramps after qualification. $LPK that volume ramps after qualification. This is just missing qualification cycles in semiconductors and how to model financials currently. As for the $LITE comparisons (which was also my long last year): $LITE literally started off selling laser dies before acquisition of Cloud Lite and other downstream optical engine components. This is where $SIVE is at today with starting off in the laser chokepoint for CPO: People are modeling laser revenue off very isolated TAM projections. Meanwhile Sivers is targeting M&A to expand revenue for TAM projections. This is not a simple component FAU + ramp valuation modeling over with a Taiwanese company. Since Laser companies like $LITE, $COHR are known to downstream expand to make their lasers more valuable, then vertically integrate (fabs, assembly) afterward. Again, Sivers worked with Ayar and these types of companies before they all became billion dollar companies. I have high conviction knowing they know what to acquire down the ELS/optical engine stack + pluggable transceiver for TAM expansion. It's just annoying when I get people who don't understand the nuances backseat commenting wrong things about my longs. I got the same thing about $AXTI is not a bottleneck! InP isn't needed! China! back at $14. Now it's $140 I got the same thing about $AAOI "is going down 50%!" back at $65. or "AOI management is shady at $30". Now it's $170 I got the "there's nothing new with $SOI" back at $45. Now it's $170. I think I'm one of the few who actually understands the nuances with photonics, since I did call out $LITE, $TSEM, Innolight, $AXTI, $AAOI, $SOI, that outperformed both photonics markets and overall markets over the past year. And now I'm long on $SIVE.

@aleabitoredditMay 23, 2026 · 4:55 PM

I don't post dollar amounts because they don't matter. What matters is return %. Speaking of that... YTD: 3840.39%. I'm probably the only one in the world. Who called out multiple names that 10x'd in a short timeframe. Do you remember these thesis anon? 1. $AXTI 2. $SIVE 3. $AAOI 4. $LITE 5. $IQE 6. $AEHR 7. $CRCL 8. $EWY 9. Unimicron 10. Nitto Boseki 11. $OSS 12. $GDRZF 13. $RPI 14. $SOI 15. $ALRIB 16. $SNDK 17. $SIMO 18. $VPG 19. $TSEM 20. $ARM 21. $MRVL 22. $INTC 23. $LPK 24. $NBIS 25. $MU They're all up 100-1000%+, because... 1. I post a thesis. 2. People can see how the stock performs months later. 3. They turn out right (thesis validation) because they're up hundreds of percent + hold their returns. I really dislike the traditional X influencer who shows large dollar amounts or fancy watches/cars/private jets. Then use that to get more by selling expensive subscriptions rather than through market returns. So trying to set a new trend off pure information discovery/synthesis from free thesis posts and the results that follow in terms of return percentages. TLDR: Market returns in terms of percentages matter the most to validate a thesis. Not the dollar amount made.

@aleabitoredditMay 19, 2026 · 2:50 PM

@ryansfinance Think you’re forgetting $NBIS went from $90 -&gt; $235 -&gt; $180. $AEHR went from $25 -&gt; $110 -&gt; $77 While $IREN went from $45 -&gt; $70 -&gt; $45

@investmentguru_May 16, 2026 · 12:04 PM

Most people are focused on silicon. The real bottleneck nobody’s talking about? Indium Phosphide. $NVDA needs faster chips. Faster chips need faster interconnects. Faster interconnects need InP-based lasers. And InP supply is critically constrained. WHAT IS InP AND WHY DOES IT MATTER? Indium Phosphide is the substrate behind high-speed optical components — the lasers and photodetectors moving data at 800G and 1.6T speeds inside AI data centers. Silicon simply can’t do what InP does at these speeds. No InP = No optical interconnects = AI infrastructure hits a wall. THE FULL InP VALUE CHAIN 𝗦𝘂𝗯𝘀𝘁𝗿𝗮𝘁𝗲𝘀 & 𝗪𝗮𝗳𝗲𝗿𝘀 → $AXTI — one of the only publicly traded InP substrate suppliers in the West → $IQEPF — epitaxial wafer supplier feeding InP laser production 𝗖𝗼𝗺𝗽𝗼𝗻𝗲𝗻𝘁𝘀 & 𝗟𝗮𝘀𝗲𝗿𝘀 → $COHR — vertically integrated, owns InP laser fabs → $IPGP — fiber and InP laser exposure → $AAOI — transceiver/laser components → $LITE — InP laser supplier to hyperscalers 𝗦𝗶𝗹𝗶𝗰𝗼𝗻 𝗣𝗵𝗼𝘁𝗼𝗻𝗶𝗰𝘀 / 𝗙𝗼𝘂𝗻𝗱𝗿𝗶𝗲𝘀 → $TSEM — InP photonics foundry capabilities → $GFS — compound semiconductor exposure 𝗣𝘂𝗿𝗲-𝗣𝗹𝗮𝘆 𝗦𝗽𝗲𝗰𝘂𝗹𝗮𝘁𝗶𝘃𝗲 → $POET — optical interposer platform designed to integrate InP lasers at scale, potentially solving the bottleneck directly → $LWLG — electro-optic polymer platform, InP alternative play → $ALMU / $SIVEF — smaller speculative names in the photonics supply chain 𝗧𝗲𝘀𝘁𝗶𝗻𝗴 & 𝗩𝗮𝗹𝗶𝗱𝗮𝘁𝗶𝗼𝗻 → $AEHR — wafer-level burn-in testing for photonic chips → $KEYS — optical and high-speed signal testing WHY SUPPLY CAN’T JUST SCALE OVERNIGHT InP isn’t silicon. → Only a handful of facilities globally can produce InP substrates → Building new InP fabs takes years and billions → Geopolitical concentration risk is real — much of the supply chain runs through Asia → Hyperscaler demand for 800G/1.6T is accelerating faster than supply can respond This is a structural bottleneck — not a temporary one. THE INVESTMENT THESIS IN ONE LINE The AI buildout runs through InP. And InP supply is stuck. The picks-and-shovels play isn’t just $NVDA. It might be the substrate nobody’s heard of yet. Not financial advice.

@aleabitoredditMay 15, 2026 · 9:34 PM

Leopold Aschenbrenner is a legend, but I'm not quite sure he can beat 3152.77% YTD in the Serenity Awareness fund. That being said, I've hit 23 different longs this year with 100-1000%+ YTD. 1. $AXTI 2. $AAOI 3. $SIVE 4. $LITE 5. $IQE 6. $AEHR 7. $CRCL 8. $EWY 9. Unimicron 10. Nitto Boseki 11. $OSS 12. $GDRZF 13. $RPI 14. $SOI 15. $ALRIB 16. $SNDK 17. $SIMO 18. $VPG 19. $TSEM 20. $ARM 21. $MRVL 22. $INTC 23. $LPK Do you remember all of these anon?

@thevalueistMay 15, 2026 · 6:42 PM

$AEHR vs. $TRT : Two Divergent Models Capturing the GAI Semi Reliability-Testing Inflection. Aehr Test Systems and Trio-Tech International both participate in the semiconductor reliability testing ecosystem, but they represent materially different economic exposures. Aehr is best framed as a specialized semiconductor test equipment and consumables company, with differentiated exposure to wafer-level burn-in, package-level burn-in, high-power AI processor reliability screening, silicon photonics, and wide-bandgap power semiconductors. Trio-Tech is best framed as a semiconductor back-end services, burn-in board, burn-in systems, reliability lab, and industrial electronics company, with exposure to outsourced testing, customer-specific reliability programs, burn-in boards, regional Asian test capacity, automotive semiconductor reliability, and industrial/aerospace electronics demand. Aehr monetizes production tools and interface hardware. Trio-Tech monetizes test capacity, engineering services, burn-in boards, customer-specific process execution, and industrial electronics distribution. That distinction dominates the comparative analysis.

@aleabitoredditMay 13, 2026 · 4:28 AM

$SIVE earning results don't matter, I don't see risks in financials other than your usual media trying to cause panic. Main indicators to look at like I've done before with $AAOI / $AEHR. Are around laser volume production going forward in 2027. Maybe anything around Win Semi capacity lock would be bullish. Or Ayar, Jabil, and other customer volume production. Any added note around new hyperscaler suppliers would be bullish.

@aleabitoredditMay 12, 2026 · 12:08 AM

$AEHR is up close to 3x in 1 month by the way. I tend to get timing extremely well?? https://t.co/bIbpSXcYtP

@aleabitoredditMay 11, 2026 · 9:23 PM

I'm literally analyzing earnings, not the chart. Looking at reactions shortly after to see if it's positive/negative is stupid. As seen with $AAOI, $AEHR or any single earnings I've looked at recently, it doesn't matter what happens right after unless you're weekly option trading. And if it drops, while it's actually structurally positive, then that's a buying opportunity as seen with $AAOI.

@aleabitoredditMay 11, 2026 · 3:22 PM

@TomSzczypka $LITE nasdaq 100 lot of inflow. Algorithms tied $SIVE, $AAOI, $AEHR, and others to Lumentum now as the sector leader. So they all got brought up together.

@aleabitoredditMay 11, 2026 · 12:31 AM

Promising… but for actually making returns: Things like Glass/Silicon Photonics is an immediate H2 2026-2029 event. 2029+ timeline seems accurate for plastic optics and cables cited by Tosoh. Better to wait until it’s about to hit volume ramp as seen with $SIVE or $AEHR. Feel like in general, going long on quantum dot, microled, and plastic way too early on… might just end up stalling out for a year in the next correction? The nuance is you find any indicators timelines have been pushed up a lot (CPO got pushed up), then that’s alpha. But it’s something to save for a year and a half later like H1 2028 for plastic optical cables.

