
The prospect of Jensen Huang returning to Nvidia was given a 0% probability by Rohan Goswami when host Ed Elson raised the question during the prediction segment of Prof G Markets. Goswami based that judgement on two sets of verifiable facts. At an industry event he had recently attended, which brought together around 400 AI and technology company CEOs, a possible Huang comeback had been a prominent discussion point. His earlier reporting also showed that Nvidia’s board had spent hours deliberating before agreeing to the leadership change. Several directors had known Huang professionally for more than 30 years, with ties dating to the company’s founding. In Goswami’s assessment, a board that had taken matters this far was unlikely to change direction merely in response to external pressure.

A contrasting reading came from Deirdre Bosa, another guest on the programme. Because Huang had stepped down voluntarily, rather than being dismissed directly by the board, she argued that a return might still be possible. Some observers also saw a potential parallel with Sam Altman, who returned only days after his removal as OpenAI’s CEO. Even so, Bosa acknowledged that public backing did not necessarily imply a reversal by the board. That support included Elon Musk signing a petition seeking Huang’s reinstatement and Donald Trump suggesting that his return “might not be a bad idea”.
A new Huang venture backed by $50 billion was a personal prediction Goswami made on the programme, not an announced plan from Huang. Nor was the amount chosen at random. It equalled the size of Nvidia’s $50 billion AI financing platform, personally introduced by Huang in August 2026.
Has Nvidia Closed the Door on Huang’s Return?
On a conference call, Nvidia Chairman Harvey Jones made the board’s position explicit: it did not intend to bring Huang back. A long-term CEO was being selected, he said, with executive search firm Heidrick & Struggles appointed to manage the process. Rather than a non-committal declaration that “all options remain open.”, this represented a firm stance supported by practical steps. The external search firm was in place and the selection exercise had begun.
The comparison with Altman cited by Bosa therefore has an important limitation. At OpenAI, the board reversed its earlier decision, enabling Altman to return within days of his dismissal. Huang, by contrast, left voluntarily; Nvidia’s board then initiated a formal search and publicly excluded his return. A chairman telling analysts during an earnings call that there is “no intention of bringing Huang back,” is doing more than responding to speculation. The message also gives customers, employees and executive recruiters assurance that the company is seeking a new CEO, not waiting for its founder to come back.
Those formal measures lent support to Goswami’s judgement that, after proceeding this far, the board was unlikely to yield to outside pressure. Neither Musk’s petition nor Trump’s supportive remarks alter the fact that the executive search has already started.
What Could Huang Do Next?
Behind Goswami’s association of Huang’s next step with $50 billion lies a question about the future of Nvidia’s $50 billion AI financing platform after its founder’s exit. Huang’s own wealth, estimated at a net worth of $145 billion, is less central to that assessment.
When Huang personally introduced the platform in August 2026, the aim was to bring together capital from several sources and supply $50 billion in financing for AI computing transactions. Money alone was not its principal asset. More important were Huang’s credibility and connections as the lead dealmaker. Participants’ readiness to finance data centres and chip assets through the platform relied substantially on their confidence in his judgement.
Uncertainty over the platform’s prospects without Huang has already drawn attention from market institutions. According to Reuters, the departure of its chief dealmaker could leave Nvidia facing greater difficulty in financing some of its riskier infrastructure investments. UBS analyst Tim Arcuri quantified the potential impact more directly: a tightening of Nvidia’s financing support could remove $50 billion or more in funding from the market.
Seen in that context, Goswami’s $50 billion prediction matches the scale of the possible funding shortfall at Nvidia’s financing platform. Were Huang to set up a new business with $50 billion, the capital might come through a network shifting away from the existing platform, rather than being raised wholly from scratch. A decision would then confront both the AI infrastructure projects previously financed through Nvidia’s platform and the investors drawn into its ecosystem by faith in Huang’s judgement: move with him to a new entity, or stay with Nvidia in a financing structure that no longer includes him.
How Far Could $50 Billion Take a New Venture?
The significance of $50 billion in initial funding for a new Huang company would depend on how it compares with the prevailing financing environment for AI chip start-ups.
Consider British AI inference chip company Fractile: in November, it closed a $720 million Series C round at a valuation of $7 billion. Following its most recent funding round, South Korea’s FuriosaAI was valued at $2.9 billion. Both rank among the closely tracked businesses in the current AI chip start-up sector, where individual rounds run from hundreds of millions to billions of dollars. Starting with $50 billion would put a venture many times above those funding levels.
At that scale, the business envisaged in Goswami’s prediction could reshape the competitive landscape from launch, rather than resemble a conventional AI chip start-up. With a $50 billion capital base, it could invest throughout the value chain, covering chip design, software development, foundry capacity and the creation of a customer ecosystem. It could do so without successive early funding rounds that dilute control.
A competitive barrier, however, cannot be built with capital alone. By stacking DRAM and logic chips in 3D, Fractile’s roadmap aims to offer more bandwidth and speed than conventional HBM, illustrating the contest between distinct technological approaches. Outspending rivals on an identical approach is not enough to ensure success. Any new company established by Huang would face a market in which Nvidia commands more than an 80% share, even as customers and investors actively look for alternatives. Microsoft has publicly called for a wider choice of AI chips, while emerging competitors including Fractile and FuriosaAI are still securing funding.
Against that backdrop, Goswami’s mention of “friends who have supported him all along” takes on particular significance. Rebuilding in a new entity would require Huang to recreate the network developed over three decades: investors prepared to provide backing, customers prepared to test new chips, and supply chain partners prepared to assign production capacity to a new market entrant.