KawzInvests 🦑 (@kawzinvests) on Surfaced

KawzInvests 🦑 (@kawzinvests) on Surfaced

Recent mentions

latest 50 of 72

@kawzinvestsJun 4, 2026 · 5:11 PM

$CIEN reported $1.57B in revenue, 39.5% YoY growth, and raised full-year guidance to $6.3B. Adjusted gross margin hit 44.9% and operating margin hit 19.5%, both above the high end of prior guidance, while OpEx grew only 7.7% YoY. Revenue grew 39.5% while OpEx grew 7.7%. That math tells the whole story. Backlog increased by more than $600M sequentially to $7.7B. Management said if they could deliver that backlog in 2026, customers would take it. Two cloud customers represented 34% of revenue which has some customer concentration risk. What changes it is Hyper-rail, which just received the industry's first multi-rail order from a leading hyperscaler, contributes $0 of revenue today, and is described as "hundreds of millions over multiple years" per deal with most major hyperscalers engaged. That revenue begins in 2027 and carries better margins than the current mix. I don't know where the stock goes short term. I do know the earnings trajectory has structurally reset higher. I added to my position on this dip. $CIEN $AAOI $COHR $LITE

@kawzinvestsJun 1, 2026 · 4:30 PM

Stock pickers market. Pick right or sit it out. $PENG $AAOI $BB $NOK https://t.co/mKNxpA0Hka

@kawzinvestsJun 1, 2026 · 2:02 PM

$AAOI is guiding to 930,000 units per month of 800G and 1.6T capacity by end of 2027. That's 9x the Q1-26 exit rate of 100,000 per month. Semis are already trading at 44x 2026 earnings. Price is running 2 years ahead of fundamentals across the entire complex. The setup isn't about catching the ramp early anymore. It's about owning the name before its 2027 number is the one the multiple sits on. Each 800G module needs 4 lasers. Each 1.6T module needs 8. To hit 930K modules, AAOI has to ship roughly 5.6 million lasers a month out of Sugar Land, Texas. Only 3 Western companies grow their own InP lasers. $AAOI is one of them. $LITE EML capacity is fully allocated through 2026. $NVDA spent $4B locking in priority access at Lumentum and $COHR in March. BofA models 1.6T transceiver units going from 1 million in 2025 to 117 million by 2030. Per-GPU transceiver intensity is doubling every cycle. $LITE $COHR $AAOI $NVDA

@kawzinvestsMay 24, 2026 · 6:47 PM

Photonics is not new. Fiber optic cable has crossed oceans since the 1980s. The alpha was a physics calculation made years before this chart went vertical. Copper degrades beyond 2 meters at 200G+ SerDes speeds. At 100K+ GPU cluster scale, NVIDIA modeled optical transceiver power consumption at 10 to 20x traditional cloud data centers. The industry's own operating philosophy was "copper if you can, optics if you must." That "must" moment was always coming. Clusters were going to get large enough to make it physics, not preference. Once they did, the entire supply chain from InP substrates to silicon photonic foundries had years of underinvestment priced in. The chart shows photonics vocabulary in earnings calls across NVIDIA, Coherent, Lumentum, Broadcom and six others going back to 2012. Flat for a decade. Then vertical. The alpha was reading the physics before the rest of the market did. $AAOI $COHR $LITE $AXTI $SLOIF

@kawzinvestsMay 24, 2026 · 5:01 PM

$SLOIF +546% YTD on anticipation alone. Commercial CPO has not shipped at volume yet. Soitec's 25% revenue CAGR through 2030 does not bake in a single CPO unit. That number gets reset when the ramp actually starts in 2027. For everyone who held this through the last twelve months, the returns have been career-altering. For anyone looking at $205 and feeling late, the CPO cycle has not started. $SLOIF $AAOI $COHR $LITE

@kawzinvestsMay 13, 2026 · 6:38 PM

Are you winning anon? $AAOI https://t.co/y2JsyhKfwW

@kawzinvestsMay 13, 2026 · 5:37 PM

@Frenchie_ $AAOI https://t.co/5V5NFoRBSQ

@kawzinvestsMay 12, 2026 · 7:54 PM

the $AAOI short graveyard 😅 https://t.co/tRHcqMU0rr

@kawzinvestsMay 12, 2026 · 1:52 PM

$AAOI just hit $200. What an insane year for optics. I started covering this name in the low $30's when fewer than five accounts on this app were even talking about it. I had less than 5,000 followers back then. Now I'm sitting at almost 100k. From $15 to $200 in 12 months. 13x. I'm so grateful for everyone who came along on this ride. The amount of you who DM'd me telling me you got in early and held through the chop means everything. This is what conviction looks like. The supercycle isn't over. Buckle up. $AAOI

@kawzinvestsMay 11, 2026 · 4:56 PM

@Frenchie_ it's ok let them panic. someone needs to buy $AAOI at $250😂

@kawzinvestsMay 11, 2026 · 3:51 PM

Where did all the optical panicans go? $LITE +16%. $AAOI +24%. $COHR +13%. There were so many of you last week during $LITE and $AAOI earnings. Now silence. https://t.co/YPtslmDbor