@investmentguru_May 9, 2026 · 3:53 PM

The Complete Semiconductor Playbook — AI Supercycle 2026 🤖 EDGE AI $QUIK · $CEVA · $SYNA · $QCOM · $INDI · $VLN · $SMTC → AI leaves the data center. Devices. Cars. Factories. Multi-year secular trend just getting started. 🏗️ AI INFRA $SGH · $SKYT · $AVGO · $NVDA · $ARM · $CRDO · $MRVL → The “Nvidia-only” era is over. The full stack is getting priced in. ⚡ POWER $POWI · $TXN · $MPWR · $VICR · $GANX · $AEHR → AI data centers are power monsters. Unsexy. Essential. Increasingly scarce. 🔬 FUTURE-TECH $LWLG · $AXTI · $POET · $TSEM · $GFS · $IPGP → Silicon photonics foundry capacity = national strategic asset. 🛠️ EQUIPMENT $AMAT · $LRCX · $KLAC · $ASML · $ONTO · $ACMR → Without these, nothing above gets built. Map the full stack. That’s where the alpha is. Not financial advice. DYOR

@kaizen_investorMay 9, 2026 · 10:11 AM

You can earn so much by just listening to earning calls and interpret the numbers. The algo’s just look at the basic earning numbers and decide to trade on that, they don’t interpret them. There are so many examples. For $FLNC the algo’s pushed it from 12 to 16. If you would have listened to the earnings you realise that 16 is still way too low, so I bought. 1 day later and Fluence now trade at 24. Same for $WOLF, the numbers in itself weren’t fantastic so algo’s pushed it down 12%. But if you know the company and listen to the earnings call, you would have realised they are on the right track, $WOLF is now up 32% since. Another example, but one I couldn’t capitalise on is $AEHR. I remember them beign down 12% in aftermarket. I wanted to buy them but I don’t have broker who supports that. $AEHR is up 50% since. Learn to interpret the numbers and what you should be looking for in an earnings call and you’ll be a better investor.

@mrmikeinvestingMay 7, 2026 · 2:23 AM

We are currently in a “once in a lifetime” AI super cycle… Phase 1 was: (already gone) Semiconductors ~ $NVDA, $AMD, $INTC, $ARM Phase 2 is: (passing by now) Memory ~ $MU, $SNDK, $WDC Photonics ~ $AAOI, $AEHR, $LITE, $MRVL The current phase is Neo Cloud/AI infrastructure: $IREN, $NBIS, $CRWV, $CIFR, $APLD Next wave (many will miss) Rare Earths ~ $USAR, $MP, $UUUU, $FCX Power & Cooling~ $VRT, $CEG, $OKLO, $OSS Finally it all concludes with these 3 sectors: Robotics ~ $TSLA, $PATH, $SERV Space ~ $RKLB, $ASTS, $PL, $LUNR Drones ~ $ONDS, $AVAV, $LMT Many will make generational wealth from this AI super cycle over the next 7 months. Save this to look back on later…

@aleabitoredditMay 5, 2026 · 7:54 AM

True… I did call: $AXTI $12 -> $105 $SOI $43-> $145 $TSEM $110 -> $218 $AAOI $30 -> $180 $IQE $13 -> $46 $LITE $363 -> $1000 $AEHR $30 -> $85 (if you count siph segment) As individual thesis posts… and others like $COHR and Innolight last year… I think $SIVE is the most debated right now, but I do think I’m right. Feels like I’m seeing something others don’t with photonics?

@aleabitoredditMay 4, 2026 · 3:19 PM

@tnglxn4 $AEHR just went up 3x after I called it? Things don’t move in a straight line. Waiting for volume ramp news.

@investmentguru_May 4, 2026 · 11:13 AM

Money still coming in this stocks $LITE $COHR $AAOI $AXTI $ALMU $SIVEF $GFS $LWLG $VIAV $AEHR

@investmentguru_May 2, 2026 · 6:29 PM

The Full Optical AI Industry Chain: 1. IC Design & Manufacturing (Where the brains are built) ASIC / xPU / Photonic ICs → $NVDA $MRVL $AVGO $LITE $COHR $INTC $AMD $CSCO Wafer Foundries → $TSM $TSEM $GFS $UMC 2. Materials (The Overlooked Layer) (The foundation of photonics) Indium Phosphide → $AXTI $IQE Gallium Arsenide → $AXTI $MTSI This layer is underfollowed and could see strong upside as optical demand expands. 3. Optical Components (The data highways) VCSEL / EML Lasers → $LITE $COHR $AVGO $AAOI Optical Module Assembly → $FN $COHR $LITE $MRVL $CSCO Fiber / WDM / Optical Connectivity → $CIEN $GLW $LITE MPO Connectors → $GLW $APH Optical Fiber → $GLW 4. Packaging & Testing (The hidden backbone) Packaging → $ASX $AMKR $FN $JBL Packaging Equipment → $KLIC Testing → $TER $KEYS $FORM $AEHR

@aleabitoredditMay 2, 2026 · 7:10 AM

@TheMathMan70 Are you kidding? I always write high conviction if I’m confident in my thesis. And I do it with $SIVE, $AEHR, $AAOI, which all end up tripling. I don’t share position sizing since I don’t want blind copy traders. People should manage their own portfolio and build their own conviction. Without just copying me.

@jasonl_capitalMay 1, 2026 · 11:48 AM

10. $AEHR - Aehr Test Systems (Test & Measurement) Every photonics chip needs to be tested before it ships. Aehr's FOX-XP platform does wafer-level burn-in - testing full wafers in parallel to catch early failures before packaging. Won a major new silicon photonics customer in March. Record $41M production order from their lead hyperscale AI customer. Second-half bookings exceeded $92M.

@investmentguru_May 1, 2026 · 10:52 AM

In 10 months 8/10 gave big returns $AEHR +841% $ASTS +270% $RKLB +267% $OUST +200% $INDI +105% $AQST +94% $QS +79% $ABCL +66% $SOUN -19% ❌ $PCVX -16% ❌

@aleabitoredditMay 1, 2026 · 6:37 AM

@vz921 I guess it’s just luck calling out $AAOI, $IQE, $AXTI, $SIVE, $SOI, $AEHR, and others that all went up 3x this year?

@aleabitoredditApr 30, 2026 · 6:26 AM

$LPK / $LPKFF earnings are out. Seeing a lot of very dumb commentary on X. If you're wondering how to analyze qualification-cycle players, it's the same as $AEHR. Nobody cares about current earnings unless there's something extremely bad. If your revenue declines -8M euros before any volume ramp, it doesn't mean anything. The only reason why LPKF is a long anyway is 2027 LIDE glass core substrate mass production. Main thing to look at is earnings call in 2 hours not current financials and indication of high volume production + customers. People made this same mistake with $AEHR selling off on previous financials instead of listening to the call.

@aleabitoredditApr 28, 2026 · 7:39 PM

I do really like $LPK. Critical monopoly chokepoint in glass substrates… Which are used for advanced packaging and CPO. I flagged it as a potential 10x back in Jan, but thought it was a bit early. However… time seems right now? "About four years ago, we began collaborating with a semiconductor company to develop mass-production equipment for direct 3D waveguide formation," Lee said. “The customer has already installed LPKF's equipment." Maybe Samsung or SKC Absolics since this was in Korea? Seems like momentum is ramping up now though like $AEHR pre-earnings, not quite high volume (2027), but around this time felt compelling for me. During the transition from qualification/pilot -> high volume.

@mrmikeinvestingApr 26, 2026 · 12:26 AM

THE FULL PHOTONICS BOTTLENECK It’s clear that the next phase of AI scaling is no longer compute, but rather optical connectivity. 1. $AAOI builds high-speed transceivers with vertically integrated lnP laser manufacturing and early 1.6T hyperscale demand. 2. $AEHR providing burn-in and reliability testing for AI and optical hardware, with Sonoma gaining traction at hyperscalers. 3. $AVGO delivers core networking and optical connectivity through custom silicone used across hyperscale AI infrastructure. 4. $COHR scaling lnP lasers, and optical engines to support next-gen AI networking demand. 5. $MRVL builds DSP and interconnect silicone that powers, high speed optical infrastructure. 6. $LITE supplies lasers and optical components backed by NVDA, and a growing optical switching backlog. 7. $CRDO enables faster data movement through cables, retimers, and interconnect silicone, expanding into silicone photonics. 8. $CSCO provides switching, routing, and optical networking gear that ties AI clusters together at scale.

@investmentguru_Apr 25, 2026 · 9:50 PM

Building a Semiconductor Watchlist: 20 Names to Watch $AMKR — Advanced packaging $GFS — Mature-node foundry $LSCC — Low-power FPGAs $AEHR — Wafer-level burn-in test equipment $AXTI — Compound semiconductor substrates (GaAs, InP) $NVTS — GaN/SiC power chips, AI data center play $POET — Optical semiconductors, photonics $INDI — Automotive/ADAS chips $VLN — Automotive connectivity $HIMX — Display drivers, IoT $VECO — Semiconductor equipment $SKYT — US-based foundry, govt contracts $PLAB — Photomasks for chipmakers $SQNS — 5G/IoT chips $GCTS — 5G connectivity $MX — Power & display chips (Korea) $ICG — ASIC crypto/AI chips $ACLS — Ion implant equipment $TTM — PCB/semiconductor packaging $VSH — Passive components, discrete semis

@aleabitoredditApr 25, 2026 · 9:02 PM

@ISITrading2717 Nope, I stand by my words. I haven’t been wrong on any high conviction stock yet from $HOOD, $RKLB, $TSM, $AAOI, $AEHR, $SIVE. And now $IREN community member IQ.

@aleabitoredditApr 25, 2026 · 8:20 PM

@workoutwithp $AEHR long term is very early. The only started volume ramp with one customer… And others will pick up (like their t1 optical transceiver one) over the next few years. For current valuations, I’d say it’s more priced in near term.

@aleabitoredditApr 25, 2026 · 7:34 PM

Uh chat... is my timing insane or what? Names like $TSEM are flat all year, I go long, then it doubles. The moment I go long on $AEHR, it almost triples. People are starting to think I'm the catalyst because this keeps happening over and over... lol this is a $22B+ company. Fun fact on how I get these right : -> Identify critical companies not really noticed -> I time my longs around catalysts like $NVDA GTC or OFC. -> And around when news about material changes comes about. -> Then look at good entry points on drops (this is where your astrology TA's get used). Going long isn't just picking a random point in time! On other names, like $ALRIB it's new information discovery (eg. Microsoft Quantum), so two different types. There's actually a strategy here that repeats... (as seen with the other 16 triple digit return stocks).

