Who mentions $SHOP most on Surfaced?

@stocksavvyshay (Shay Boloor) mentions $SHOP most — 27 mentions in the past 12 months.

When was $SHOP first mentioned on Surfaced?

The earliest tracked mention of $SHOP in the Surfaced archive was on Aug 6, 2025.

How many voices track $SHOP?

19 FinTwit voices have mentioned $SHOP 120 times in the past 12 months.

What kind of voices mention $SHOP most?

$SHOP is mentioned most by Long Research voices on Surfaced.

$SHOP — FinTwit mentions on Surfaced

@kaizen_investorJul 10, 2026 · 1:54 PM

Let me dive a bit deeper in what I think is the difference in business model between $NBIS and $IREN. I've seen a couple of panels, quick fire chats, and keynotes from both companies and they are emphasizing different kind of things. I write this with a neutral stance where I see positives and negatives in both models. The first thing I want to emphasize is that the demand is real. During the conference there was never a question on whether the demand is real and/or sustainable. The main question was how to get the best ROI out of AI. Which business model and pricing model works best and where do companies find gaps in the market. A second thing is that I look at the bigger picture, I don't go into the details/calculation of any deals. I'm just connecting the dots. The Layers To understand the difference in business model, you have to be aware of the three layers both companies speak about. 1. The hardware: The first layer is the hardware layer. Basically the GPU's. I'm sure a lot can be said and done on this layer and there are differences between both but I don't see a major difference in business model between the two companies here. Let's just assume they both have access to hardware. 2. The cloud: This is the core of both Iren and Nbis. Both are selling cloud solutions. This basically means that companies can rent the GPU perfomances per hour. The big difference with Hyperscalers like AWS and CGP is that Iren and Nebius have way more access to power and compute. The goal for this power should be to train models on that raw power. It is not possible on the older datacenters of AWS and CGP because they don't have the compute to possibly do this. The compute they have available is meant to store HR/IT/Sales platforms on for enterprises. 3. The application: The application layer is what you build on top of the cloud layer. It's for enterprises who don't want to manage raw severs. You can immedatly design and install the nessecary applications. The space is very crowded here, you compete directly to competitors like Anthropic and SiliconFlow. The Hyperscalers To my understanding, up to 70% of the demand comes from Hyperscalers who are looking to buy raw compute for the Cloud. They got their own software to build on top of the cloud and they just need the compute to train their models on. In this sence both business models do not differentiate much from each other. What I did notice is that $IREN is a bit more focussed on protecting the bottom line, while $NBIS of focused on getting sold-out and getting deals done. The demand from Hyperscalers is big and they are coming to Iren and Nebius to find deals. But there are some other players in the field, Hyperscalers are actively looking for the best deals. I sometimes see new deals around with new datacenter companies I never heard about. When experts do the calculations, it looks like the gross profit of those deals aren't specatacular. Hyperscalers know this and they will use this to their advantage. Iren being this protective will defend their bottom line a lot but will automatically make you miss out on some potential deals. But the demand is only rising and the available power is shrinking, so patience might be the best deal here. The enterprises The big difference on the business models is how they are looking to approach the other 30% of the market. IREN $IREN is solely focusing on the second layer: "cloud for everyone". Ken spoke about: "Wanting to become a Hyperscaler". The deal with Mirantis is another clear statement that they want to be focussed on this segment. The thing here is that you narrow yourself a bit down to AI labs who are looking to train open source models on the cloud of $IREN. Companies like Blackbox AI, Cline, Mistral all fit this category, they are all looking for compute to train their open source models on. Mirantis helps smaller open source companies to build cloud software on top of IREN's infastructure and power. So the cloud is basically the raw compute you buy in GPU per running hour. Mirantis helps to build a software on top of this cloud. Some premium customers from Mirantis are $ADBE and $PYPL, who don't nessecary buy cloud from $IREN. I think outside the Hyperscalers, the open-source companies are the main ones $IREN is looking for but Ken talking about wanting to become a Hyperscaler and wanting to have cloud for everyone, does not really fit this picture. Hosting HR/IT/Sales platforms on your cloud has a way lower ROI than hosting training models. The raw power and compute Iren has, is an overfit for this business model. It also does not really fit the picture of protecting your bottom line, as you are going for lower ROI sales. I think becoming a Hyperscaler might be a long-term dream but the main focus right now is on the open source models. Nebius $NBIS solves this enterprise problem in a different way: tokens. They opened the Nebius token factory where you and I can start interacting with models directly via a simple API. The positive here is that you can target enterprises from all levels and even individuals. Some examples from token customers here are Revolut, $BKNG, and $SHOP. According to Rod Evans of Nvidia the ROI of selling tokens vs selling cloud is 2x higher. So big customers who have their own infastructure do not want to buy tokens but cloud directly. The smaller customers can buy tokens directly. The positive here is that you create a tailor made solution for every customer. They don't want to buy raw power, they only want to buy what they need. The customer also does not has the hustle of builing and managing raw servers. As I said before, the space here is extremely crowded. Anthropic and Siliconflow are some examples. Conclusion For all investors who follow both companies a lot, this was probably already clear. But for me this brought me some more insights in both companies. It has been said a couple of times during the conference, it is also a business model evolution. From SK Hynix to Nvidia to Cybersecurity companies. Whoever executes best on business model, will win. Demand is real, growth will come, the right business models will stay.

About this stock

Frequently asked

Related stocks