@kawzinvestsMay 8, 2026 · 12:17 AM

$AAOI Q1 is in. Here's my take after listening to the full call. Q1 came in at the low end at $151.1M, 29.2% gross margin, -$0.07 EPS. Stock got hit on the print. None of that matters. This is a forward-looking story and the ramp doesn't start until Q3. Management raised FY26 from $1B+ to $1.1B+ revenue and $120M+ to $140M+ non-GAAP operating income. Thompson said internal target is $1.2B. Actual customer demand is $1.4-1.5B. They're guiding the only number they can control, their own capacity. The mid-2027 monthly transceiver run rate went from $378M to $471M. That's ~$1.1B in annualized revenue added to the model in one quarter. 1.6T alone in 2027 is a $2B+ business per Thompson. AAOI's TTM revenue is $507M. Q3 guided up 60-80% sequentially. Q4 similar. If Q2 hits the $189M midpoint, Q4 alone exceeds all of FY25. 40%+ gross margin target by Q4. Three 10%+ customers already in Q1 before the 1.6T ramp. Three multi-year CPO/laser agreements in negotiation. Execution risk is still there. They have to procure InP substrates, hit qualification timelines, and stand up new facilities on schedule. If they execute, this is a very solid long. Sold out through mid-2027 with customer demand 30%+ above what they can produce. Added under $140. Full breakdown linked below. $AAOI $LITE $COHR $AEHR $VIAV

@kawzinvestsMay 6, 2026 · 8:03 PM

$AAOI reports earnings tomorrow and this is what I'm expecting. 2025 revenue was $456M. Management guided $1B+ for 2026 with $120M+ in non-GAAP operating profit. The business is set to more than double in a single year after running $190-250M annually from 2019 through 2024. > Q1 guide is $150-165M vs Street at $157M. CATV running hot, 800G firmware slip from Q4 pushed shipments into Q1, and they beat EPS by 91% last print. Setup favors a beat. > Q2 is what matters. Management has telegraphed it as the first profitable quarter and the start of the 800G/1.6T ramp. Above $185M with positive NG EPS and the move continues. > OFC capacity slide showed 650K combined 800G + 1.6T units/month by Q4 26, 30% above prior guidance. 930K/month by Q4 27. That's a ~$6B annualized run rate from two product lines against a $14B market cap. Full pre-earnings breakdown linked below. $AAOI $LITE $COHR $AEHR $VIAV

@kawzinvestsMay 1, 2026 · 4:09 PM

This is one of the most insane runs of the entire cycle. $AAOI up 382% YTD. From $11 to $190 in 12 months. I started buying in the low 30s when fewer than 5 accounts on this app were covering the name. The 1.6T supercycle is here. https://t.co/BXICEhPgNX

@kawzinvestsApr 29, 2026 · 4:08 PM

The State of Texas just gave $AAOI a $20.9M grant to expand its Sugar Land facility. This stacks on top of the 900,000 square foot Houston footprint they announced two weeks ago. Here's why this matters. $AAOI guided full-year 2026 revenue to exceed $1 billion on their Q4 call. That is more than double 2025's $455.7M, and it is anchored to capacity ramping to 500,000 units per month of 800G and 1.6T transceivers by year end. Now look at where 2027 is going. By 2027, $AAOI is guiding 700,000 units per month from Houston alone. That is 40% more capacity than what powers the entire $1B 2026 guide, and Houston is only one of two manufacturing hubs (Taiwan is the other). Run the math at 2027 ASPs of roughly $263 per 800G unit and $344 per 1.6T unit. If most production stays on 800G, Houston alone produces around $200M per month, or ~$2.4B annualized. If the mix shifts heavily to 1.6T (the AI generation), Houston alone produces around $225M per month, or ~$2.7B annualized. That is Houston only, at 5x to 6x current full-company revenue. CFO Stefan Murry on the Q4 call said revenue is "limited by our production capacity and supply chain, not market demand." So when the State of Texas writes them a check to scale faster, that is what is being unlocked. $AAOI is the only true Made in America transceiver player at this scale. Nobody builds 900,000 square feet on speculation. $LITE $COHR $AAOI

@kawzinvestsApr 29, 2026 · 12:21 AM

Insane hyperscaler capex. Insane token/inference demand. And the AI buildout has barely even begun. Market was shaky today on an OpenAI revenue miss headline and oil ripping on Iran fears. Both are noise when you consider capex commitments and the fact that not a single 2027-planned data center is online yet. McKinsey projects global AI data center capex hits $5.2T by 2030. 2026 alone is tracking $600B+ (up from $410B in '25). Tomorrow night $MSFT, $META, $GOOGL, $AMZN all print after close. I'll be watching every one closely: > $MSFT capex into FY27 and Azure demand against the $ 625B commercial RPO > $META revision against the $115B to $135B 2026 guide > $GOOGL TPU v7 ramp and optical content per pod > $AMZN AWS reacceleration and Trainium volume Upstream, $COHR, $LITE, and $AAOI are all aggressively expanding capacity, backed by multi-year supply agreements and $NVDA equity stakes through end of decade. Nobody funds expansions of that scale on hope. Optical TAM at $14B in 2025 to $73B by 2030 (39% CAGR). CPO going from basically zero to $15B. I've also been digging into advanced packaging names. We are still in inning 1 of this supercycle.