@aleabitoredditApr 24, 2026 · 3:55 PM

Photonics go brrr. $MXL +76.2% (lol) $AAOI +15.03% $SIVE +12.69% $AEHR +7.07% $SOI +6.19% $LITE +4.96% Hope you didn’t sell and try and chase other bottlenecks anon? https://t.co/Ft2p94bDWR

@kaizen_investorApr 22, 2026 · 2:43 PM

We all make mistakes. Should have followed my gut feeling here and just buy $AEHR before earnings. The drop just hours before earnings scared me a bit. Stock is now up almost 100%. https://t.co/Q328kF4ZgP

@thevalueistApr 21, 2026 · 1:19 AM

I'm a firm believer in reflexivity. The SMID-cap GAI infrastructure trades that are at beginning of building their businesses potentially stand to benefit greatly. Still hard to pick the long-term winners. https://t.co/A4bayRdSLl $AEHR $ATXI $SIVEF $LWLG What Is Reflexivity? Reflexivity refers to a theory in economics where investors' perceptions influence economic fundamentals and in turn, those perceptions and market prices. Reflexivity originates in sociology, but it is prominent in the field of finance. George Soros, its leading advocate, believes that the deviations from economic equilibrium are due to positive feedback loops. According to Soros, reflexivity theory challenges equilibrium and the efficient market hypothesis.

@investmentguru_Apr 21, 2026 · 12:22 AM

Photonics names showing clear relative strength today — this rotation is getting noticeable. $CRDO $MRVL $AEHR $SIVEF $GFS $POET $LWLG

@mrmikeinvestingApr 20, 2026 · 11:44 PM

I’ve said this before, &amp; I’ll say it once more… If you missed: $RKLB at $30. $NBIS at $60. $SNDK at $200. $AEHR at $30. Then you CANNOT miss out on $ONDS here at $10. Ondas is directly involved in one of the fastest growing themes, &amp; that being drones. 10x potential… https://t.co/lKEnl4pULM

@aleabitoredditApr 20, 2026 · 6:11 PM

Told ya. $AEHR is cooking. Once you reach a certain point with MC it’s mostly institutions driving price. If you find a name that a ton of hyperscalers use… it’s feels bound to get repriced? https://t.co/AHqpgbqHJK

@investmentguru_Apr 20, 2026 · 11:19 AM

Missed Testing Testing: $AEHR $KEYS $VIAV $MKSI

@aleabitoredditApr 19, 2026 · 11:18 PM

@MisterPariPassu A lot of the upstream supply chains are not in the US. Mostly concentrated in Japan, Korea, Vietnam, Taiwan, etc. Normally US-listed equities more priced in at a certain point in time. It's just rarer to find an $AAOI, $AEHR, sitting around.

@aleabitoredditApr 19, 2026 · 8:40 PM

Wow, majority of these 30 stocks I’ve liked are up a lot in just two weeks (just a recap to new folks) By the way, my long term opinion doesn’t change on any of them from $MRVL, $AMD, $ARM and others. Short term entry points do though with names like $AAOI to $AEHR. And they make the difference between +10-20%. I focus a lot about the “undiscovered” ones like Riber or $SIVE or $RPI or $IQE in analysis when I make a new entry -> wait for it to play out. But the same thesis around $LITE or $NBIS or $AXTI from last year is still the same. And I don’t need to post that same thesis multiple times, since it’s not new anymore. But the reason they’re not new is because markets have validated the thesis and are repricing the stocks live because of them.

@investmentguru_Apr 19, 2026 · 2:53 PM

@pinetree_cap $AEHR

@investmentguru_Apr 18, 2026 · 9:34 PM

$ATYR aTyr Pharma is advancing ATYR1923 through Phase 2 clinical trials for interstitial lung disease — a rare, underserved condition with few effective treatments. The company’s novel mechanism is built on tRNA synthetase biology, a class of proteins with unique roles in immune modulation. Orphan drug designation provides meaningful regulatory advantages. Binary risk by nature — but the upside on a successful Phase 2 readout is genuinely transformational for a company this size. $ARQT Arcutis is the only commercial-stage company on this list with three FDA-approved products already generating revenue. ZORYVE cream, foam, and lotion are all on market in dermatology — a specialty segment with high barriers to entry, strong physician loyalty, and durable pricing. Prescription volumes are growing, payer coverage is expanding, and the company is the closest name on this entire list to reaching profitability. The anchor of the biotech portion of this framework. $POET POET’s Optical Interposer is a patented platform that integrates lasers, modulators, and photodiodes onto a single chip — effectively semiconductorizing photonics for the AI data center era. The company has transitioned from a prototype story to a commercially validated platform, with initial production orders signaling a meaningful revenue ramp into 2026. Management expects to ship more than 30,000 optical engines this year across 800G, 1.6T, and high-power segments. Active module development partnerships include LITEON, Semtech, Lessengers, and NTT Innovative Devices. The company is also co-developing 3.2Tbps optical engines with Quantum Computing Inc. — a format whose global market is projected near $12 billion by 2030. Revenue is still early, losses are real, and dilution is a risk — but the pipeline is now layered, and for a $564M company playing in a market headed for $12B+, the upside math is compelling. $AEHR Aehr Test Systems provides wafer-level and package-level burn-in and reliability testing — the critical quality control layer for AI chips, silicon photonics transceivers, and power semiconductors. By early 2026, the company’s narrative had shifted decisively from its original silicon carbide focus toward AI processors and silicon photonics for optical interconnects — a much larger and faster-growing market. Q3 2026 bookings hit $37.2M with a book-to-bill ratio exceeding 3.5x and a record backlog of $50.9M. Management sees a clear path to sustained growth as its installed base expands across AI, silicon photonics, and power semiconductors. Full-year FY2026 revenue is guided to the high end of the $45–50M range, with second-half bookings expected at the high end of the $60–80M range. Picks and shovels for the photonics wave — the testing layer that no one talks about. $BKSY BlackSky delivers real-time, on-demand AI satellite intelligence to defense, government, and commercial customers worldwide. Its new Gen-3 constellation delivers 35-centimeter very-high-resolution imagery and achieved general availability in March 2026 after rapidly commissioning four satellites, while simultaneously securing an eight-figure sovereign intelligence contract. Full-year 2025 revenue came in at $106.6M with a backlog of $345M, up 32% year-over-year. International revenues surged over 50%, now exceeding half of total revenue — a signal that defense diversification beyond the U.S. government is working. A sole-source $99M U.S. government IDIQ contract for next-generation earth observation payloads underlines its defense credibility further. Four analysts carry Strong Buy ratings with a $28.50 price target.

@blackpanthercapApr 18, 2026 · 12:56 PM

the simplest institutional edge you can steal as a retail investor: follow the supply chain, not the stock price. most retail investors watch $NVDA go up and ask: what other AI stocks should i buy? institutional analysts do something different. they trace the supply chain upstream/downstream from nvidia and ask: what does nvidia need that it can't build itself? the answer is almost always a company you've never heard of, with a market cap under $2B, that makes something very specific and very irreplaceable. and we seen this lately; $AAOI, $SIVE, $AXTI, $AEHR probe cards. photomasks. power conversion ICs. optical interposers. the parts that sit between the fab and the finished product. those companies get priced last. they move after the institutional allocation has already happened. by the time nvidia is in your news feed, the supply chain is three moves ahead. that's the edge. trace the chain. find the chokepoint. own it before it's obvious. -BP

@aleabitoredditApr 17, 2026 · 7:52 PM

Genuinely thanks for nice comments. I share my ideas for free in the end though since I want to help out the retail community. $TSEM hit triple digit return... so that's 16 different names YTD. So my YTD hit 1525%+ as a result. Just to recap all the endless abuse and harassment along the way: 1. $AXTI - "Pump and Dump", "Scam Chinese Stock", Got banned from WSB $RDDT after Mods got mad investors actually made money AXT going from $12->$80. 2. $AAOI - "Pumping stock with no fundamentals, Meme stock" 3. $SIVEF - "Pump and Dump" "Meme Stock" 4. $LITE - "Photonics Bubble" 5. $IQE - "Just pumping low MC stocks" 6. $AEHR - "Stock with negative revenue growth, why is anyone following this guy and not paying $2,000+ for my subscription?" 7. $CRCL - "TA says it's going down to $30" 8. $EWY - "Just from followers" (hint, it's the South Korean Index) 9. Unimicron - "Idea is useless give me US stocks" 10. Nitto Boseki - "Idea is useless give me US stocks" 11. $OSS - Stealing Ideas (no, my synthesis around Venezuela was novel) 12. $GDRZF - "You're a terrible human trying to profit off of the War in Venezuela" 13. $RPI - "Meme stock all because of a Meme Trader" (FT, European Media). 14. $SOI - "Pump and dump", "no novel idea" (random analysts) 15. $ALRIB - "Pumping low MC stocks" (no, it's $MSFT quantum information discovery) 16. $TSEM - "Pumping based on followers alone" (bro it's $25B+, these are institutions) Or how about... the idea around fundamentals was right all along? And I'm just sharing information synthesis/discovery before institutions find out about them. Retail and media should be celebrating when 16+ different ideas return 100%+ YTD, since stocks are positive sum. Everyone from retail, the companies, and local economies benefits. Instead, negativity is through the roof and people keep trying to diminish/downplay the ideas like frontrunning the photonics supercycle… even when they actually turn out right? The trolls are starting to get to me, from $IREN folks creating new accounts every day just to send IRL threats, to European media disinformation about "pumping and dumping"... since I do read every comment. But notice... how 95% of things keep going up? And institutions like Point72 and Apollo end up buying the names I mention? Comments like this do make it helpful to stay on X, and I do enjoy taking victory laps on the haters.

@aleabitoredditApr 17, 2026 · 4:48 PM

$HPS.A is up 21% since my thesis 2 weeks ago. Unfortunately, not everything can go up 100% every 2 weeks like $AXTI? If the transformers bottleneck is that extreme for multiple years… Sometimes it’s better to wait for markets to price that in. $AEHR or $AAOI triple digit rallies in few week timeframes probably set unrealistic expectations… Even 10% return yearly is usually good in markets? This has been 21% so far and I have comments asking if this is still good, but I’m hoping it’s another triple digit return in due time.

