Where Nvidia’s $100 Billion Dealmaking Juggernaut Will Go Next
Valida Pau and Phoebe Liu (with Alix Coutures, Stephanie Palazzolo and Amir Efrati) report that Nvidia signed more than $140B of deals in two months, including a $105B credit guarantee for OpenAI leases at SB Energy’s Ohio data center project, and as of late July held nearly $100B of equity investments plus $25B of future commitments. According to bankers, lawyers and investors who have worked with it, Nvidia is now looking at robotics, self-driving technology and AI models that run on local devices. Previously unreported details: Nvidia showed interest in OpenRouter in July but never made a formal offer before Stripe won with an $8B bid; it has discussed investing another $1B in Figure, which is trying to raise at about $38B pre-money; it was recently in talks to invest about $2.5B in Thinking Machines Lab; and Perplexity’s Aravind Srinivas suggested Nvidia buy Perplexity outright before the two discussed a license-and-hire deal worth at least $20B and settled on a partnership with a roughly $3B investment at a $35B pre-money valuation. Huang pushed to close Hugging Face fast at $12.9B, more than 80 times its $150M annualized revenue, after Salesforce and others showed interest and OpenAI held early talks about a $100M investment. Nvidia’s corporate development team is led by Vishal Bhagwati (ex-HPE, Oracle), but Huang often sets prices himself. The piece flags limits: CDS spreads widened in August, and the Ohio guarantee was cut to $105B for phase one from $250B first discussed. Most deal-talk details rest on unnamed people familiar.
Why it matters
It is the clearest map yet of the most active strategic investor in AI and where its next checks will land (robotics, autonomy, on-device AI, Nemotron-adjacent model makers). Dealroom should check that Nvidia’s investor and acquirer links (Hugging Face, Poolside license-and-hire, Perplexity, Figure, Thinking Machines Lab, SB Energy, Anthropic, xAI, Reflection AI, OpenAI) are complete, and it should note Stripe’s acquisition of OpenRouter at $8B.
Executive takeaways
- Nvidia signed more than $140B of deals in two months, including a $105B credit guarantee; it held nearly $100B of equity stakes plus $25B of commitments as of late July, with $99B in cash and securities.
- Next targets, per bankers, lawyers and investors: robotics, self-driving technology, local or on-device AI models, and startups that help its Nemotron open-source models.
- New details: interest in OpenRouter without a formal offer (Stripe won at $8B); talks to put another $1B into Figure at about $38B pre-money; about $2.5B into Thinking Machines Lab; a Perplexity buyout idea that became a roughly $3B investment at $35B pre-money.
- Hugging Face at $12.9B was more than 80 times $150M annualized revenue; Huang moved fast once Salesforce and others circled.
- The strategy is to seed thousands of AI models so OpenAI and a few others don’t dominate: three customers made up 44% of sales in the six months to July.
- Limits are showing: CDS spreads widened in August, and the Ohio guarantee shrank to $105B for phase one from the $250B first discussed.
- Huang negotiates big deals himself, and corporate development is led by Vishal Bhagwati.
What The Information may be missing
The piece doesn’t quantify returns or marks on Nvidia’s nearly $100B of equity positions, or how much of its revenue comes back from companies it funds (the circular-financing question). It names no specific robotics, autonomy or on-device startups beyond Figure and Perplexity, and it doesn’t look outside the US, for example at European or Asian targets or how antitrust regulators may view the Hugging Face and license-and-hire deals. Most deal talks rest on unnamed sources. Dealroom data could show how Nvidia’s portfolio and co-investors have changed over time and which robotics or edge-AI startups already count it as an investor.