@kawzinvestsApr 28, 2026 · 2:34 AM

$AAOI's 1.6T pluggable optics line at OFC 26 is running on $VIAV ONE-1600 test equipment. This is the unsexy chokepoint of the optical transition. Nobody ships a 1.6T module without validating 224G SERDES, and $AAOI runs the largest US-based 1.6T transceiver production. “The demand for optical connectivity in data centers has exceeded our expectations,” said Dr. Thompson Lin, Founder, Chairman, and CEO of AOI. “We are focused on ensuring every critical function from supply chain and manufacturing capacity to quality, reliability, and customer support is positioned to scale rapidly to meet this demand. With the acquisition of these properties, we will have the footprint to expand our capacity to produce up to 700,000 units of 800G and 1.6T transceivers per month in the Houston area, while also expanding our laser fabrication capacity by around 350% by the end of 2027.” $AAOI guided 650,000 units per month by end of 2026 and 930,000 by end of 2027. That is one customer. $COHR , $LITE , $CIEN, and even $CRDO are running parallel 1.6T programs that need the same class of test capacity. As 1.6T volumes go from essentially zero to nine figures of annual shipments, the test floor scales with them. I do not know what share $VIAV holds versus $KEYS in production test. But the ONE-1600 sitting on $AAOI's OFC booth is interesting... $VIAV reports earnings Wednesday. I am watching it closely. $VIAV $AAOI $COHR $LITE $CIEN $KEYS

@kawzinvestsApr 22, 2026 · 12:51 AM

$CRDO up 74% in the past month. Credo's entire business was built on copper cables. Copper works today, but every hyperscaler is moving to optical for 1.6T and beyond. That was the bear case. One technology shift and the whole thesis breaks. The DustPhotonics deal kills that risk. Credo now owns a silicon photonics portfolio covering 400G, 800G, and 1.6T with a 3.2T roadmap. They are no longer a copper company hoping optical is far away. They are a connectivity company that sells you whichever one you need. 74% in a month is the market pricing in exactly that. $CRDO $AAOI $COHR $LITE

@kawzinvestsApr 21, 2026 · 11:35 PM

One question decides who wins the next decade of AI infrastructure. Who captures the 1.6T testing TAM? I asked a Cisco guy at GTC how $AAOI $LITE $COHR hold their margins when every new generation of optical parts does the work of four old ones. Fewer units should mean shrinking revenue. It doesn't. Every generation is harder to build and much more expensive per unit. The new high-power laser are priced at roughly four times the price of the older lasers it replaces. Fewer parts shipped, more dollars per part, margins expand. The same logic applies to the test equipment that validates every one of those parts before they ship. 400G test gear cannot test 1.6T. 1.6T gear cannot test 3.2T. Co-packaged optics adds an entirely new testing category that did not exist before. Fewer units, higher prices, new generations on an accelerating cadence. Interesting enough testing is just consolidated into a two-player market. $KEYS closed its $1.46B Spirent acquisition on October 15. The DOJ forced a divestiture of Spirent's high-speed Ethernet, network security, and channel emulation business, which $VIAV bought the next day for $425M. Three independent vendors became two. $VIAV $KEYS $COHR $LITE $AAOI

@kawzinvestsApr 17, 2026 · 11:30 AM

$AAOI just announced its Houston footprint is expanding to 900,000 sq ft with InP laser fab capacity growing ~350% by end of 2027. InP lasers are the physical bottleneck of the entire AI optical supply chain. Global fab capacity sits in dozens of facilities. NVDA pre-secured all of $LITE’s EML output in 2024 and the rest of the hyperscalers are still scrambling for supply. The new buildings will support up to 700,000 units per month of 800G and 1.6T transceivers. Laser fab, design, and assembly all under one roof. (800G at $263, 1.6T at $344), 700k units per month pencils out to roughly $213M per month on a 50/50 mix, or closer to $240M on a 1.6T-heavy mix. A $2.5B to $2.9B annualized run rate from Houston alone. Only fully vertical US producer in the space. $AAOI $LITE $COHR $AXTI

@kawzinvestsApr 14, 2026 · 7:56 PM

The rotation into CPO names is inevitable. From what I am reading, it may be coming faster than most expect. Soitec $SLOIF up 10% today. It is the only company on earth supplying the substrate TSMC, GlobalFoundries, STMicro, and Intel use for silicon photonics at volume. There is no second source. Every pluggable transceiver shipping into AI data centers today uses 50mm2 of it. Every CPO engine uses 200mm2. Same monopoly. 4x the content per device. The CPO market goes from essentially nothing today to $15 billion by 2030 (263% CARG). Soitec's Photonics-SOI revenue is already approaching €86 million in FY26 growing 25% annually. That is before the CPO ramp. We published a full breakdown of where Soitec sits in the CPO supply chain and what the content step-up means for the next two years. Link in reply. $SOI $TSEM $MRVL $LITE $COHR $AAOI