@jasonl_capitalApr 17, 2026 · 1:14 PM

Growth stocks I'm watching right now: $IREN | $48.18 - AI data center infrastructure, 4.5 GW pipeline, largest single-site build in the US $CRDO | $164.09 - AI connectivity, 1.6T optical DSPs in 3nm, just acquired DustPhotonics for $750M $AEHR | $84.49 - AI chip testing, TAM expanding 3-5x beyond SiC into GaN and silicon photonics $HIMS | $27.12 - Telehealth + owned pharmacy network, licensed care model that's hard to replicate $ZETA | $17.87 - AI-powered marketing platform, 56% off its all-time high $OUST | $24.41 - Digital LiDAR on a silicon chip, SAFE LiDAR Act is a policy tailwind $OSS | $9.95 - Ruggedized edge AI compute, guiding 20-25% revenue growth, new energy node contracts $IBRX | $7.31 - ANKTIVA immunotherapy, $44M Q1 revenue, EU conditional approval in Feb

@investmentguru_Apr 17, 2026 · 12:04 AM

$AEHR just dropped a BIG signal for the AI infrastructure trade Aehr Test Systems announced a record $41M production order from a lead hyperscale AI customer — its largest ever. Here’s why this matters - • Total 2H bookings now exceed $92M (already above guidance) • Direct exposure to custom AI ASICs used in data center training + inference • Strong validation of Sonoma platform for high-power AI chips • Next-gen AI accelerators already in pipeline → more orders likely This isn’t just a one-off order — it confirms a multi-year demand cycle driven by: → Hyperscalers building custom AI chips → Exploding power requirements of AI workloads → Need for reliability + burn-in at scale Translation for the market: $AEHR is positioning itself as a picks-and-shovels play in AI infrastructure — similar to how semicap names benefited in prior cycles. And it’s not just AI chips: Silicon photonics, SiC, GaN, memory → all expanding TAM. This is what early-stage capacity ramp + demand visibility looks like. If AI infrastructure continues its trajectory, $AEHR could be entering a high-growth phase into FY2027 and beyond.

@aleabitoredditApr 16, 2026 · 7:44 PM

@45daysignals Fun fact I got banned on WSB subreddit after $AXTI went from $12 to $80+ because Mods got mad it kept going up and played out correctly. Names like $RPI got endless harassment over "meme stock" narratives. Including FT and news media... but it ended up doubling anyway. Names like $SOI, endless pump and dump harassment. Names like $SIVE, endless Swedish salt harassment. Names like $AEHR, endless harassment. Names like $SOI, endless harassment thinking someone else went long before I did. Idk... whenever you're first to something, it's just a ton of hate comments. And only after it's a $5-10B+ company after going up 1000%+ is it "okay" to go long.

@aleabitoredditApr 16, 2026 · 7:36 PM

Glad to hear it! I've went long and wrote thesis posts on about out 15 different stocks that hit 100-1000%+ YTD? 1. $AXTI 2. $AAOI 3. $SIVEF 4. $LITE 5. $IQE 6. $AEHR 7. $CRCL 8. $EWY 9. Unimicron 10. Nitto Boseki 11. $OSS 12. $GDRZF 13. $RPI 14. $SOI 15. $ALRIB Not including others like $TSEM that are about to hit triple digit returns too in a month. The amount of hate people like myself get for posting free ideas over the internet is pretty insane TBH. Starting to make sense why people just set up $20,000 paywalls and sell info to Western institutions instead of helping out salty retail investors (especially over in Europe). But helps me keep motivated to keep posting with these positive comments.

@investmentguru_Apr 16, 2026 · 6:52 PM

$AEHR https://t.co/7QBIlJQSQR

@investmentguru_Apr 16, 2026 · 3:38 PM

Photonics and Space bouncing today $AAOI $COHR $LITE $LWLG $POET $AXTI $AEHR $SATL $RKLB $ASTS $SIDU $SPIR

@aleabitoredditApr 16, 2026 · 3:06 PM

$AEHR receives $41M production order from lead hyperscaler customer. As I said after earnings, they indicated volume ramp “shortly after”. This is it. $AEHR +13% today on the news. https://t.co/rpF3qTHoJz

@thevalueistApr 16, 2026 · 2:40 PM

This $AEHR deal is only $41mm, and the stock is up ~18%. The deals will simply get bigger and more frequent. How many tens (hundreds?) of billions of dollars' worth of optical networking equipment, and now the coming HBM/HBF for AEHR (listen to the last earnings call if you don't know what that is about), will be installed over the next decade?

@thevalueistApr 16, 2026 · 12:21 PM

$AEHR (Bloomberg) -- Aehr Test Systems shares jump 18% before the bell after the maker of semiconductor equipment said it received a $41 million production order and expects second half bookings to exceed $92 million. •Aehr said it received a record $41 million follow-on production order from its lead hyperscale customer for package-level burn-in (PLBI) of custom AI processor ASICs •Notes the order brings its bookings in the second half of Aehr’s fiscal year to more than $92 million to date

@aleabitoredditApr 16, 2026 · 8:32 AM

It doesn't need to be detailed with $ALRIB? > Riber effective, profitable duopoly with $VECO on MBE equipment, that traded at low fwd multiples. > $MSFT Quantum as core hyperscaler buyer of Riber from public information discovery > $IQE, QD Laser (Quantum dot), IntelliEPI and others using it. Thought process: What other companies under $1B supply directly to hyperscalers like $GOOGL to $MSFT for their frontier programs? And are critical suppliers that can't be replaced? Can't really name any aside from $AEHR, but that's now $2.3B from $600m... So thought $ALRIB was compelling for Quantum/Silicon Photonics exposure. If a profitable ~300m company is powering a hyperscaler's quantum frontier program… Probably going to get re-rated triple digits over time.

@investmentguru_Apr 15, 2026 · 11:50 PM

Photonics names took a breather today, consolidating while the broader market ripped — with key players like $AAOI, $COHR, $LITE, $AXTI, $AEHR, and $OPTX in the red; looks more like rotation than weakness as capital chased higher-beta momentum in quantum and AI, and if anything, this kind of pause in photonics after a strong run could set up the next leg once money rotates back into the optical/AI infrastructure trade.

@investmentguru_Apr 15, 2026 · 10:44 AM

The AI supply chain is still the hottest sector in the market From optics to semis to testing — capital is flowing across the entire stack- $AAOI $AXTI $OPTX $SNDK $LWLG $AEHR $ICHR $BE $VIAV $LITE $MRVL $FORM $WDC $SIVEF This isn’t just a GPU story anymore… It’s about everything behind AI infrastructure: • Optical connectivity • Semiconductor materials • Testing & packaging • Data + storage The message is clear: AI demand is pulling the entire ecosystem higher Follow the supply chain — that’s where the next winners are emerging

@aleabitoredditApr 14, 2026 · 4:46 PM

"Silicon photonics scaling hits wafer testing bottleneck" - $FORM (helps that ficontec left), $AEHR - Over in Taiwan MPI (6223), MA-tek (3587), MSScorps (6830) - Then large cap Cap - $TER, Advantest (6857) Are some decent exposure off the top of my head if you’re curious. I’ll make a deeper dive ETF for you all soon.

@aleabitoredditApr 13, 2026 · 4:53 PM

They’re still there. It’s just hard to say anything…. When all my recent thesis posts from $HPS.A, $IQE, $AXTI, $SIVE, $AAOI, $LITE, $NBIS, Win, Shunsin, $AEHR, $TSEM, $SOI, and many many others I call out. Just hard outperforms the market. Year to date of +1,116.29% isn’t too bad, right chat?

@kaizen_investorApr 12, 2026 · 1:48 PM

Some key catalysts I'm looking forward to in 2026-2027. Q2 2026 - $POET: high-volume light source scaling. POET unveiled its next-generation Starlight and Blazar hybrid external light sources at the March 2026 OFC conference, heavily pushing forward their high-volume scaling for the AI connectivity market. - $IREN: Sweetwater 1 1.4GW facility energization. IREN is finalizing site works for the Sweetwater 1 bulk substation, with energization on track for April 2026 (Q2). - $KRKNF: Covelya Group $615M acquisition close The strategic acquisition of the UK-based Covelya Group rapidly expands their vertically integrated product suite to include navigation, positioning, and subsea communications. - $AEHR: ultra-high power FOX-XP shipments In March 2026 (Aehr's fiscal Q3), the company announced a follow-on order and a new major silicon photonics customer win for its FOX-XP systems, with shipments scheduled for its fiscal Q4 ending in May 2026. Q3 2026 - $CIFR: AWS AI hosting lease commencement Cipher secured a 15-year, 300MW lease with AWS for an AI/HPC campus, with initial rent commencement confirmed for August 2026 (Q3). - $OUST: Start SAFE LiDAR ACT Ouster should see the first positive signs of the SAFE LiDAR ACT in the Q3 results. Q4 2026 - $RKLB: Neutron maiden flight Rocket Lab is developing the Neutron rocket to provide a reusable, medium-lift alternative to SpaceX's Falcon 9 for constellation deployments. A successful maiden flight proves the vehicle's viability, opening up a lucrative pipeline of commercial and government launch contracts. - $ASTS: commercial service activation (block 2) The activation of its larger Block 2 BlueBird constellation enables the initiation of consistent commercial service for telecom partners globally. - $NBIS: 800MW to 1GW capacity operational In early 2026, Nebius significantly raised its contracted power guidance, officially targeting 800MW to 1GW of available connected capacity by year-end 2026. - $AAOI: 5,000 ELSFP units per month production Reaching a production run rate of 5,000 External Laser Small Form Factor Pluggable (ELSFP) units monthly proves their automated U.S. manufacturing can scale to meet co-packaged optics demand. The goal is 400,000 units by the end of 2027. - $FLY: Blue Ghost Mission 2 lunar launch Blue Ghost Mission 2 will deliver payloads to the far side of the moon while demonstrating the first on-orbit AI data processing via its Elytra spacecraft. - $RDW: SabreSat DARPA Otter VLEO demonstration The SabreSat Otter demonstration for DARPA will validate highly advanced air-breathing electric propulsion, allowing satellites to maneuver indefinitely in Very Low Earth Orbit (VLEO) by harvesting atmospheric air. Q1 2027 - $CRDO: Cardinal 1.6T DSP production ramp Credo launched its Cardinal 1.6T DSP family in March 2026. Industry analysts predict the major 1.6T transceiver upgrade cycle and corresponding production ramp will heavily take hold in 2027. - $PL: Pelican-4 edge AI revenue ramp Planet successfully demonstrated in-orbit AI object detection on its Pelican-4 satellite, supporting the planned ramp of its Planetary Intelligence edge AI revenue streams into 2027. Q2 2027 $VRT: Ironton liquid cooling expansion operational The expansion of its Ironton facility will massively boost production capacity for liquid cooling systems, a necessity for the booming deployment of high-density AI servers. Q3 2027 $SATL: NextGen Merlin constellation operational Satellogic unveiled its new Merlin constellation, explicitly targeting the first launch in October 2026 with full operational capacity scheduled for the first half of 2027. $BSKY: AROS country-scale mapping constellation launch The launch of the AROS constellation supplements their targeted site monitoring with country-scale broad area mapping, securing their position against retiring legacy competitor satellites. $LWLG: material volume production and licensing revenues During their recent early 2026 updates, LWLG management explicitly stated they are building infrastructure for long-term commercialization, with meaningful volume production and licensing revenues anticipated to begin in 2027.

