From Hugging Face to Figure Robots: How NVIDIA Is Betting Billions on the Future of the AI Ecosystem

Nvidia is reshaping the AI industry landscape at an unprecedented speed and scale. In the past few months, the chip giant has completed over $140 billion in transactions and holds nearly $100 billion in equity investments. Its strategic intent has far surpassed simply selling chips—Jensen Huang is seeking to use capital to build an ecosystem composed of thousands of AI models, aiming to mitigate the risk of relying on a small number of key clients.

On October 7, according to tech media The Information, Nvidia's next phase of investment will focus on humanoid robots, autonomous driving technologies, and AI models that can run on local devices, as confirmed by multiple bankers, lawyers, and investors working with Nvidia. Among these, Nvidia is in discussions to invest an additional $1 billion in humanoid robot company Figure, which had a pre-financing valuation of about $38 billion this round.

Behind this investment offensive is Jensen Huang's lingering concern over Nvidia's market position. Nvidia disclosed that, in the six months ending July this year, three customers accounted for 44% of its total sales. This overly concentrated client structure is the core driver behind this hundred-billion capital layout. Meanwhile, Nvidia's credit default swap spreads widened in August, with some investors expressing concern about its financial risk exposure.

Acquisition of Hugging Face: A Dual Game of Speed and Trust

Nvidia’s acquisition of Hugging Face is one of the most representative cases in this deal frenzy.

Hugging Face is a decade-old open-source AI model platform that has long attracted acquisition interest. In early summer this year, after OpenAI breached Hugging Face’s system, it immediately began investment negotiations, proposing a $100 million injection; competitors like Salesforce also expressed acquisition intentions.

The key turning point came after Hugging Face co-founder Clem Delangue proactively reached out to Jensen Huang. According to reports from insiders, Huang advanced the negotiations rapidly, assuring Delangue that Nvidia was the only partner who could guarantee Hugging Face's open-source model community would continue operating. Eventually, Nvidia completed the acquisition for $12.9 billion, more than 80 times Hugging Face’s annualized revenue of about $150 million.

Delangue stated at the press conference:

"Throughout the development of Hugging Face, we've received many investment and acquisition offers, which we previously declined, but this past summer, the timing was right."

Ecosystem Logic: Counteracting Client Concentration Risk and Betting on the 'Age of Thousands of Models'

Nvidia's investment logic is rooted in a clear awareness of its own business model’s vulnerability.

Nvidia holds about $99 billion in cash and marketable securities, and enjoys strong cash flow. Jensen Huang hopes to leverage this financial strength to diversify the AI ecosystem—enabling thousands of models to serve thousands of application scenarios, rather than being dominated by just a few giants.

According to reports, an AI infrastructure investor familiar with Nvidia's deal operations stated, "If I were in Jensen Huang's strategic room, I would spare no effort to tilt the scale towards a world with thousands of models serving thousands of purposes."

At the model layer, Nvidia has already invested in Anthropic, Elon Musk's xAI, and open-source model company Reflection AI. Additionally, Nvidia completed a $6 billion deal for software licensing and talent acquisition from Poolside, partly to advance development of its self-developed open-source model Nemotron—Poolside employees released the Laguna open-weight AI model this year.

According to The Information, Nvidia is also in negotiations to invest about $2.5 billion in AI lab Thinking Machines Lab, founded by former OpenAI CTO Mira Murati.

Local AI and Edge Computing: The Next Battleground

With the popularization of AI Agents, demand for local computation is surging, and Nvidia is actively laying out its strategy for the terminal market beyond data centers.

This year, AI Agents capable of handling multi-step tasks on local devices have rapidly proliferated, driving up user demand for local computing hardware. Nvidia has launched its DGX Spark series, designed specifically for running AI Agents locally.

In June, engineers from AI search engine company Perplexity demonstrated to Nvidia how to run their AI Agent software on two DGX Spark computers. The results greatly impressed Jensen Huang, launching a summer of intense negotiations between the two parties.

According to reports, insiders revealed that Perplexity co-founder and CEO Aravind Srinivas once proposed that Nvidia acquire Perplexity outright, with subsequent discussions about a technology licensing and talent acquisition deal valued at at least $20 billion.

Ultimately, both parties announced a partnership agreement at the end of August. Perplexity released a new app Portable Computer, optimized for DGX Spark, while Nvidia plans to participate in Perplexity's new round of financing with about $3 billion, valuing the company at $35 billion pre-financing.

Betting on Data Centers: Financial Risks Emerge

Nvidia's scale of credit backing for large data center projects has already raised risk concerns among some investors.

According to reports, in the large data center project developed by SB Energy, a SoftBank subsidiary, on federal land in Ohio, Nvidia initially discussed providing credit support of up to $250 billion with SoftBank to facilitate OpenAI renting the center for model operation and training. Jensen Huang noted on his blog that the data center campus could accommodate roughly $600 billion worth of Nvidia computing power.

However, Nvidia's credit default swap spreads widened in August, and Huang frequently inquired about spread dynamics with colleagues. Eventually, Nvidia scaled phase one credit guarantees back to $105 billion and is advancing the project in stages to buy time for further decision-making. At the same time, Nvidia also committed to investing $3 billion around SB Energy's IPO.

In addition, Nvidia participated in OpenAI's recent financing round with an investment of $30 billion, with the final $10 billion being delivered on October 1.

To spread financing pressure, Jensen Huang convened six Wall Street firms, including Blackstone, Apollo Global Management, and Goldman Sachs, in early August to jointly support $500 billion in hardware financing. Nvidia stated it may provide up to a 25% backstop guarantee for some related deals.

At the Goldman Sachs San Francisco Tech Conference in September, Jensen Huang responded publicly to risk concerns:

"People are starting to realize that anywhere I invest isn't a bad investment target, because I am an informed investor. We aren't taking any risks—I need certainty."

The Operation Mechanism of the Deal Machine: Jensen Huang Is Hands-On

Nvidia’s deal system is coordinated by the corporate development team led by former HPE and Oracle executive Vishal Bhagwati, but Jensen Huang is deeply involved in the negotiation details of key deals.

In high-value acquisitions like Hugging Face, Jensen Huang personally led pricing and negotiation efforts. He also regularly meets with startup founders, investors, and top executives of private equity-backed enterprises to learn about their product usage and potential cooperation opportunities—a non-formal information-gathering approach common to technology leaders like Microsoft's CEO Satya Nadella.

It’s worth noting that Nvidia does not always win the race. In July, Nvidia showed interest in competing for acquisition of AI model marketplace OpenRouter, but missed out due to needing more time for due diligence. Stripe eventually acquired it for $8 billion. Nvidia immediately shifted its attention to other targets and has completed over $140 billion in transactions in the following two months.

As of the end of July this year, Nvidia’s equity investment holdings were close to $100 billion, with another $25 billion in future investment commitments pending. According to reports, several insiders have indicated that more deals are still in the pipeline.