@kawzinvestsApr 13, 2026 · 9:09 PM

Credo $CRDO is acquiring DustPhotonics for $750 million cash plus equity. This is the biggest move they've made. My concern with CRDO was always copper. Their business was built on Active Electrical Cables, and the shift toward silicon photonics was the one thing that could break the thesis. This deal answers that. DustPhotonics is a fabless Israeli company with 70 engineers and a SiPho PIC portfolio spanning 400G, 800G, and 1.6T with a 3.2T roadmap. Their L3C laser coupling technology lowers power, simplifies manufacturing, and removes the EML supply chain problem that has been holding the industry back. CEO William Brennan said it plainly: "a vertically integrated connectivity platform that spans from copper to optical and from chip to cluster." That is the whole thesis now. $CRDO $COHR $LITE $AAOI

@kawzinvestsApr 10, 2026 · 2:29 PM

@CitronResearch is short $AAOI on three claims. Here is what the capacity ramp plan actually shows why he is wrong. Claim one: Oracle is the anchor and Oracle is financially stressed. The ramp AAOI published guides to 420,000 800G units per month and 230,000 1.6T units per month by Q4 2026. Oracle's entire AI buildout cannot absorb that volume. A ramp of this scale requires multiple hyperscaler customers. The concentration thesis does not survive contact with the output numbers. Claim two: 31% gross margins at 112x earnings is uninvestable. This ignores what happens to revenue mix as speeds scale. 100G sells at $20 per unit. 800G sells at $320. 1.6T sells at $640. Each generation expands ASP far beyond the throughput increase. The ELSFP module contributes essentially zero to today's revenue and scales to 400,000 units per month by Q4 2027. That mix shift changes the earnings picture entirely. Claim three: 112x forward earnings is beyond peak bubble Nvidia. That multiple is being applied to today's earnings base on a company mid-ramp. By Q4 2026 the capacity guide implies roughly $303 million per month in revenue. The multiple compresses faster than any static comparison accounts for. The real bear case is manufacturing execution. If the fab ramp slips, the numbers slip. Citron is not making that argument. They are using today's earnings to value a company whose revenue mix will be unrecognizable in two quarters. $AAOI $LITE $COHR $NVDA

@kawzinvestsApr 9, 2026 · 8:45 PM

@crux_capital_ Still remember when it was just us and a handful of others covering $AAOI back in the 30s, little did we know...

@kawzinvestsApr 8, 2026 · 10:04 PM

This is a multi-year cycle and every layer runs. Transceiver testing is happening now. CPO testing is the next leg. InP laser burn-in after that. The market keeps repricing the same supply chain as each wave becomes visible. We're in wave one. $AEHR $VIAV $COHR $LITE $AAOI

@kawzinvestsApr 8, 2026 · 7:41 PM

Now up almost ~50% in 5 days $AAOI https://t.co/tk8MIut4Ur

@kawzinvestsApr 8, 2026 · 2:34 PM

$AAOI is up 37.66% in 5 days. The OFC numbers we posted are why. ~$64 million per month in current revenue. ~$303 million per month by Q4 2026. ~$701 million per month by Q4 2027. These are not annual figures. The mechanism is ASP mix shift, not volume growth. The ELSFP module goes from effectively zero today to ~$256 million per month by Q4 2027. Each product generation that ships carries materially more revenue per unit than the one before it. The bear case is manufacturing execution. That has not changed. But even at 70% of guidance, a $9B market cap does not reflect where this revenue mix is going. The thesis has not changed. $AAOI $COHR $LITE

@kawzinvestsApr 2, 2026 · 9:33 PM

$AAOI received a $71M order for 800G transceivers from ANOTHER Major Hyperscale Customer In the last 3 weeks alone: - March 9: First-ever volume order for 1.6T transceivers - March 23: New 800G volume order - April 2: Upsized $71M 800G order All from the same undisclosed "major hyperscale customer." That customer is almost certainly $ORCL. Oracle is mid-execution on the largest cloud buildout in its history. The order sizes, the timing, the acceleration. It all fits. The CEO said it himself today. "We also recently shipped the first 10,000 units of an 800G single-mode transceiver order to another hyperscale datacenter customer." Two hyperscalers. Both ordering now. Both accelerating. This is a multi-year cycle. 400G became 800G. 800G becomes 1.6T. 1.6T is the last stop before the industry moves to Co-Packaged Optics, CPO, where optics get embedded directly onto the switch chip. Less power. More bandwidth. The hyperscalers have been pushing toward this for years. $AAOI already has their first 1.6T volume order. They broke ground on a 210,000 sq ft factory in February. Their anchor customer keeps coming back and ordering bigger every single time.