@investmentguru_Apr 12, 2026 · 1:19 PM

10 Stocks That Have Nearly DOUBLED in 2026 YTD. While the market bleeds, these names are printing $SNDK Pure-play NAND flash powerhouse spun off from Western Digital. Revenue up 61% YoY driven by AI data center demand. Joining Nasdaq-100 on April 20. The AI memory supercycle is real. $AXTI Makes indium phosphide substrates — the critical material powering silicon photonics. Plans to double manufacturing capacity in 2026. Every AI data center needs what they make. $AAOI Optical transceiver maker riding the hyperscaler buildout wave. Direct picks-and-shovels play on AI fiber infrastructure. Revenue exploding as data centers scale. 💡 $IBRX Anktiva bladder cancer drug revenue up 168% YoY. Now approved across 34 countries with FDA label expansion pending. From clinical-stage to commercial inflection — fast. 💊 $AEHR Wafer-level burn-in testing leader for silicon carbide and photonics chips. Picks-and-shovels for the semiconductor ramp no one talks about. Quietly one of 2026’s best setups. $SATL Geospatial intelligence + DoD contract momentum driving this space stock. Defense and surveillance demand exploding globally. Under the radar but not for long. $LITE Pure-play photonics and optical components powerhouse. Direct supplier to hyperscalers building out AI data center interconnects at massive scale. Transceiver and laser demand exploding as cloud providers race to upgrade optical infrastructure. Revenue recovering strongly after a multi-year downcycle — the comeback trade in photonics. The “layer below the chips” thesis made into a stock. $OPTX Optical component supplier benefiting from photonics buildout AND defense optics demand. Dual catalyst — AI infrastructure + government contracts. Tiny float, big moves. $LWLG Electro-optic polymer company powering next-gen silicon photonics modulators for AI data centers. Signed a deal with Tower Semiconductor targeting 400G per lane optical modules. A Fortune Global 500 company now in Stage 3 of their design win cycle. Pre-revenue but catalysts stacking fast. $NBIS European AI hyperscaler projecting revenue run rate to grow from $1.25B to $7-9B by end of 2026. GPU cloud infrastructure at scale. Think AWS but built purely for AI workloads. ☁️

@aleabitoredditApr 11, 2026 · 10:37 PM

Serenity's Follower Picked Hyperbolic 10x ETF Performance. Week 1: +12.39% $AEHR: +56.72% ($45.08 -> $70.65) $AAOI: +39.63% ($108.86 -> $152.00) $SIVE: +35.35% (9.9 SEK -> 13.4 SEK) $ENAFF: +31.58% ($1.71 -> $2.25) $AL2SI: +25.44% (28.70 EUR -> 36 EUR) $ENVX: +21.30% ($5.07 -> $6.15) $BZAI: +18.99% ($1.79 -> $2.13) $POET: +16.04% ($6.11 -> $7.09) $WATT: +14.81% ($15.8 -> $18.14) $HGRAF: +14.48% ($4.49 -> $5.14) $VLN: +13.79% ($1.16 -> $1.32) $LPK.DE: +13.20% (6.59 EUR -> 7.46 EUR) $FLY: +13.09% ($33.16 -> $37.50) $VPG: +11.63% ($44.7 -> $49.90) $PLAB: +9.86% ($40.87 -> $44.90) $TRT: +8.33% ($5.88 -> $6.37) $EQR.AX: +7.94% (.315 AUD -> .34 AUD) $LASR: +7.92% ($60.7 -> $65.51) $ASPI: +6.67% ($4.2 -> $4.48) $P4O.DE: +5.69% (6.85 EUR -> 7.24 EUR) $EOS.AX: +3.11% ($9.00-> $9.28) $ADUR: -0.29% ($10.37 -> $10.34) $MITK: -2.52% ($13.9 -> $13.55) $ALCJ: -3.41% (2.05 EUR -> 1.98 EUR) $TMC: -5.01% ($4.59 -> $4.36) $QURE: -9.94% ($17.21 -> $15.50) $EONR: -20.00% ($.9 -> $.72) Top 3: 1. $AEHR: +56.72% 2. $AAOI: +39.63% 3. $SIVE: +35.35% Honorable mention $ENAFF with a 31.58% return. Weighted average was 12.39%. Honestly not bad everyone, you beat year index returns in just 1 week.

@investmentguru_Apr 11, 2026 · 12:53 AM

PHOTONICS — THE MOMENTUM TRADE SMART MONEY IS ROTATING INTO Institutional capital is quietly flowing into optical networking. This isn’t hype — it’s AI infrastructure. Why photonics is leading right now → AI data centers are hitting a copper bandwidth wall → Optical interconnects = faster, scalable, lower power (light > electrons) → Next-gen systems are being built around this shift → 800G+ demand is accelerating faster than supply can match The supply-demand imbalance → Capacity is tight across key components → Long lead times = sustained demand visibility → Pricing power shifts to suppliers → This is how multi-month sector runs start The names to watch → $LITE — core laser tech, supply constrained, strong demand tailwind → $COHR — scale player, diversified + aligned with AI ecosystem → $FN — manufacturing backbone, major beneficiary of volume ramp → $CIEN — network infrastructure layer, hyperscaler exposure → $MRVL — silicon photonics + custom AI data center chips → $AEHR — testing demand ramp tied to new photonics cycle $POET , $LWLG , $SIVEF , $AAOI , $AXTI What confirms the trend → Breakouts across multiple names — not just one → Moves happening without headlines = institutional accumulation → Strong relative performance vs $SPY Don’t overcomplicate it → This is a sector move, not a single-stock story → You don’t need perfect entries — you need the right theme → Momentum phases last longer than most expect The play → Focus on strength within the sector → Let momentum + capital flows work → Stay until relative strength breaks — then rotate Photonics is the plumbing of AI. And this cycle is just getting started. Not financial advice. DYOR.

@aleabitoredditApr 10, 2026 · 9:27 PM

@Ren_aramb Swiggly line drawers that charge $500 a month are possibly the greatest annoyance to me... Especially when they comment underneath my thesis posts like $AEHR earnings. Stocks move based on fundamentals and catalysts. TA is astrology self-fulfilling prophecies but if $AMZN has a massive purchase order, nobody cares about "overbought" based on TA it's about modeling in how much FCF that brings relative to MC. Same with anyone who tries drawing a TA for $IREN when there's literally $6,000,000,000 in dilution that a chart doesn't account for.

@aleabitoredditApr 10, 2026 · 8:30 PM

I feel like I've called out the most triple digit stock returns YTD... Out of anyone in history? Hence why I have 150k+ followers now! In just a short timeframe: $AXTI -> 5x+ $AAOI -> 5x $SIVE -> 2x+ $LITE -> 2x+ $IQE -> 2x+ $AEHR -> 2x+ $CRCL -> 2x+ $EWY IV -> 2x Unimicron -> 2x+ Nitto Boseki -> 2x+ $OSS -> 2x+ $GDRZF -> 2x+ $AEHR -> 2x With many more like $TSEM, $RPI having close to triple digit returns. Not including many others last year like $HOOD or $RKLB for triple digit returns, just this YTD. There's stuff like $FORM and others like Macronix... and $NBIS that actually doubled from the bottom at $70. But I won't take credit since I didn't do a specific post about it during the timeframes. There's a difference between just mentioning among many other tickers. Then having conviction like myself, writing a specific thesis post about it, getting catalyst timing right, and going long yourself. But proud if this helped retail going the right direction. Especially that they don’t need to pay $2,000+ just to see tickers people go long on or join some “special club” for company discussion.

@investmentguru_Apr 10, 2026 · 10:45 AM

Follow the money — smart flow is rotating into next-gen tech infrastructure $AAOI → Massive momentum on AI/data center demand + 800G transceivers → Recently saw explosive upside driven by hyperscaler orders $AEHR → Quiet leader in AI chip testing + silicon photonics → New orders tied to AI processors + data centers $AXTI → Substrate play for compound semis (AI, photonics, 5G) $COHR → Core player in lasers, optics, and photonics infrastructure $SNDK → Memory + storage demand rising with AI + data growth Money is clearly flowing into AI infrastructure backbone — optical, photonics, semis, and data center supply chain. These aren’t hype names… they’re picks & shovels of AI.

@aleabitoredditApr 10, 2026 · 10:31 AM

@yuntungshieh Honestly would not be surprised if $AEHR or $ALRIB 10x either. But $IQE had a lot of latent capacity for InP, they just needed to restructure and refactor their reactors.

@aleabitoredditApr 10, 2026 · 7:37 AM

There's many nuanced architectural changes for "supercycles" in photonics. If you feel like you missed $LITE from last year, just frontrun the next cycle with CPO / lasers. $SIVE was my lesser known pick for laser exposure. They're in $JBL 1.6T LRO, $MRVL via Celestial, O-Net -> Hyperscalers. Which is extremely abnormal for a ~$350M MC company. If I had to draw parallels: it's similar to $AAOI 2025 or $AEHR now with current qualification cycles before volume. Which is why retail misunderstands it when you look at purely financial numbers today with negative eps or capex spend. And a lot of that execution uncertainty with volume production / capex spend for scaling dfb laser is de-risked from Win Semi. However, it looks to just keep ramping past 2029+ starting from H2 2026. Might be a little early... But I think it's a decent read on what's up and coming. Dilution is a real risk. But a lot of fears are priced into its ~$350M MC, especially when a company like $AVGO or AlChip can just buy $SIVE for vertical integration or to troll Marvell’s CPO program.

