Shimmy Says: Nvidia Is the Beating Heart of AI
Nvidia is the beating heart of AI — and in this Shimmy Says, Alan Shimel explains why the recent “Could Nvidia default?” clickbait misses the far more consequential story. After $96.2 billion in quarterly revenue, nearly $60 billion in net income, a forecast for approximately 70% growth next year, and a reported $12.9 billion acquisition of Hugging Face, Nvidia is no longer just selling AI chips. It is becoming the supplier, financier, marketplace and circulatory system of the entire AI economy.
Earnings demolish the default premise
Nvidia generated $96.2 billion in revenue in a single quarter, up 106% year over year. Net income hit $59.7 billion. Data center revenue was $89 billion, up 117%. Gross margin held at 75%. Guidance for the current quarter is approximately $108 billion — with no assumed China data center revenue. Management now forecasts roughly 70% revenue growth in fiscal 2028, versus the ~45% Wall Street expected. That is not the financial profile of a company approaching default.
Demand is being locked in years ahead
AWS and Nvidia announced plans to deploy 2 million additional Nvidia GPUs across AWS infrastructure in 2027–2028, on top of the more than 1 million planned to begin in 2026. Anthropic reportedly signed a ~$45 billion, six-year compute agreement with Nscale, running on Nvidia Vera Rubin systems starting in late 2027. Hyperscalers and frontier labs are reserving Nvidia-based capacity years in advance — reinforcing both the durability of demand and the concentration risk at the center of the story.
Why Hugging Face is not a software acquisition
Hugging Face reportedly generates about $150 million in annualized revenue — meaning Nvidia is paying roughly 86 times revenue. That price only makes sense if Nvidia sees Hugging Face as the demand and distribution layer for the open-model economy. Sitting between millions of developers and an enormous library of models and datasets, Hugging Face could direct workloads toward Nvidia hardware, help find customers for unused capacity, and turn Nvidia’s claim that its compute is “fungible” into an operational reality.
The balance sheet becomes a moat
Nvidia disclosed approximately $366 billion in future commitments — $279 billion in supply and capacity commitments alone, up from $119 billion three months earlier, primarily to secure memory. Nvidia said AI labs supported through its balance sheet could account for roughly one-quarter of its business next year. Its financial support for the AI ecosystem is no longer a side program — it is becoming material to Nvidia’s growth.
The risk is concentration, not default
AI labs depend on Nvidia for compute. Neoclouds depend on Nvidia systems and financing. Infrastructure developers depend on its commitments. Memory suppliers build around its forecasts. Hyperscalers depend on Nvidia to expand their AI businesses. Developers may increasingly depend on an Nvidia-owned marketplace. Nvidia can almost certainly absorb an individual customer failure. The more serious scenario is several assumptions failing together — and that would ripple through the entire AI economy.
The indispensability trap
Becoming indispensable is the ultimate competitive achievement. But indispensability pulls responsibilities and risks toward you. Nvidia has cemented itself as the indispensable infrastructure company of the AI era. Nvidia is the beating heart of AI. The risk worth watching is not whether that heart suddenly stops — it is how much of the AI economy now depends on it never missing a beat. That’s what Shimmy says.
Transcript
This week, a headline asked a question: could Nvidia default? Then the numbers came in. 9 billion for HuggingFace, 366 billion in future commitments.
Nvidia isn't the company that might fail, it's the one everything else now depends on. That's what Shimmy says. Hey, everyone.
It's Shimmy Says. It's Thursday, must be time for Shimmy Says. Thanks for joining us.
I wanted to talk today. Well, a lot of people are talking about it today. Nvidia reported earnings yesterday, and a bunch of other stuff came out after that.
" And the author made some points that I think in another time, another place might hold more water. Nvidia is backstopping this big data center in Ohio. They've put their balance sheet up and they're doing a lot of, you want to call it circular financing, where I give OpenAI money, OpenAI buys back stuff from me.
They've been doing it with everyone. There's a lot of commitments there, and on a normal-sized company, they'd be significant. 5 trillion plus valuation, they're significant.
