Following its historic large-scale acquisition, this chip giant may next enter the fields of robotics, autonomous driving technology, and on-device local AI models.
In July this year, NVIDIA learned that the AI model marketplace OpenRouter was about to be acquired. This three-year-old startup had become a popular hub for AI models, hosting a wide selection of high-value open-source models. At the time, Databricks and Stripe were already in discussions to acquire the company, leaving NVIDIA late to the table. NVIDIA’s CEO, Jensen Huang, wanted to get involved in the acquisition.
A source said that NVIDIA executives expressed interest in acquiring OpenRouter and were prepared to make a generous offer, but needed more time to evaluate the deal. OpenRouter’s founders were unwilling to wait. Ultimately, NVIDIA did not submit a formal acquisition offer—previously unreported—and Stripe secured OpenRouter with an $8 billion bid.
NVIDIA’s M&A team quickly shifted focus to other targets. In the following two months, the company finalized a variety of transactions totaling over $140 billion, including a $10.5 billion credit guarantee. By the end of July, NVIDIA held nearly $10 billion in equity investments and had an additional $2.5 billion in future investment commitments. More transactions are expected to follow.
Despite investor concerns that NVIDIA and other tech giants are making massive investments in AI projects, leading to mounting financial pressure, industry insiders expect NVIDIA to continue this epic wave of mergers and acquisitions over the coming months. Several investment bankers, lawyers, and investors who have worked with NVIDIA reveal that the company is currently identifying startups to invest in or acquire, particularly those focused on robotics, autonomous driving technologies, and AI models capable of running directly on local devices such as smartphones and home computers.
A previously undisclosed report indicates that, according to informed sources, NVIDIA is discussing an additional $1 billion investment in the humanoid robotics company Figure. Prior to this new funding round, Figure’s valuation was approximately $38 billion. (NVIDIA is already an existing investor in Figure.)
These individuals noted that NVIDIA is also seeking startups to support its self-developed open-source Nemotron large model project and to expand into other AI applications. One reason NVIDIA agreed to spend $6 billion to acquire Poolside’s software licenses and absorb its team was to advance the Nemotron iteration; Poolside’s team launched the open-source Laguna weight model this year.
This series of investment moves is led by a decision-maker who is constantly concerned about NVIDIA’s industry position. Huang Renxun has previously admitted that he has always been anxious about NVIDIA losing its competitive edge—whether due to a cooling in AI demand or the emergence of competing chips capable of replacing NVIDIA’s.
With $9.9 billion in cash and marketable securities, along with a steady stream of cash flow, Jensen Huang aims to leverage NVIDIA’s strong financial position to create a future where thousands of high-quality AI models coexist, rather than a landscape dominated by a few large models. This approach helps mitigate the risk of NVIDIA’s overall performance being heavily influenced by a small number of major clients such as OpenAI. In the six months ending in July, NVIDIA’s three largest customers accounted for 44% of total revenue.
An investor familiar with NVIDIA’s acquisition strategy and investments in AI infrastructure and applications said: “If I were in Jensen Huang’s war room, I would do everything in my power to create a world with thousands of models serving millions of use cases.”
This also explains why NVIDIA's acquisition engine is running at full speed.
Recently, NVIDIA has been negotiating an investment of approximately $2.5 billion in Thinking Machines Lab, an AI laboratory founded by Mira Murati, former CTO of OpenAI. If finalized, this would be another major investment by NVIDIA in a large model company, following prior investments in Anthropic, Elon Musk’s xAI, and several open-source model startups.
NVIDIA has also become a core investor in large-scale data center projects. NVIDIA is negotiating a $3 billion investment in SB Energy, participating in the OpenAI data center project, and providing substantial financing guarantees for the project.
Jensen Huang wrote in his blog that the data center campus developed by SB Energy in collaboration with OpenAI can deploy approximately $600 billion worth of NVIDIA computing power. NVIDIA chose to support OpenAI because “leading AI labs have enormous demands for training and inference computing, but many organizations are expanding faster than their balance sheets and long-term creditworthiness can support.”
Although he missed out on OpenRouter, when his desired targets attract other bidders, Huang Renxun acts decisively in other transactions.
Using the recent acquisition of Hugging Face as an example, investment banks and individuals close to the company say that this decade-old startup has become a mainstream open-source AI model repository and has consistently received acquisition inquiries. Insiders indicate that executives at NVIDIA’s corporate development division have long been interested in investing in Hugging Face.
In early summer this year, after OpenAI’s intelligent agents accessed the Hugging Face platform, OpenAI and Hugging Face initiated preliminary discussions regarding a potential $100 million investment. Around July, existing Hugging Face investors such as Salesforce, along with other competitors, expressed interest in acquiring the company.