@investmentguru_Apr 10, 2026 · 1:20 AM

Photonics — still one of the strongest sectors to keep on watch 👀 $AAOI $COHR $LITE $POET $MRVL $LWLG $TSEM $SIVEF $NOK $OPTX $AEHR $LASR $OPTX

@investmentguru_Apr 10, 2026 · 1:12 AM

$AEHR strong and long https://t.co/xVrSX0NpzT

@aleabitoredditApr 10, 2026 · 1:01 AM

@justaladd_ Glad $AEHR worked out for you! Yeah this rally was pretty incredible. It was funny to see so many retail investors going short because they don’t understand qualification cycles https://t.co/xZZcSF3swF

@aleabitoredditApr 10, 2026 · 12:42 AM

@0xInitialAce How many more triple digit stocks do you need to escape the underclass? Feels like $AAOI 4x, $AXTI 5x, $AEHR 2x, $IQE 2x, $LITE 2x, and others should be enough?

@aleabitoredditApr 10, 2026 · 12:32 AM

No, I’m still at a conservative 759% YTD after today’s overall rally. Lost too much from hedging. Would probably be up more if indexes and many individual names like $RDDT weren’t so down from macro. But if you pick selective winners like $AAOI or $AEHR … in hyperscaler supply chains: Turns out it’s possible outperform markets? I do think it’s a tad harder than it looks finding important players in each sector and timing catalysts. $TSEM was basically flat the entire year until I bought, then it rallied 90% in 3 weeks, so timing important too (eg. OFC announcements). And I’ve developed PTSD after finding the unknown $TSM COUPE glass substrates supplier… Only to watch them get bought out by Apollo shortly after. So not always having a good time. But glad if my ideas helped others outperform indexes or see where frontier industries are heading to.

@aleabitoredditApr 9, 2026 · 7:32 PM

Woah, $AEHR almost triple digit return in 1 week. This is speed running “Escape the Permanent Underclass with Serenity 101”. That aside, it helps to understand qualification cycles when you look at stocks or ER. A lot of retail investors went short after ER by looking at current revenue rather than volume ramp + all ongoing hyperscalers testing out orders.

@aleabitoredditApr 9, 2026 · 10:27 AM

Here's a bunch of random 30 US-available random stocks I like today and why: 1. $INTC - America's hope for foundry, national security 2. $MRVL - scales rev from future maia asics and add ons like cpo, they do everything lost count 3. $TSM - backbone of semis/ai 4. $COHR - They do everything vertically integrated + captures optical cycle 5. $RKLB - the final frontier of space will be around 5 years from now and 20 years from now. 6. $DRAM - memory exposure for samsung/sk hynix 7. $AVGO - hyperscalers dont like nvidia gpu tax 8. $AMZN - nobody can compete against the overnight shipping of toilet paper. robotics will lower opex over time 9. $ARM - AGI CPUs scale revenue quite a bit over the next decade 10. $TSEM - you're going to need a foundry for light based stuff 11. $IBIT - bitcoin, we all know by now 12. $NBIS - i think it's the next AWS. Also they do self-driving cars with uber, own scaling DB companies, data labeling. It's almost like a mini Google. 13. $GOOGL - youtube is not going away, gemini is great. they're vertically integrated with TPUs and fund buildout with operating income so i like it. 14. $AMKR - super facilities coming online in late 2027-2028. benefits from made in america 15. $HOOD - i dont like short term, but long term i'm a fan of Robinhood since they captured retail + have more products like banking, etc that they're scaling up. product innovation is wild. 16. $CRCL - I happen to really like stablecoins and see them as the future for both payments/holding (depends on clarity act) 17. $META - people aren't going to stop using instagram or whatsapp, or others anytime soon. 18. $LITE - $GOOGL TPU exposure decently high part of BOM. As long as Google's AI program keeps running I think $LITE will do well. 19. $LPTH - Germanium and China export controls will always be an issue so US made engineered alternatives will always be important 20. $FN - Someone needs to assemble optical stuff 21. $JBL - same as above, but added with ip from Intel's SiPh acqusition so might end up like innolight? 22. $MP - American rare earths program is extremely important, similar to $INTC national security risks 23. $HIMS - Okay here me out they just acquired a ton of companies, and at $19 they have global DTC channel. short sellers really hate this company, but I think it's actually promising as a contrarian long 24. $SMTC - LRO/LPO transition 25. $POWL - US alternative to hammond for switchgear DC type bottleneck 26. $VPG - Humanoids will be a thing down the road maybe 2027-2028, this makes the sensors. 27. $MOG.A - Feels like i see them everywhere in robotics, to spacex supply chains 28. $MSFT - At $375, one day we'll look back and see this as a buying opportunity. 29. $CVX - oil might crash after war but these oil companies are going to be extremely important, especially when Venezulea is a goldmine. 30. $XLU - i think rate cuts might be back online, we need power/grid for AI so these names will always be improtant from $CEG to $NEE Just throwing out other thoughts aside from $AAOI and $AEHR.

@aleabitoredditApr 9, 2026 · 9:09 AM

@NabQ321 Good question! $AAOI, $AEHR are two extremely high beta, fast growth stocks. Maybe some larger companies like $COHR, $TSEM, $SMTC, $MRVL, $INTC or $FN on the next drop? Companies like AOI or AEHR are pretty rare to come by.

@investmentguru_Apr 9, 2026 · 12:36 AM

My favourite Top 5 Plays $AAOI - Massive growth + strong contract wins - new ATH today $MRVL - NVDA-linked photonics backbone play - Celestial AI deal = long-term dominance $RKLB - Strong revenue growth, macro-driven pullback - Neutron + space sector catalysts in play $ASTS - Satellite-to-cell = massive TAM - Any launch confirmation = explosive move - $AEHR - Picks-and-shovels of photonics boom - Already showing relative strength Not financial advice. DYOR.

@investmentguru_Apr 9, 2026 · 12:28 AM

Photonics / Optical / Semiconductor — Full Stack View Foundries $TSEM $GFS → Scaling silicon photonics capacity to meet demand surge Optical Components & Lasers $COHR $LITE $IPGP $LASR → Core building blocks: lasers, modulators, photodiodes Transceivers & Optical Engines $AAOI $POET → Where electrical meets optical — key bottleneck + opportunity Chip Design & SiPh Integration $MRVL $MTSI → Designing the brains behind optical interconnects Manufacturing & Packaging $FN → Advanced assembly becomes critical as complexity rises Networking & Systems $CIEN $GLW → Moving massive data through fiber at scale Testing & Materials $AEHR $AXTI $LWLG → Enabling performance, reliability, and next-gen efficiency Key Drivers → AI data center buildout (800G → 1.6T → CPO) → Demand > supply through 2027 → Aggressive M&A across the ecosystem This isn’t just a sector — it’s an entire supply chain being repriced in real time. Photonics isn’t coming… it’s already here. Not financial advice. DYOR.

@jasonl_capitalApr 8, 2026 · 10:20 PM

4. $AEHR at $60 - The AI Chip Quality Gatekeeper Every AI chip needs to be tested and burned in before it ships. Aehr Test Systems is the market leader in wafer-level burn-in for AI processors, silicon photonics, and power semiconductors. Just reported $37.2M in quarterly bookings at a 3.5x book-to-bill ratio. Backlog of $50.9M. Expanding into burn-in for GPUs, AI accelerators, and optical transceivers for data center interconnects. The more chips get built, the more testing they need.

@investmentguru_Apr 8, 2026 · 6:26 PM

Photonics emerging stronger today $AAOI $COHR $LITE $AEHR $AXTI $POET $SIVEF $LWLG

@aleabitoredditApr 8, 2026 · 2:12 PM

@Yolo365247isme @raghavrmehta $10B MC on $AEHR seems possible in 1 1/2 years.

@thevalueistApr 8, 2026 · 2:06 PM

I am confident a large portion of this $AEHR move is a short squeeze, and I still highly recommend you listen to the entire call from yesterday. There was great insight into AEHR, of course, as well as significant color on the rest of the optical networking and the broader GAI picks-and-shovels ecosystem. AEHR is pushing hard to land burn-in testing deals for HBM4 and HBF (they already have a meaningful NAND business). Hockey stick optionality if those hit. $NVDA $MU $SNDK $LITE

@aleabitoredditApr 8, 2026 · 2:01 PM

Did you listen anon? $AAOI and $AEHR both up 50%+ this week alone. I post my thesis publicly BEFORE any movement happens and without charging thousands to see stock picks. If you’re listening to $IREN investors why these are “short or going to dump because of 2025 P/E ratios” maybe stick to PayPal or $SPY.

@kaizen_investorApr 8, 2026 · 1:49 PM

Well, so far for my $AEHR buying opportunity. Was looking for an entry but looks like I’m too late. Like I said, earnings were great imo. Aehr Test Systems designs and manufactures specialized test and burn-in equipment for the semiconductor industry to ensure the reliability of integrated circuits, particularly silicon carbide power devices and silicon photonics. A critical monopoly in a growing sector.

@aleabitoredditApr 8, 2026 · 1:45 PM

@si0xac4742 Yeah a lot of the $IREN investors got mad at me and tried shorting $AEHR or $AAOI... I think they're just adding fuel to the fire at this point since $AEHR is up 24%.

@aleabitoredditApr 8, 2026 · 1:41 PM

@AsianbeBlazin yeah i decided have a higher concentration on $AEHR after the new silicon photonics t1 customer like $LITE, $AVGO, or $COHR. a ton of of retail investors misunderstood qualification cycles -> volume ramp so they went short. its crazy to see retail try and short $AAOI and $AEHR just to face infinite losses in a supercycle.

@aleabitoredditApr 8, 2026 · 1:37 PM

Not too shabby with $AEHR. Up +20% today. For all the bears, this is what happens when you misunderstand qualification cycles in semi names. Then try and look at current revenue numbers only. https://t.co/9hU4UwNyRE

@aleabitoredditApr 8, 2026 · 3:16 AM

@TD_btc24 Most recent 5 thesis posts I've shared: 1. $HPS.A ($1.77B) - Transformer/Switchgear DC bottleneck 2. $ARM ($152B)- AI CPU ramp 3. Win Semi ($5.7B) - Foundry for CW lasers and other supply chains from SpaceX to humanoids 4. $SIVE ($295M) - CW Laser ramp for H2 2026 and 2027. 5. $TSEM ($22B) - photonics foundry Apart from those, names I've positively mentioned like $MRVL, $AAOI, $RDDT, $NBIS, $RPI, $AEHR, $LITE, $COHR, SK Hynix, $LASR, $SOI, $IQE, and others might be decent additions.