@kawzinvestsApr 2, 2026 · 4:31 PM

@michaelsikand $AAOI up 15% today insane

@kawzinvestsMar 27, 2026 · 7:42 PM

AI optical networks have a zero-tolerance policy for failure if a single 800G or 1.6T link underperforms, an entire multi-million-dollar array of GPUs sits idle waiting for data. This is why Viavi Solutions $VIAV is a critical hidden tollbooth for the AI buildout. As a global leader in test, measurement, and optical assurance, their technology secures and validates the infrastructure of these massive data center ecosystems. Because moving to higher speeds and Co-Packaged Optics introduces severe signal integrity issues like jitter and noise, testing isn't just a one-time hurdle. It is a massive, overlapping supercycle that must scale directly with optical complexity to prevent catastrophic bottlenecks. $AAOI $COHR $LITE $SLOIF

@kawzinvestsMar 27, 2026 · 2:06 AM

Building the next-generation optical chips for AI (800G and 1.6T) is incredibly difficult. At these extreme speeds, signals easily distort, and just one bad chip can crash a massive, multi-million-dollar AI cluster. Stop trying to guess which optical manufacturer (like $AAOI or $COHR) will win the final cloud contracts. Instead, buy the companies that make the testing and inspection equipment. These testing companies are "winner-agnostic tollbooths." Because there is zero room for error in AI data centers, every manufacturer is forced to buy this testing gear right now to validate their new production lines. The testing companies get paid first, no matter whose chip ultimately wins. $VIAV & $AEHR: Provide the essential fiber network testers and "burn-in" equipment to ensure these parts actually survive in the real world $KEYS (Keysight): Sells the advanced oscilloscopes engineers absolutely need to test 1.6T signals in the lab. $CAMT (Camtek): Acts as a "tax on chip complexity" by inspecting the microscopic layers of advanced AI chip packaging before they are stacked.

@kawzinvestsMar 26, 2026 · 3:19 PM

We are currently in the middle of an unprecedented, multi-year compounding supercycle Here is how the distinct optical cycles actually overlap: 400G: The foundational baseline that is mainstream today 800G: The current workhorse. Bears think this cycle peaks soon, but 800G unit volumes are projected to continue growing and won't actually peak until 2028 at ~84 million units 1.6T: The next massive inflection point. Expected to become the standard for new AI cluster buildouts, volume ramping begins in 2026/2027 to feed next-gen GPU clusters, driving a heavy shift toward Silicon Photonics 3.2T: The ultimate pluggable endgame. This technology is already in development and expected to hit meaningful commercial ramps by 2028+ before Co-Packaged Optics (CPO) fully takes over the backend Historically, optical cycles replaced each other and pricing power died quickly. But in the AI era, network bandwidth demand is doubling every two years. These deployments aren't cannibalizing each other they are heavily overlapping and stacking on top of one another to build a $45.4 billion transceiver market by 2030 $LITE $COHR $AAOI $CIEN $TSEM

@kawzinvestsMar 25, 2026 · 1:28 PM

Why are traditional pluggable optics doomed in the AI era? Pushing data through copper traces to a pluggable module causes 22dB of signal loss and requires power-hungry DSPs that burn ~30W per port. For @NVIDIA's upcoming million-GPU AI factories, this is a thermodynamic nightmare. The inevitable solution is Co-Packaged Optics (CPO) By integrating the optical engine directly onto the switch substrate, CPO slashes power consumption by 3.5x and cuts signal loss down to just ~4dB. But this transition won't happen overnight... We are entering a multi-year supercycle where 800G, 1.6T, and 3.2T optical pluggable deployments will heavily overlap and compound through 2030 before CPO fully takes over. Why shouldn't you be bearish on pluggable transceivers yet? Meaningful CPO volume inflection will not occur until ~2028. Until then, pluggables offer unmatched field serviceability, flexible "pay-as-you-grow" bandwidth, and a mature, standardized ecosystem that operators rely on. Because of these advantages, we are entering a multi-year supercycle where 800G, 1.6T, and 3.2T pluggable deployments will heavily overlap and compound So, who actually wins this transition? You have to look at the component supply chain: $LITE, $COHR, $AAOI, and $SOITEC. NVIDIA knows InP lasers are the chokepoint, which is why they committed $4B to secure capacity from $LITE and $COHR. $LITE is the exclusive supplier of ultra-high-performance (UHP) CW lasers for NVIDIA's Spectrum-X CPO switches. $COHR is ramping a massive 6-inch fab in Texas to supply the External Laser Sources (ELS) needed for CPO. • $AAOI is vertically integrated (making its own lasers in Texas, avoiding China risk) and is currently developing 400mW pump lasers specifically for CPO applications. But the ultimate hidden monopoly is Soitec ($SOI). Every silicon photonics chip requires a specialized "Photonics-SOI" wafer. Soitec holds a >95% market share. The kicker? Moving from a standard pluggable to a CPO switch requires 4x the wafer content (scaling from 50mm² to 200mm²). When CPO ramps, their revenue per port quadruples.