Hugging Face co-founder Clem Delangue reached out to Jensen Huang, informing him that multiple potential acquisition offers had been received. According to insiders, Huang swiftly moved to close the deal, assuring Delangue that NVIDIA was the only trustworthy partner capable of sustaining Hugging Face’s open-source model community. Huang offered $12.9 billion for the startup, which generates $150 million in annual revenue—a premium of over 80 times, making the offer impossible to refuse.
Delong said at the press conference: "Throughout its journey, Hugging Face has received numerous investment and acquisition offers, all of which we declined in the past. But this summer, everything came together."
Home computing segment
People involved with NVIDIA's collaboration revealed that over the coming months, NVIDIA will advance more transactions to extend GPUs beyond data centers and into home devices. An increasing number of users will run AI locally on their computers and small devices.
This year, AI intelligent agents have rapidly gained popularity, driving a surge in demand for local AI. These software programs can perform multi-step tasks such as booking flights and organizing email inboxes. A large number of users are purchasing hardware optimized for local AI tasks.
NVIDIA has also launched new products such as the DGX Spark, specifically designed for running AI agents locally. Recent transactions by NVIDIA reflect its intent to expand its product portfolio in this space.
According to insiders, Perplexity, the startup that began with AI search and has since developed the AI intelligent agent Perplexity Computer, demonstrated to NVIDIA engineers in June that its software can run on two DGX Spark devices.
After hearing the demonstration, Huang Renxun became highly interested, and the two companies continued discussions on a potential deal throughout the summer. According to sources, Aravind Srinivas, co-founder and CEO of Perplexity, proposed to Huang Renxun that NVIDIA could acquire Perplexity directly.
The two parties then shifted to discussing a "technology licensing + talent acquisition" proposal: NVIDIA would invest at least $20 billion, or more, to obtain the rights to Perplexity's technology.
Both parties officially announced their partnership at the end of August. Under the agreement, Perplexity will release a customized version of its app to run more efficiently on DGX Spark hardware. The Information first reported this investment: NVIDIA plans to invest in Perplexity at a pre-investment valuation of $35 billion.
Financial risk
Like many tech companies, NVIDIA has a corporate development team responsible for mergers and acquisitions. This department is led by Vishal Bagwati, a former executive at Hewlett Packard Enterprise and Oracle. Nevertheless, Jensen Huang frequently takes a deep hands-on role in transaction details, personally finalizing prices and leading high-stakes negotiations for acquisitions such as Hugging Face.
Jensen Huang also frequently meets with startup founders, investors, and private equity executives to understand how they use NVIDIA’s products and what support NVIDIA can provide. Executives such as Microsoft CEO Satya Nadella adopt similar informal, on-the-ground research approaches.
Over the past few months, Jensen Huang has had to confront a new reality that may constrain NVIDIA’s acquisition ambitions: even a financially powerful company like NVIDIA may hit a safety spending cap when it comes to investments, acquisitions, and guaranteeing large-scale projects.
SB Energy, a subsidiary of SoftBank, is developing a large data center project on federal land in Ohio; this financing has exposed market concerns. Earlier this summer, SoftBank and NVIDIA initially discussed NVIDIA providing up to $250 billion in credit support to OpenAI, which plans to lease the data center to train and run models on NVIDIA chips. (NVIDIA also participated in OpenAI’s latest funding round, investing $3 billion, with the final $1 billion settled on October 1.)
In August, NVIDIA's credit default swap (CDS) spread widened, reflecting investor concerns about the company taking on excessive risk. Jensen Huang also noticed this change in the spread.
NVIDIA ultimately provided only $10.5 billion in credit guarantees for Phase One of the project. While still a massive amount, it is less than half of the original negotiated scale. The phased implementation of the project gives NVIDIA several years to decide whether to support Phase Two.
Meanwhile, NVIDIA is advancing another investment, planning to inject $3 billion into SB Energy before and during its IPO.
Jensen Huang emphasized that sustained progress in AI requires more institutions to jointly fund the development of chips, data centers, and power infrastructure. In early August, Huang brought together six Wall Street institutions—including Blackstone, Apollo, and Goldman Sachs—to raise capital for hardware projects. NVIDIA stated that, in certain related transactions, it may provide a guarantee of up to 25% of the total project financing.
But Huang Renxun downplayed external concerns, noting that well-capitalized investors like NVIDIA have not been overly aggressive.
In September, Jensen Huang told the audience at Goldman Sachs’ annual technology conference in San Francisco: “People are gradually realizing that wherever I invest, it’s never a bad bet, because I have information advantages. I don’t take unnecessary risks. We’re not as smart as people think we are. I pursue trades that are sure to win.”
Editor-in-Chief: Guo Mingyu