@aleabitoredditApr 8, 2026 · 2:33 AM

Getting tired of 100K+ follower accounts that who charge $2,000+ a year just to see their stock picks… That then get frustrated over someone who shares names from $AAOI to $AXTI for free… And then judge $AEHR or $SIVE by previous year revenue without understanding CPO/SiPH ramp in 2027. Feels like these accounts will get eroded over time as people realize they can get alpha for free instead of paying thousands for it.

@thevalueistApr 8, 2026 · 12:42 AM

$AEHR EXECUTIVE CALL SUMMARY: Aehr Test Systems (04/07/26) The central conclusion from the call is that Q3 FY26 was a demand inflection quarter, not a clean earnings inflection quarter. Reported revenue remained weak at $10.3 million, down 44% y/y, and non-GAAP net loss was $1.5 million, or $0.05 per diluted share. Against that, bookings surged to $37.2 million, quarter-end backlog reached $38.7 million, and effective backlog rose to $50.9 million after another $12.2 million of bookings in the first 5 weeks of Q4. On the official numbers, book-to-bill was roughly 3.6x. Management reiterated 2H FY26 revenue guidance of $25 million to $30 million and 2H non-GAAP EPS guidance of -$0.09 to -$0.05, but qualitatively tightened the outlook by saying FY26 revenue should finish on the high side of the prior $45 million to $50 million range and 2H bookings on the high side of the prior $60 million to $80 million range. The correct read-through is that end-demand and order momentum improved materially, while reported revenue, margin capture, and operating leverage remained under pressure.  A more precise framing is that the quarter widened the gap between demand formation and revenue realization. Through the first 9 months of FY26, revenue was $31.2 million versus $44.9 million in the prior-year period, and non-GAAP EPS was -$0.09 versus +$0.16. Non-GAAP gross margin was 36.5% in Q3, which was 620 bps below FY25 Q3’s 42.7%, although it did recover 661 bps from the 29.8% level in Q2. By contrast, Q3 bookings increased 54% y/y versus FY25 Q3 and 500% sequentially versus FY26 Q2. Effective backlog of $50.9 million now equals roughly 86% of all FY25 revenue. The debate has therefore shifted away from whether the company has demand and toward 3 narrower questions: how much of backlog converts in the next 1 to 3 quarters, how much of that conversion is system revenue versus consumables, and whether mix can support the promised return to profitability.  GUIDANCE AND WHAT CHANGED The most important guidance point is that the formal numbers did not rise. Q2 had already reinstated 2H FY26 revenue guidance of $25 million to $30 million and 2H non-GAAP EPS guidance of -$0.09 to -$0.05, while indicating expected 2H bookings of $60 million to $80 million. Q3 kept those numerical ranges intact. What changed was management’s confidence inside the ranges. The call explicitly moved revenue expectation to the high side of the prior FY26 range and bookings expectation to the high side of the prior 2H bookings range. That is a meaningful tonal improvement, but not a headline raise. Since 9M FY26 revenue is $31.2 million, the official full-year revenue range still requires roughly $14 million to $19 million of Q4 revenue, which implies a significant sequential step-up from Q3. The practical implication is that the call improved FY27 setup more than it de-risked FY26 execution.  A second, more subtle guidance change sits inside the wafer-level AI benchmark commentary. In the January 8 Q2 release, management had said it expected to complete data collection that quarter on the benchmark with a top-tier AI processor supplier. On this call, management disclosed that the program “has taken longer than we originally expected” because of “a technical misunderstanding on the clock configurations,” and that WaferPaks are now being redesigned while additional data is collected over the next several months. That is not a thesis break, but it is a real timing slip versus the prior quarter’s implied schedule. The positive implication is that the program is still active. The negative implication is that some of the highest-value wafer-level opportunities are still engineering-sensitive and harder to underwrite on a quarter-by-quarter basis.  ORDER BOOK QUALITY AND REVENUE CONVERSION The record backlog is important, but its revenue timing is uneven. The February 26 $14 million AI wafer-level order is scheduled to ship within 6 months, which means a portion is near-term and a portion likely falls into the transition period or early FY27. The March 31 silicon photonics order from a major new customer is scheduled to ship in fiscal Q4 ending May 29, 2026 and is therefore the clearest immediate revenue candidate. By contrast, the February 11 next-generation Sonoma AI ASIC order is scheduled for summer 2026 delivery, and the March 3 silicon photonics follow-on order is scheduled for 2H calendar 2026. The analytical consequence is straightforward: the backlog surge is real, but a meaningful share of the announced February-March orders is structurally more supportive of transition-period and FY27 revenue than of FY26 Q4 revenue.  That timing asymmetry is especially important because Aehr’s own FY25 10-K emphasizes that most customer purchase orders are subject to cancellation or rescheduling with limited penalties and that backlog at any particular date is not necessarily indicative of actual sales in a succeeding period. The same filing also highlights the company’s persistent concentration risk: 77% of FY25 revenue came from the top 5 customers, with 2 customers accounting for 39% and 15%, respectively. In other words, the quarter substantially improved confidence in demand, but did not eliminate the company’s historical sensitivity to a small number of order movements and shipment deferrals. For an investment committee, that distinction is critical. Backlog should be treated as strong directional evidence, not as fully contracted revenue.  AI PACKAGE-LEVEL BURN-IN Package-level burn-in on Sonoma remains the fastest and cleanest monetization vector in the AI opportunity set. The company has now announced an initial production order for a lead hyperscaler’s next-generation higher-power AI ASIC, and management stated on the call that it also expects a “significant near-term follow-on order” for the current device family. The qualitative Q&A color was also important. Management said that many AI ASIC suppliers still do not run production burn-in today, but that “the common theme is they’re all moving to burn-in.” That comment matters because it implies the current Sonoma opportunity is not confined to one customer or one device cycle. It suggests a category shift in high-power AI silicon reliability screening as thermal density and current levels rise.  The other important package-level read-through is that management increasingly framed Sonoma not as a standalone niche tool, but as the first monetization step in a larger account lifecycle. Qualification burn-in often starts at package level because customer test modes and production flows are already understood there. Wafer-level is economically superior in many cases, but more design-for-test dependent and more likely to require engineering iteration. In that context, Sonoma appears to be the lower-friction entry point, with FOX wafer-level potentially following on later generations. That interpretation was reinforced by management’s disclosure that the lead hyperscaler is already discussing DFT for a 3rd device so that wafer-level burn-in can be considered on the roadmap. The implication is that package-level and wafer-level should not be modeled as competing products. The more likely pathway is package-level first, wafer-level later where economics justify pre-package screening. Manufacturing capacity also appears materially less constrained than in prior periods. The company said it expects to begin shipping Sonoma systems this quarter from a newly upgraded contract manufacturing facility with capacity for more than 20 additional systems per month, in addition to Fremont. If that capacity comes online as described, supply should stop being the binding constraint for Sonoma. The gating factors would then shift to customer qualification timing, module availability, and end-customer ramp cadence. That is constructive because it increases the probability that FY27 revenue is limited by demand conversion rather than internal manufacturing bottlenecks.  AI WAFER-LEVEL BURN-IN AI wafer-level burn-in remains the potentially larger long-term economic vector, but it is also the segment with the greatest execution complexity. Management was emphatic, stating that “AI wafer-level burn-in is really hot right now,” and the $14 million follow-on order from the lead AI wafer-level customer clearly demonstrates that the company has moved beyond proof-of-concept with at least 1 customer. The strategic logic is compelling and was articulated clearly on the call: screening high-power AI devices before they are combined with expensive substrates, HBM stacks, photonics, or other die inside advanced packages can save substantial downstream cost and avoid multi-die package scrap. That logic becomes more powerful as package content and packaging cost rise.  However, the call also provided the clearest evidence to date of what can slow this adoption curve. The top-tier benchmark did not fail, but it did slip. The explanation was operationally specific rather than market-driven: “a technical misunderstanding on the clock configurations” forced redesign of the WaferPak approach. That makes the right analytical conclusion neither fully bullish nor fully bearish. The conclusion is that wafer-level AI remains highly valuable and apparently technically viable, but schedule predictability is lower because each program can involve customer-specific DFT assumptions, custom contactor work, power/current constraints, and a learning curve on both sides. The stock can support a higher multiple on wafer-level AI only after more of these benchmark programs convert into repeatable production orders, not merely because the TAM is large. The Q&A also sharpened the adoption model. Management effectively described a market still in discovery mode. Some customers start by assuming package-level burn-in is the only option, then pivot toward wafer-level once they see the economic trade-off. Others are already asking what DFT changes should be embedded into next-generation devices so wafer-level screening becomes easier. The practical implication is that wafer-level AI revenue should be expected to compound in a stair-step pattern rather than in a straight line. Large orders are possible, but they are likely to remain lumpy until customer design flows standardize around wafer-level burn-in requirements. SILICON PHOTONICS Silicon photonics was arguably the cleanest positive surprise in the call. The March 31 order from a major new customer was not a narrow pilot. It included both engineering qualification and high-volume production tools, and it is scheduled to ship in fiscal Q4. Management also discussed a follow-on order from the lead silicon photonics customer and described integration into fully automated, lights-out operation with autonomous guided robots. This combination of new-customer acquisition, production intent, automation integration, and near-term shipment timing makes silicon photonics look more de-risked than several of the still-evolving AI wafer-level benchmarks.  The segment matters for 3 reasons. First, it gives the wafer-level business a near-term growth leg that is not entirely dependent on AI processor qualification cycles. Second, it is inherently aligned with the same macro driver powering AI infrastructure buildout, namely the shift from copper to optical interconnect as bandwidth, power, and thermal constraints intensify in hyperscale data centers. Third, it appears to be moving more quickly from evaluation to production than some AI wafer-level programs. Management’s tone on silicon photonics was notably less exploratory and more operational. The company now has both an installed base and new-customer proof points in this market. For valuation purposes, silicon photonics now deserves to be treated as a real growth contributor rather than residual optionality.  POWER SEMICONDUCTORS AND MEMORY Power semiconductors and memory should be treated as separate buckets. On power, management was conspicuously disciplined. There was constructive commentary on GaN and a new SiC customer in Taiwan, but the call also explicitly said the company is “not yet counting on significant revenue from this segment to return yet.” That is an important change in narrative emphasis. Historically, Aehr’s equity story was heavily tied to SiC recovery. By FY25 Q3, management had already disclosed that SiC wafer-level burn-in, which represented over 90% of fiscal 2024 business, was tracking to less than 40% of fiscal 2025 business, while AI burn-in had already reached over 35% in its first year. The present call confirmed that the center of gravity has moved decisively away from waiting for a SiC rebound and toward AI, silicon photonics, and broader data-center infrastructure demand.  On memory, the tone was more strategically interesting than near-term bankable. Management said it achieved the correlation requested by a key memory supplier and is now discussing test system specifications and a development agreement. The call also expanded the memory narrative from flash into HBM-adjacent opportunity. That is conceptually powerful because HBM sits inside the same advanced packaging ecosystem where wafer-level screening becomes more economically attractive. Yet the timeline described on the call still looks long. Management’s own language suggested potential fiscal 2027 orders and fiscal 2028 ramps after a 12 to 18 month development effort. That places memory firmly in the category of meaningful upside optionality, but not something that should be heavily capitalized into the next few quarters of estimates. MARGIN, CASH FLOW, AND CAPITAL STRUCTURE The margin profile improved sequentially but remained well below the level needed to validate a durable earnings inflection. Q3 non-GAAP gross margin was 36.5%, versus 29.8% in Q2 and 42.7% in FY25 Q3. Q1 FY26 had already been 37.5%, so Q3 does not yet establish a new upward trend; it mainly confirms that Q2 was an unusually weak trough. Non-GAAP operating expenses were $6.3 million in Q3, essentially flat y/y but above Q2. Management attributed ongoing spend to AI benchmark and memory projects. The economic model still works if revenue scales, but it is not currently showing material operating leverage.  The consumables thesis remains central to the margin case, but evidence is still incomplete. Management said this year has been light in consumables, particularly WaferPaks, and described that as an outlier tied to customers growing into previously installed system capacity. Management also reiterated the longer-term goal that consumables should exceed 30% of revenue and lift margins as the installed base expands. That framework is plausible, especially if Sonoma and FOX deployments become more production oriented. However, the reported Q3 mix still does not fully demonstrate that operating model. The company is currently asking the market to underwrite future consumables normalization rather than showing it cleanly in current results. Management also said it expects to return to non-GAAP profitability in Q4. That is directionally reasonable, but the room for error is limited. Officially reported 9M FY26 non-GAAP EPS is already -$0.09. That means the Q4 outcome needs to cluster near breakeven or modest profitability for the full-year framework to hold. The result is that Q4 revenue timing, shipment mix, and consumables attachment matter disproportionately. A modest revenue shortfall or an adverse mix outcome could still prevent the quarter from delivering the desired profitability signal.  Liquidity risk has improved, but not because the operating model has already self-funded. Cash, cash equivalents, and restricted cash rose to $37.1 million from $31.0 million sequentially. On the call, management said the increase was primarily due to ATM issuance and disclosed that the $40 million ATM has now been fully utilized, with more than 1.13 million shares sold at an average price of $35.38. Relative to the 29.877 million shares outstanding at May 30, 2025, that implies roughly 4% dilution before considering other share movements. This is an acceptable trade if FY27 growth materializes, because it removes near-term balance-sheet pressure ahead of a potentially larger production cycle. It is less attractive if bookings remain lumpy and revenue conversion disappoints.  The fiscal-year-end change should be viewed as a modeling issue rather than a fundamental issue. The board changed the year-end from the Friday nearest May 31 to the Friday nearest June 30, which creates a 28-day transition period from May 30 to June 26, 2026 before FY27 begins on June 27. The operational business impact should be minimal, but comparability will become temporarily more difficult. Revenue phasing, growth rates, and shipment timing around the transition month will require extra care in model construction and in interpreting quarter-to-quarter changes.  INVESTMENT IMPLICATIONS The bullish interpretation of the call is that Aehr now has enough concrete evidence to support a credible FY27 reacceleration thesis. The evidence base is materially stronger than it was 90 days ago: Q3 bookings of $37.2 million, effective backlog of $50.9 million, a $14 million AI wafer-level follow-on order, an announced production order for a lead hyperscaler’s next-generation Sonoma ramp, a new silicon photonics customer with Q4 shipment timing, and incremental Sonoma manufacturing capacity. That is a much stronger setup than a generic “pipeline” story. If these orders convert on schedule, the company can move from subscale FY26 profitability to meaningfully improved FY27 volume absorption, broader market diversification, and better consumables leverage.  The bearish interpretation is that the company still has not fully solved its oldest public-market problem: translating exciting technology and large customer opportunities into smooth, predictable quarterly financials. Q4 still requires a sharp revenue increase. The highest-value wafer-level programs remain engineering-intensive and timing-sensitive. Backlog carries limited cancellation protection. Customer concentration remains high. The margin thesis still depends on a future consumables mix improvement that is not yet obvious in reported gross margin. In that sense, the call materially de-risked demand, but did not fully de-risk execution. The immediate market reaction reflected that balance: AEHR traded at $50.25 at 6:27 p.m. ET on April 7, down 3.6% from the prior close.  The most important confirmatory signals from here are straightforward. 1st, Q4 revenue needs to convert into roughly a mid-to-high-teens level for management’s qualitative tightening to be validated. 2nd, the promised near-term follow-on Sonoma order from the lead hyperscaler would further strengthen the package-level ramp case. 3rd, the top-tier AI wafer-level benchmark needs to exit redesign and move toward production order visibility. 4th, silicon photonics needs to prove it is repeatable across multiple customers, not just a single quarter spike. 5th, consumables need to become a visibly larger share of revenue if the longer-term margin algorithm is to command a premium multiple. Absent those signals, this quarter risks being remembered as a peak-bookings, delayed-revenue quarter. If they appear, the call will likely prove to have been the point where Aehr’s story shifted from narrative optionality to a tangible FY27 growth setup.  COMPANY PARTICIPANTS •Gayn Erickson, President and Chief Executive Officer •Chris Siu, Chief Financial Officer •Jim Byers, Investor Relations, PondelWilkinson •Unidentified Speaker, title not disclosed in the transcript RESEARCH ANALYSTS •Christian Schwab, Craig-Hallum •Larry Chlebina, Chlebina Capital •Mark Schuder, William Blair •Max Michaelis, Lake Street Capital Markets