And I guess if you were a credit manager or someone who's evaluating that, you might want to take this into consideration. But the fact of the matter is, guys, this is Nvidia we're talking about. And, especially after yesterday's news and reporting, you go back and look at that article, and the article kind of looks like it's something you'd see on The Onion or something, right?
Nvidia is not defaulting. It was a minuscule chance of a default before this, but it's certainly not defaulting now. I think the real question we should be asking is, what is Nvidia becoming?
Because it's not just a chip supplier anymore. And is the entire AI economy based upon them, where if they have a hiccup, it's going to be felt all the way downstream. So let's talk first about Nvidia earnings.
Man, you want to talk about dream earnings? This is dream earnings. 2 billion.
That's almost $100 billion quarter, 400 billion. Revenue up 106% year-over-year, and that's not because last year was a bad year. Last year was a great year, and it still doubled this year.
7 billion to the bottom line. Data center revenue, just by itself, up 117%, $89 billion, a gross margin of 75%. That's dream territory.
This is a company that's making about a billion dollars a day, maybe more. More than a billion a day. And their forecast for next year is crazy.
They're forecasting 70% year-over-year growth. And by the way, their forecast assumes zero revenue from China. They've written China off as a market.
Those are impressive numbers in any time. They just blew Wall Street away. Wall Street was looking for maybe a 45% growth rate, and they got 70.
And here's another crazy thing. You know the only thing stopping Nvidia from making more money? They can't get any more components.
They can't buy any more memory. They can't buy the components they need because the market would buy them if they could. That's crazy when you think about it.
The only thing stopping them from selling more is they just can't make any more because they can't get enough supplies for it. So my friends, this isn't any company looking for default. And it's not just their earnings.
Nvidia's at the top of their game, like a well-oiled machine. They announced right after earnings deals came out, AWS bought something like two million GPUs in the next year or so. Because their product is so scarce that companies are lining up to buy out future production.
Anthropic did another similar deal. Nvidia is going to sell out whatever they can produce over the next year or two. It's crazy.
And it's not just the Nvidia GPUs. That also brings Vera, their CPUs, into the mix as well. So they're expanding that beachhead of what silicon they sell.
But if you still think Nvidia is just about silicon, think again, because the other big news broke yesterday, too, after earnings. Around 8:00 last night, I started seeing it. It's not been confirmed yet, but it's the worst-kept secret in technology.
They're buying HuggingFace for about $13 billion. Now, this deal in and of itself is a little crazy. HuggingFace had about $150 million in revenue.
So think about it, $150 million company for $13 billion. That's quite a multiple. But it's really not about the multiple like the VCs always like to tell you.
This was a strategic deal. And what was strategic about it is HuggingFace becomes the distribution point for all of these AI models that are running. Now, NVIDIA mentioned something in their earnings yesterday, that what they're building is fungible computing.
If OpenAI doesn't want to use the Ohio data centers, Anthropic will. If Meta doesn't want to use the GPUs, NVIDIA can plug someone else in. These neo clouds that are really kind of GPU-as-a-service kind of things can all be moved around.
But how would you know what to put where, where there's a need, where there's a want? HuggingFace. HuggingFace gives NVIDIA the insight and the ability to truly move to fungible computing.
Now, there are people who are a little upset with this Hugging Face deal, right? Because it moves NVIDIA into this seat where they could determine the success or not success of various models. HuggingFace has, I think, two million different models in there.
So what it is, what you're dealing with there is, if you're the company controlling what models are being brought to people's attention, what models are available, and in turn, you have economic interest in those models, well, you kind of lose some of that independence and credibility. And so, there are people asking questions about it, and they're right to ask the questions. They're right to ask the questions.
The market will take care of this, right? Because people can recreate a HuggingFace. There could be other similar sites that do what HuggingFace does.
But right now, NVIDIA will potentially control that as well. So think about this now. They've got the silicon GPUs and CPUs.
They build whole racks. They're the financial backstop for many of these huge AI factory data centers going up. They have circular financing and equity agreements in the largest AI companies.