@kawzinvestsMar 24, 2026 · 5:59 PM

INSANE $AAOI is up 17.7% Same stock people were dumping three weeks ago over ATM dilution and insider sells. Here is what they got wrong. The fear was understandable. AAOI has burned people before. One customer drove most of their revenue in 2017 and when that relationship softened, the stock collapsed. They promised 400G traction in 2022 and were still in early 800G sampling a year later. The share count has ballooned over time. So when management raised $250M through an ATM and then set up another one, and insiders started selling, people assumed the worst. But the capital raise was not a warning sign. It was construction funding. AOI broke ground on a 210,000 square foot manufacturing facility in Sugar Land, Texas this February. That plant is expected to be operational by summer. They are not raising money to stay alive. They are raising money to build capacity fast enough to keep up with what customers are actually asking for. Since then the orders have come in back to back. March 9th, a major hyperscaler placed a $200 million order for 1.6T transceivers. Deliveries start in early Q3. Today, that same customer placed a $53 million 800G order specifically for GPU cluster buildout. Two orders from one customer in under three weeks. The firmware issue that was blocking the 800G ramp is resolved. Management confirmed it publicly. Capacity targets were revised upward at OFC from 500K to 650K units per month by year end. Full year revenue guidance of over $1 billion stands. They exit 2026 expecting three separate hyperscalers each above 10 percent of total revenue. The customer concentration problem that destroyed them in 2017 is being structurally fixed in real time. This is one of my largest positions. The orders exist. The facility is being built. The customer pipeline is expanding. The 2027 margin targets assume gross margins reaching 35 to 40 percent as the product mix shifts toward higher speed transceivers. The risk is execution. It always is with AAOI. A qualification delay or a customer pause can move this stock violently in either direction. But if they deliver, this is nowhere near fairly valued. That gap is why I have not sold a single share. Wrote a full breakdown on Substack. Link in bio and will drop it in the comments. $AAOI $LITE $COHR $SNDK $MU

@kawzinvestsMar 24, 2026 · 12:12 AM

$AAOI just landed a $53M+ order for 800G transceivers from a major hyperscaler. This comes right after last week's 1.6T order from the same customer.

@kawzinvestsMar 23, 2026 · 10:43 PM

The Nvidia GTC report just dropped on Asymmetrical Bets. 36 hours on-site. Private press briefing with Jensen. One-on-one with Nvidia's SVP of Networking. A closed-door session with Meta, CoreWeave, Microsoft, and SoftBank that most people at GTC did not know was happening. We asked Jensen directly why he wrote $4 billion in checks to two photonics companies. We got the answer. The report covers everything: CPO, HBM, inference infrastructure, continental-scale compute, and our #1 stock pick on co-packaged optics a name most people have never heard of. Link in below & in bio. $NVDA $AAOI $COHR $LITE $CIEN $MU $SNDK $TSEM

@kawzinvestsMar 19, 2026 · 9:35 PM

NVIDIA GTC is wrapping up. Starting today, @michaelsikand and I are going into full lockdown every hour we have goes into turning what I witnessed this week into the most thorough AI infrastructure breakdown we have published. @nvidia extended an invitation this year. What I saw on-site confirmed several theses I have been building for months and introduced structural setups I had not fully modeled. The report covers all of it: silicon photonics and Co-Packaged Optics, HBM memory demand, inference compute scaling, network scale-up and scale-across architecture, continental and orbital data center buildouts. Every trade thesis. Every structural setup. Where the alpha is heading. Follow and enable notifications. Asymmetrical Bets Substack Report drops soon. $NVDA $AAOI $COHR $MU $LITE $CIEN $SNDK $TSEM

