Nvidia’s $6B Poolside Deal: The Cost of Control
$6 billion, 109 job offers, and one brutal founder question: when a strategic partner licenses the engine and hires the crew, what do you still own?
Nvidia is paying $6 billion to license Poolside’s “Model Factory” and making offers to hire 109 Poolside employees involved in building Laguna. If your strategic partner can license the engine and hire the crew, you do not own as much of the business as you think.
Poolside can call itself independent after the deal. Fine. But when the world’s most powerful AI infrastructure company buys access to your engine room and makes offers to a large chunk of the crew, independence starts to look a lot like keeping the sign above the door.
The deal: $6 billion for the factory, $1 billion for the option
According to an investor letter reported this week, Nvidia has agreed to pay Poolside $6 billion to license its AI model-development software, known internally as its “Model Factory.” Nvidia is also investing $1 billion in Poolside at a $12 billion pre-money valuation.
The letter says Nvidia will make offers to hire 109 Poolside employees involved in building Laguna, Poolside’s AI model. Poolside’s founders are expected to remain and the company says the arrangement is neither an acquisition nor an acquihire.
That distinction matters legally. Commercially, I’m less convinced.
Nvidia is not merely buying a few clever models or a bundle of code. It is buying the machinery that makes the models: the systems, workflows and people required to turn vast amounts of compute into a useful AI product. That is the scarce bit.
Anyone can declare they are building an AI company. Plenty can rent GPUs. Very few can build a repeatable machine for producing competitive models, training runs, data pipelines, evaluation systems and deployment infrastructure. Poolside spent years trying to make that machine. Nvidia has decided it is worth $6 billion to get access now rather than wait to build or replicate it.
Poolside was founded by former GitHub CTO Jason Warner and Eiso Kant. In October 2024, it raised $500 million at a reported $3 billion valuation to pursue AI systems focused on software development. That was enormous money at the time. Now Nvidia is attaching a $6 billion licence payment and a $1 billion equity investment to a business that, only recently, looked like one more ambitious contender in the AI coding pile.
That is how quickly the goalposts move when your customer, supplier, investor and potential competitor are all the same bloke.
Poolside has found the uncomfortable middle ground
The romantic startup story says there are two outcomes: you build a standalone giant or you sell the company.
Reality is messier. Poolside has landed in the middle: it keeps corporate independence while giving Nvidia deep access to its technology and a major slice of the team behind it.
There are obvious upsides. The reported $6 billion payment is slated to be distributed to investors by the end of 2027. That gives Poolside’s backers a path to liquidity without waiting for an IPO, praying for an acquisition, or demanding that the company spend another five years lighting capital on fire. For a venture market that has spent too long pretending every mark-up will become a public-market exit, that is not nothing.
The $1 billion investment also buys the company time. Building frontier AI models is not a garage-business exercise. Compute, researchers, infrastructure and data are brutally expensive. Nvidia’s money and proximity can give Poolside a much longer runway than most competitors will ever see.
But don’t confuse runway with freedom.
If the most capable people working on your core product move across the table, if your key technology is licensed to the giant financing the industry, and if that giant has every incentive to integrate your advantage into its own stack, the remaining company has to answer one hard question: what is uniquely ours now?
That is not a criticism of Poolside. It is the question every founder should be asking before a cheque arrives.
Nvidia is not investing like a passive shareholder
Nvidia has become something much bigger than a chip supplier. It is increasingly acting as the banker, landlord, supplier, strategic investor and ecosystem architect of AI.
Earlier this month, Nvidia announced partnerships with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR aimed at mobilising more than $500 billion in third-party capital for AI infrastructure. Nvidia’s pitch is straightforward: compute should be financed as productive infrastructure, like other big-ticket assets with predictable demand and long-lived utility.
There is a sensible version of that story. AI needs absurd amounts of capital. Banks, private credit firms and infrastructure investors are better suited to fund giant build-outs than early-stage VCs with a slide deck and a Patagonia vest.
But there is also a concentrated-power version of the story.
Nvidia can help finance the infrastructure. Its chips sit inside it. Its software makes the chips more useful. It invests in the companies that need the infrastructure. And now, through arrangements like Poolside’s, it can license the tools used to create the models those companies run.
That is a hell of a flywheel.
It also means founders must understand the difference between a strategic partner and a gravitational field. One helps you move faster. The other pulls your entire business model into its orbit.
The overlooked angle: this may be better for investors than founders
Here is the bit people will politely avoid saying: a $6 billion licence payment that returns cash to investors can be a cracking outcome for investors while creating a much murkier future for the operating company.
Investors care about realised returns. They should. Paper valuations are nice for conference panels and LinkedIn posts; cash is what counts.
Founders and employees, however, need to care about what remains after the transaction.
Who owns the customer relationship? Who controls the roadmap? Can the company still recruit the best people after 109 colleagues leave for Nvidia? Can it sell a differentiated product if Nvidia has licensed the production system beneath it? And does the next generation of talent join Poolside to build an independent category leader, or see it as a very polished staging ground for Nvidia?
None of those questions means the deal is bad. It means the headline number is not the whole deal.
I have seen enough founders get drunk on valuation to know this: money does not solve strategic ambiguity. Usually it amplifies it.
The best founders are not just asking, “What are they paying?” They are asking, “What do they get forever, what do we lose forever, and what business is left when the press release is old news?”
Don’t copy the headline. Copy the leverage.
Most founders will read this and make the wrong takeaway. They will think: “Right, I need Nvidia on my cap table.”
No. You need an asset a company like Nvidia cannot easily reproduce.
Poolside’s leverage was not a slick AI demo. Every second startup has one of those. Its leverage was the accumulated capability behind the demo: people who understand model training, systems that turn compute into outputs, and a working process valuable enough for Nvidia to licence rather than rebuild.
That is the real lesson.
Your business needs a difficult-to-copy asset. It might be proprietary data with a legal right to use it. It might be a distribution channel your competitors cannot buy. It might be operational know-how that makes your margins structurally better. It might be a trusted brand in a regulated market. It might be a team that has solved a hard problem together for years.
If your moat is “we use the latest model,” you do not have a moat. You have a monthly software bill.
And if your only plan is to raise money at a higher valuation, you are not building a company. You are running an auction.
What this means for you
Whether you are a founder, operator or investor, use the Poolside deal as a practical checklist.
First, identify the asset beneath your product. Write down what a giant would actually pay to own or license. Not your slogan. Not your TAM. The specific thing that would save them years, reduce risk or give them a commercial edge.
Second, separate cash from control. Before accepting strategic capital, map what the investor gets: licences, data access, hiring rights, exclusivity, board influence, supply commitments and commercial options. The valuation is one line item. The rights are the deal.
Third, protect the team that creates the advantage. If your value sits in a small group of engineers, salespeople or operators, treat retention as a core business problem. Incentives, clarity, meaningful ownership and a credible mission are not HR fluff. They are asset protection.
Fourth, build a business that survives its best partner. Ask the brutal question: if our biggest supplier or investor copied us, hired our best people or changed terms tomorrow, would customers still choose us? If the answer is no, fix that before you celebrate the partnership.
Poolside has just shown that a startup can create enormous value without a clean, conventional exit. Good on them.
But the smarter lesson is not that every founder should chase a $6 billion licence deal. It is that you should build something so useful, so hard to reproduce and so strategically important that the giants cannot ignore you.
Then make sure they do not own the whole bloody thing by the time they are done helping.