As I discussed on the ARD #110 podcast yesterday, tech incumbents are executing totally in ‘Founder Mode’. around this AI Tech Wave. I outlined how established big tech companies are executing aggressively like tech startups. Even 33+ year old tech incumbents like Nvidia, led by founder/CEO Jensen Huang. Perhaps one of the most paranoid big tech CEOs around today.

Nvidia’s latest moves are echoing the creative acquihires and partnerships of its top customers. Who are also its top ‘Frenemies’. And since Nvidia is the one selling them the pickaxes and shovels in this AI Gold Rush, it has quite the balance sheet to do more creative things to leverage their long-term position.

Inside the NVIDIA Vera Rubin Platform: Six New Chips, One AI Supercomputer  | NVIDIA Technical Blog

The Information lays this out in “Nvidia Says It Will Take a Cut of Some Customers’ Cloud Revenues”:

  • Nvidia offers financial backstops for GPU purchases, taking a cut of cloud revenue.”

  • “Nvidia’s program empowers smaller cloud providers, reducing reliance on tech giants.”

  • “Nvidia’s guarantees on unsold GPU capacity ease financing for chip buyers.”

Nvidia Says It Will Take a Cut of Some Customers’ Cloud Revenues

Nvidia is using its powerful balance sheet to help more companies buy its expensive AI chips, and possibly make some money back too.”

I’ve already discussed Nvidia’s increasing efforts to play ‘Kingmaker’ up and down the AI Tech Stack, with everything from Neoclouds like CoreWeave in Box no. 2 below all the way to Reflection AI in Boxes 3 and 6 below.

“It’s promising to financially backstop young cloud providers that rent out its graphics processing units in exchange for a share of their revenues. The backstop would come in the form of promising to rent back unused GPUs if the companies can’t find AI developers to rent them, according to GPU cloud providers Firmus and Sharon AI, which are participants in the new program, and three executives at other firms that do business with Nvidia.”

“The deals are part of a program some people at Nvidia have dubbed the AI Compute Partnership, one of these people said. Nvidia has offered to pay a guaranteed rate for the cloud providers’ unsold GPU capacity.”

Yes, these can be characterized as ‘Boomerang’, aka ‘circular’ deals, which has been a criticism of the deal-making frenzy of late in the AI world.

“An Nvidia spokesperson also confirmed the program, which shows how the world’s most valuable firm is increasingly leaning on its financial position to become a kind of central bank to the hundreds of companies that buy its chips in bulk.”

“GPUs are typically the most expensive part of an AI data center, so guaranteeing to rent unsold GPU capacity makes it much easier for chip buyers with subpar credit ratings to get the loans they need.”

The logic typically is as follows up and down the Nvidia ‘cake’ tech stack:

“Nvidia kills two birds with one stone” in such deals, said a data center executive. If Nvidia only backstops the leases tied to facilities, “then you’re still stuck with, ‘How do you finance the GPUs?’” But if Nvidia guarantees it will pay for unused computing capacity in the facilities, “the GPUs get financed and the data center gets financed,” this person said.”

“Today, a handful of large cloud providers such as Amazon, Microsoft, SpaceX, Oracle, Meta and Google buy a majority of Nvidia’s chips. Nvidia wants to change that—in part because many of those firms are developing competing AI chips—and has spent the past few years empowering an array of upstart GPU cloud providers, including CoreWeave, to lessen its reliance on those giant customers.”

We’ve already seen Nvidia do these transactions with the largest companies in the AI landscape, the frontier model companies.

Image

“In another example of Nvidia’s effort in that regard, it recently has been in talks to provide a financial guarantee for OpenAI’s lease of a large proposed data center in Ohio that would cost $500 billion if fully built out based on today’s prices for chips, labor, power and other materials, The Information previously reported.”

““Nvidia kills two birds with one stone” in such deals, said a data center executive. If Nvidia only backstops the leases tied to facilities, “then you’re still stuck with, ‘How do you finance the GPUs?’”

“Such deals are new terrain for Nvidia, which has signaled its interest in helping finance its customers’ data center projects but has yet to commit to a large-scale project like the one in Ohio.”

A different form of these deals have already been done at a fairly robust clip.

Nvidia CEO Jensen Huang introduces a ...

“So far, Nvidia has invested billions of dollars in neoclouds in exchange for equity. And in some cases it has agreed to rent back chips from these firms, including CoreWeave and Lambda, in deals totalling billions of dollars. In Lambda’s case, The Information previously reported that Nvidia’s own researchers used the GPU servers it rented back.”

As well as the ‘backstop’ deals like Nvidia did with CoreWeave ahead of its IPO months ago.

‘Project Osprey:’ How Nvidia Seeded CoreWeave’s Rise

“Last fall, Nvidia started to do backstop deals. In September, Nvidia said it would buy all of CoreWeave’s unsold capacity through 2032, worth $6.3 billion at the time, if the cloud provider couldn’t find AI developers to rent the capacity. The move calmed investors’ worries about the levels of risk baked into CoreWeave’s highly-leveraged business model, sending its share price up nearly 30% over the following week.”

And the numbers are adding up.

“Nvidia has since guaranteed another $3.5 billion to guarantee customer data center leases in exchange for rights to buy their stock, according to a May regulatory filing about the quarter that ended in April.”

This latest ‘AI Compute Partnership’, or ACP is but the latest official incarnation and formalization of deals already underway. A trend worth watching this AI Tech Wave, for all its implications. Stay tuned.

(NOTE: The discussions here are for information purposes only, and not meant as investment advice at any time. Thanks for joining us here)