@kawzinvestsMar 18, 2026 · 6:12 AM

Building a memory fab takes 4 years. Building a photonics fab takes 9 months. There is a MASSIVE difference between the build out for Photonics vs Memory A memory fab is a precision lithography operation. You are packing billions of transistors at single-digit nanometer nodes. EUV tools alone take 12-18 months to procure and calibrate. The yield ramp after that takes years. The bottleneck is physics and it cannot be compressed. A photonics fab is an INTEGRATION PROBLEM. You are building devices that manipulate light, not electrons. Indium phosphide. Optical waveguides. Alignment tolerances measured in nanometers of coupling efficiency, not transistor density. No EUV required. The practical timeline difference: Samsung Electronics, $MU, SK Hynix 3 to 5 years from groundbreak to meaningful output. The lithography learning curve is non-negotiable. $AAOI with an existing warehouse 9 months. Not because construction is faster. Because they skip the construction problem entirely. Cleanroom retrofit, tool installation, and process bring-up run in parallel. Most companies do these sequentially. AOI does not. $AAOI has a massive automation advantage AOI runs internal testing systems at 20x the throughput of standard industry equipment. Their product platforms are standardized to the point where each new production line is not a new engineering problem it is a deployment. When they enter a new facility, they are not figuring out the process. They are executing a template they have already optimized across years of production in Taiwan. That is exactly what is happening with their new Texas facility. AOI is not building something new. They are replicating the same factory format, tooling layout, automation systems, and process templates that are already running and yielding in Taiwan. The institutional knowledge, the yield data, the calibration baselines all of it transfers. A semiconductor company standing up a new node from scratch has none of that. AOI walks in with the answer key. Vertical integration across lasers, PCBA, and final assembly means there is no external dependency introducing variance into yield. They own the entire feedback loop from wafer to finished transceiver. That matters because of what the real bottleneck actually is. Most people stop the analysis at fab timelines or InP supply. Both are real constraints. Neither is the hardest part. The hardest part is thermal qualification. A transceiver operating inside a hyperscaler switch runs continuously. These switches need to operate at full load 24 hours a day for the unit economics to justify the infrastructure spend. If the switch is down, the compute behind it is idle. At the scale hyperscalers operate, idle compute is not an inconvenience it is a direct hit to the return on billions of dollars of capex. The failure mode that defines vendor selection is thermal. Transceivers generate heat. Heat degrades the laser. A degraded laser causes signal loss. Signal loss in a switch port takes that segment of the switching fabric offline. Hyperscalers do not tolerate partial switch failures they replace the vendor. This is why qualification cycles are the longest stage of the entire ramp, not manufacturing. Hyperscalers test interoperability, sustained thermal performance, and reliability under continuous full load before committing volume. A vendor that cannot demonstrate 24/7 thermal stability does not get the contract regardless of how fast they built the factory. AOI's vertical integration is a direct solution to this problem. Because they control lasers, PCBA, and assembly in-house, they control the thermal envelope of the finished product end to end. Competitors are integrating components from separate vendors and discovering thermal variance late in qualification. AOI is designing the thermal system, not assembling one from parts. Their automated testing infrastructure means thermal issues surface during production, not during the customer's qualification cycle. That compresses the single longest stage in the entire ramp. And because the Texas facility is a copy of Taiwan, that thermal system arrives pre-validated. They are not learning how to build a thermally stable transceiver in Texas. They already know. They are just doing it closer to the customer. Memory Manufacuturing bottleneck = lithography Photonics Manufacuturing bottleneck = thermal qualification The structural thesis is sound. But there is always a layer of entropy no model accounts for. Execution risk does not disappear because the framework is good.

@kawzinvestsMar 18, 2026 · 2:43 AM

Covered the full $AAOI thesis in depth on Substack: https://t.co/q7tAnU0vtE

@kawzinvestsMar 18, 2026 · 2:03 AM

Applied Optoelectronics $AAOI released their capacity ramp plan. We are in Q2 2026, which means this timeline has already started. Current monthly revenue: $64 million. By Q4 2026, guidance puts that at $303 million per month. By Q4 2027, $701 million per month. These are monthly figures, not annual. Here is why those numbers are achievable. Transceiver pricing does not compress as speeds scale. It expands. 100G sells at $20 per unit. 400G at $64. 800G at $320. 1.6T at $640. Each speed generation commands a price that far outpaces the throughput increase, because the engineering complexity at each step grows significantly. The revenue mix tells the real story. Today, 800G contributes $44 million per month and 1.6T contributes $6.4 million per month. By Q4 2026, 800G reaches $134 million and 1.6T reaches $147 million per month. The ELSFP module, which is not in the revenue mix today, scales from $3.2 million per month in Q4 2026 to $256 million per month by Q4 2027. This is ASP mix shift, not volume growth. Each product cycle that ships carries materially more revenue per unit than the one before it. The execution risk is the manufacturing facility ramping on schedule. Even at 70% of guidance, the revenue trajectory is not priced into the current valuation $AAOI $COHR $LITE

@kawzinvestsMar 17, 2026 · 7:49 PM

NVIDIA $NVDA CEO JENSEN HUANG JUST SAID: "We need many many TRILLIONS OF DOLLARS OF CAPACITY...We work with EVERY MEMORY VENDOR, BECAUSE WE USE SO MUCH MEMORY. We are going to work with every connector vendor, EVERY SINGLE SILICON PHOTONICS COMPANY." $AAOI $COHR $LITE $CIEN $FN $MU $SNDK $EWY

@kawzinvestsMar 16, 2026 · 8:24 PM

Jensen just ended the copper vs. optical debate at GTC. The answer is not one or the other. Nvidia is scaling both simultaneously copper for near-interconnect density, co-packaged optics for long-distance scale-up, and traditional optical for scale-out. $LITE $COHR $AAOI $CIEN $TSEM

@kawzinvestsMar 16, 2026 · 7:57 PM

“We need a lot more capacity for copper, we need alot more capacity for optics” - Jensen @ GTC 2026 $LITE $COHR $AAOI $CIEN $TSEM https://t.co/DZRyaTL6rL

