
0% was the probability Rohan Goswami assigned to a Jensen Huang return to Nvidia. The question came from host Ed Elson in the prediction segment of Prof G Markets. Goswami cited two sets of verifiable facts to explain his assessment. One came from an industry event he had recently attended, with around 400 AI and technology company CEOs present. Huang’s possible return featured prominently in their discussions. The other came from his earlier reporting: before approving the leadership change, Nvidia’s board had debated it for hours. Some directors’ relationships with Huang dated back more than 30 years to the company’s founding. Having reached that point, Goswami considered the board unlikely to backtrack solely because of external pressure.

A voluntary resignation could leave the door open in a way that a direct dismissal by the board might not, fellow guest Deirdre Bosa argued. Some observers saw another reason to consider a comeback possible: Sam Altman had returned as OpenAI’s CEO just days after his removal. Bosa nevertheless recognised the limits of public support. Elon Musk had signed a petition seeking Huang’s reinstatement, and Donald Trump had said a return “might not be a bad idea”, but neither necessarily meant that directors would revisit their decision.
No plan for a new $50 billion venture had been announced by Huang. That possibility was Goswami’s own prediction during the programme. His choice of figure did, however, have a basis: Huang had personally launched Nvidia’s $50 billion AI financing platform in August 2026.
Nvidia: CEO Recruitment Proceeds, Huang Return Excluded
Selecting a long-term CEO was already the task at hand, Nvidia Chairman Harvey Jones explained on a conference call. Executive search firm Heidrick & Struggles had been appointed to run the process, and Jones expressly ruled out any intention by the board to bring Huang back. An external recruiter’s appointment and an active selection process gave substance to that position. This was not an ambiguous assurance that “all options remain open.”
The sequence matters when assessing Bosa’s comparison with Altman. OpenAI’s board changed its original decision, allowing him to regain his role within days of his dismissal. At Nvidia, Huang’s voluntary departure was followed by a formal search for a successor and an explicit public rejection of a comeback. Telling analysts on an earnings call that there is “no intention of bringing Huang back,” also has a purpose beyond dealing with speculation. Such a statement assures customers, employees and executive recruiters that a replacement CEO is genuinely being sought, rather than the company simply waiting for its founder to return.
These official actions strengthened the case Goswami had made: the board was unlikely to abandon its course under external pressure after taking the process so far. Whatever support Musk’s petition and Trump’s comments offered, recruitment for a new executive had already begun.
$50 Billion: The Financing Question Behind Huang’s Next Move
The fate of Nvidia’s $50 billion AI financing platform provides the main context for Goswami’s prediction about Huang’s next move. Less important to his choice of $50 billion is Huang’s estimated net worth of $145 billion; the central issue is how the platform fares after the founder’s departure.
Pooling money from several sources to finance $50 billion of AI computing deals was the platform’s objective when Huang introduced it personally in August 2026. Yet its most valuable resource was Huang himself: his standing and relationships as chief dealmaker, not just the available capital. Participants’ confidence in his judgement largely determined their willingness to use the platform to finance data centres and chip assets.
The uncertainty created by his exit has already come to the attention of market institutions. Financing some of Nvidia’s riskier infrastructure investments could become more difficult without its chief dealmaker, Reuters reported. UBS analyst Tim Arcuri attached a more precise figure to the risk. A reduction in Nvidia’s financing support, he estimated, could take $50 billion or more of funding out of the market.
A potential shortfall at Nvidia’s financing platform is therefore comparable in size to Goswami’s $50 billion forecast. Launching a business with $50 billion might not require Huang to assemble the funding entirely anew. Instead, an existing capital network could migrate from the platform to his new company. That prospect would leave two groups with a decision: AI infrastructure projects previously financed through Nvidia’s platform, and investors who had joined its ecosystem on the strength of Huang’s judgement. They would have to choose between joining him in a new entity and staying with Nvidia in a financing arrangement from which he was absent.
$50 Billion: A Starting Base Far Beyond AI Chip Peers
Today’s AI chip start-up funding market is the relevant yardstick for a hypothetical Huang company launching with $50 billion of initial capital.
A $720 million Series C round completed in November put British AI inference chip company Fractile’s valuation at $7 billion. South Korea’s FuriosaAI, meanwhile, was valued at $2.9 billion after its latest funding round. These closely followed AI chip start-ups operate in a sector where rounds raise hundreds of millions to billions of dollars. An opening fund of $50 billion would be a multiple of those amounts.
Rather than a standard AI chip start-up, Goswami’s forecast thus envisages a company large enough to change the competitive picture immediately. Capital of $50 billion could fund the whole value chain: chip design, software development, foundry capacity and a customer ecosystem. It could support those investments without the repeated early-stage fundraising that dilutes control.
Money is not sufficient, however, to establish a competitive barrier. Technological choices also shape the contest. Fractile’s roadmap illustrates this through its use of DRAM and logic chips stacked in 3D, intended to deliver greater bandwidth and speed than conventional HBM. Pursuing the same approach with a larger budget is not, by itself, a route to success. For any company Huang might establish, Nvidia’s market share of more than 80% would be part of the competitive landscape. So too would active demand from customers and investors for alternatives. Microsoft has publicly indicated that it wants more AI chip choices, while newcomers including Fractile and FuriosaAI continue to secure financing.
Goswami’s reference to “friends who have supported him all along” is significant for that reason. In a new company, Huang would have to rebuild the network accumulated over three decades. It comprises investors ready to finance him, customers prepared to test new chips and supply chain partners willing to reserve production capacity for a new entrant.