They partner strategically and also sell to AWS, Google, Microsoft, all of the hyperscalers, Meta, and it goes on. They partner, their business development team is cranking on all cylinders. They partner with just about everyone.
They are using their balance sheet. They've weaponized their balance sheet. It's become a moat that people can't cross.
You've got CUDA, right? As this whole stack and its open-source elements as well that people are building the next AI stack on. So NVIDIA is no longer a silicon company.
They are an integrated vertical play. The likes of which I don't know if we've seen in this world in a very long time, right? In such an indispensable sector that's happening right now, right?
They are literally the beating heart of the AI industry. Now, as I've mentioned on here before, I've got a book coming out next month called "The Indispensability Trap," and it talks about exactly these kinds of scenarios. When something like this becomes so indispensable, when a company like this becomes so indispensable, look at the electric utilities, look at AT&T back in the day, look at the railroads, look at the internet, the government almost can't help but to come in and regulate this.
Can they allow one company to just control that whole vertical integration? Now, the fact of the matter is, in the US, look, NVIDIA's kind of bought a lot of favor with this administration, right? They invested in Intel.
They do with the Chinese government embargo, they play nicely there. I don't see this government, this administration, cracking down on NVIDIA, but I could see the EU doing it. I could see others doing it, right?
Because this, I don't throw the M word around easily, but this could very well be. It's almost like, remember when Microsoft propped up Apple for a few years, so they'd be a valid competitor. You might see NVIDIA propping up Intel, if not AMD, just to have a competitor.
The fact of the matter is AMD right now seems content to play second fiddle. Second fiddle is good enough for them And I'm sure Nvidia's happy to have an actual competitor to thwart off talk of monopoly and that kind of thing. But I think this HuggingFace thing is going to bring up an issue, which is neutrality, and it'll be interesting to see how the market reacts to it, because you know the market's fickle.
Today, they love you, tomorrow, they hate you, and the next day, they love you again. Right now, everyone loves Nvidia, but when people see this much power concentrated in one player like this, everyone gets a little green with envy. Everyone gets a little jelly apple.
Everyone gets worried. Who's going to pull the rug out from them if we don't play nice with this company? And that's Nvidia today.
It supplies the computes, the components, the infrastructure. The consumers are using it. The enterprises are using it.
It's going to control HuggingFace and access to models. It lends money to the Neo clouds so the Neo clouds can buy chips from Nvidia, and then they have to give Nvidia a portion of the rental revenue that that Neo cloud rents out the chips for. What a great business model.
Could it go on for foreseeable future? I think so, but I do think we're going to see the indispensability trap kind of kick in here shortly. Too many people are depending on Nvidia.
Too many segments of this market. The old too big to fail. It's too big to fail, but it's also too big to just be left alone on its own, I think.
And I think you'll start seeing governments, maybe the next US administration as well as the EU, take some action if this doesn't change. But in the meantime, you've got to give Jensen Huang and the Nvidia team all the credit in the world. They are a company that took an opportunity, have run with it, are firing on all cylinders, are using every tool and smart move they can make, and it is paying off.
It's paying off handsomely. I think when you have a company that the only question about them being able to sell more is their suppliers can't supply more, that kind of says it all. As long as what they're selling is scarce, scarcity trumps commodity every time.
But when it is commodity, you do have the indispensability trap, and I do think that's going to come in here. So is Nvidia going to default on any of its obligations? Not today, bucko.
But that's even the wrong question to ask. Nvidia is the indispensable heart of the AI infrastructure, and that heart's not going to stop beating. But what happens if it does skip a beat, though?
Those shockwaves are going to reverberate right to the edges of this, and it'll be interesting to see what happens. But in the meantime, enjoy the ride. Hey, if you were in early on Nvidia, good for you.
I think that stock still has a lot more to go. If it's going to grow 75% year over year, it's probably underpriced. But don't take my stock advice.
I'm still here working instead of retired. " Shimmy says, Shimmy says, Shimmy, Shimmy, Shimmy says, ask me almost anything.