@kawzinvestsMar 16, 2026 · 5:15 PM

Six companies just standardized the physical layer of AGI infrastructure. That is what happened on March 12th. $AMD $AVGO $META $MSFT $NVDA and OpenAI formed the Optical Compute Interconnect MSA. A single agreed specification for replacing copper with light inside AI data centers. Copper has a reach problem. At the scale these clusters operate, electrical signals degrade, power costs compound, and cluster size gets capped by the physics of the wire itself. The architecture hits a wall before the model does. OCI solves it at the source. Multiple wavelengths multiplexed onto one fiber. 200 Gbps per fiber at generation one. 800 Gbps at generation two. 3.2 Tbps on the long-term roadmap. Pluggable, on-board, and co-packaged optics all within spec. The part worth sitting with: every major hyperscaler just locked onto the same supply chain simultaneously. Suppliers now have a stable multi-generational target backed by the largest infrastructure budgets in the world. OpenAI's Head of Hardware said AGI-scale systems cannot be built without the bandwidth this enables. That is the procurement thesis stated plainly by the buyer. $AAOI $COHR $LITE $CIEN $FN $AVGO

@kawzinvestsMar 16, 2026 · 4:28 PM

GTC is live and the photonics basket is responding. $TSEM leading the board at +11%. $CIEN +5.1%. $LITE +4.6%. $MRVL +4.5%. $COHR +3%. The names catching the largest bids are the ones closest to the CPO and silicon photonics supply chain. The market is repricing the procurement schedule in real time. $AAOI is the notable lag down 1.76% while the rest of the basket runs. Divergence worth watching. $TSEM $CIEN $LITE $MRVL $COHR $GLW $AAOI

@kawzinvestsMar 15, 2026 · 11:25 PM

@DP3188144159988 BofA projects the optical TAM to grow from $14 billion in 2025 to $73 billion by 2030 at a 39% CAGR the returns you see in $LITE, $COHR, and $AAOI are the market beginning to price a buildout that is still in early innings

@kawzinvestsMar 15, 2026 · 10:52 PM

Photonics has quietly become one of the hottest sectors on FinX right now. $LITE +800%. $COHR +262%. $AAOI +360%. $GLW +175%. $CIEN +400%. This sector pushed me out of my comfort zone and forced me to put in countless hours of research to actually understand it. If you are a beginner in photonics or just want to build a real foundation in this space, this video is a MUST WATCH. Genuinely wish I had something like this when I first started digging into it.

@kawzinvestsMar 13, 2026 · 8:00 PM

$AAOI I have been seeing a lot of bearish takes on this name recently. I am still long. The stock is down 26% from its all-time high of $127. The bear case has gotten louder. Here is why I think the bears are missing the asymmetry. At $94, AAOI trades at roughly 8x FY2026 consensus revenue of $856M. Management guided above $1B, which brings that multiple down to around 7x. LITE trades at approximately 14.5x forward revenue. COHR trades at roughly 7x. AAOI is growing faster than both. Now look at what half-execution on 2027 targets actually means: Management said monthly transceiver revenue reaches $378M by mid-2027. That annualizes to $4.5B. At current prices you are paying roughly 1.6x that run rate. If they deliver half of that you are still paying well under 3x revenue for a company growing triple digits with an in-house InP fab. That does not stay unpriced. The Q4 results were real. Revenue of $134M, up 33.9% year over year, beating estimates. EPS of -$0.01 vs consensus of -$0.12. 1.6T volume order secured from a major hyperscale customer. Shipments starting Q3 2026. Three hyperscale customers each expected to exceed 10% of FY2026 revenue. The single customer concentration risk the bears have cited for years is structurally gone. The company is investing $300M to triple InP laser capacity in Texas by mid-2027. They are presenting at OFC on March 17th. Jensen speaks Monday. The bear case requires management to miss badly on every single target simultaneously. The bull case only requires them to be half right. Full breakdown in the report. Link in bio.

@kawzinvestsMar 13, 2026 · 5:00 PM

$AAOI $LITE $COHR $POET $CIEN NVIDIA just committed $4 billion to lock up laser supply from $LITE and $COHR. $AAOI makes the exact same category of component in-house and trades at 3.6x 2027 revenue vs $LITE at 14.6x. $LITE holds 50% market share in high-speed EML lasers and has $400M+ in OCS backlog that the Street has not modeled. $COHR is the only company that can supply both the silicon photonic engine and the laser source for NVIDIA's CPO architecture. They have already secured what management called an exceptionally large CPO purchase order. The optical AI market goes from $14B today to $73B by 2030 at a 39% CAGR. We cover all of it in depth in our latest report. There will be money to be made and money to be lost next week. This is a must read. For Photonic Investors

@kawzinvestsMar 13, 2026 · 11:50 AM

$LITE $COHR $AAOI $POET $CIEN Next week we have NVIDIA GTC and OFC happening in the same week. GTC is where Jensen Huang will talk about Silicon Photonics, Co-Packaged Optics, and AI networking infrastructure. OFC is where the entire optical industry converges 800G, 1.6T, LPO, CPO, OCS. Every technology that matters to this supply chain will be discussed in one place. We just published an institutional grade deep dive on the photonics supply chain to prepare you for both events. This is a must read before next week. Link in bio.