@aleabitoredditApr 7, 2026 · 10:14 PM

@peedgdev Bullish on $AEHR, but not my #1 pick. I'm actually a fan of $LITE and $AAOI in the photonics space the most. Then highest upside is $SIVE. I like laser exposure the most.

@aleabitoredditApr 7, 2026 · 10:03 PM

@beauty_oe もし、シリコンフォトニクス分野などの顧客がテスト段階から大量発注（量産）の段階へと移行すれば、それは同社に数年間にわたる収益拡大をもたらすはずです。 $AEHR はまだ初期段階にあり、現在多くのハイパースケーラーが同社のテストを行っているところですので、あなたの要約は的を射ています。現在の財務状況は、顧客プロファイル（顧客層）や量産拡大の指標ほど重要ではないと私は考えています。

@investmentguru_Apr 7, 2026 · 9:21 PM

$AEHR Q3 EARNINGS • Revenue: $10.3M vs. Est. $10.9M • EPS: ($0.05) vs. Est. ($0.07) • Bookings: $37M (+54% YoY) • Backlog: $51M (+113% YoY

@aleabitoredditApr 7, 2026 · 9:10 PM

@curiosito26 It's nuanced. $POET has low downside risk because everyone already got diluted. So their cash balance sheet is extremely high ~$400m vs. MC, which cushions downside risk. So EV is like $400-500m which is great for the sole source $MRVL Celestial supplier, even if they get engineered out in a few years. I haven't seen them have more lead customer diversification yet, which is concerning. But if they do easily few billion company, and other ppl can take that risk. If you look at the names I personally pick like $AEHR ( maybe $GOOGl, $AMD, $INTC, and another optical transceiver customer). With stuff like $SIVE, it's the same $MRVL, $JBL, Ayar -> Alchip, and others at a lower MC.

@aleabitoredditApr 7, 2026 · 9:02 PM

@4intheflames Again, if mass orders happen 2027, $AEHR will re-rate today. And we'll get more indication about it in their earnings call now as they signaled "significant forecasts from our lead hyperscaler". $AAOI got re-rated 2-3x after earnings because the CEO signaled $378m/month H2 2027. I would not look at current near-term qualification orders in financials to indicate where it's heading next year.

@aleabitoredditApr 7, 2026 · 8:53 PM

@straightcash385 If $AEHR lands a major optical company like $LITE or $COHR. It’s not going to show up in current financial numbers because these are small qualification orders before mass volume ramp. This is misunderstanding how qualification cycles work. Markets are forward looking so if they state mass order ramp for 2027 in their earnings call, it’s going to get priced in now.

@aleabitoredditApr 7, 2026 · 8:37 PM

@KiakotosC It’s nuanced. If you look at current numbers alone, it’s not bullish because $AEHR is currently in their qualification cycles. It’s all about volume ramp, and they said they’re expecting significant near term volume from one of the hyperscalers (this part is bullish) This is not reflected in current financial statements but in next quarter earnings.

@kaizen_investorApr 7, 2026 · 8:30 PM

$AEHR currently down 5% after a revenue miss. Don’t mind that at the moment. Guidance looks great. - Bookings: A massive $37.2 million for the quarter, representing a very strong book-to-bill ratio of over 3.5x. This reflects heavy customer demand across both wafer-level burn-in (WLBI) and package-level burn-in (PLBI) applications. - Backlog: Stood at $38.7 million as of February 27 (Effective backlog is up to $50.9 million when including bookings received since the quarter ended). - Liquidity: The company increased its cash position, closing the quarter with $37.1 million in total cash, cash equivalents, and restricted cash (up from $31.0 million at the end of Q2). - Hyperscale Demand: Management noted they expect a significant near-term follow-on production order from a lead hyperscale customer for PLBI systems used in custom AI processors. - H2 Bookings Projection: Because of the strong demand, AEHR now expects its total bookings for the second half of fiscal 2026 (ending May 29, 2026) to hit the high end of their previously stated $60 million to $80 million guidance range. If the downtrend continues, I’ll be looking for an entry.