Nvidia without Jensen Huang: what it means for your SME
The departure of Nvidia's CEO raises questions about talent, the AI supply chain and computing prices. We analyse the real impact on SMEs and IT teams.
5 min read

In this article
An earthquake at the top of AI
Jensen Huang's departure as CEO of Nvidia is not just another corporate news story. We are talking about the executive who for 33 years has steered the company whose chips train and run a good part of the artificial intelligence models used by OpenAI, Microsoft, Google, Meta and Amazon. The temporary replacement by Colette Kress, until now chief financial officer, leaves the company in an unprecedented situation: good products, but a business that is harder to sustain without the figure who articulated it.
The trigger, according to what has been published, is several romantic dalliances by the executive, one of them with a possible Chinese spy, which adds a layer of concern about the national security of the United States. For a Spanish SME this may sound distant, but it is worth not losing sight of the thread: when the main supplier of AI infrastructure falters, the echo reaches the entire chain.
What analysts say (and what they don't)
Independent analyst Fernando Maldonado points to two factors that will determine the future: that companies maintain the pace of AI spending towards 2027 and that the investigations affect only Huang or the entire company. Gartner supports the first part: most global investment in AI will be concentrated in infrastructure, accounting for up to 59% of global spending in 2027. In other words, demand for data centres, servers, GPUs and networks will continue to be enormous.
Maldonado notes that Nvidia's advantage lies in its engineers, its upcoming products and CUDA, the software used by millions of programmers and which makes it difficult to change suppliers. But the most immediate risk is the talent drain: many employees are paid in shares that are worth more than a third less than in October. If the best start to leave, the consequences will take time to show.
«Nvidia sells the chip, helps pay for it and promises to buy the capacity that no one rents», Maldonado describes about the company's cross-investment model.
The business that feeds itself
Nvidia has accumulated some 99 billion dollars in holdings and, according to Morgan Stanley, another 170 billion in commitments and guarantees. A relevant part goes to companies such as CoreWeave, Nebius or Nscale, which buy its chips to market computing capacity. To CoreWeave, for example, it guarantees up to 6.3 billion if it does not find other customers. It is a circle: Nvidia invests in those who buy its chips, those purchases raise sales and share price, and that allows it to continue investing. If the process reverses, it will have less margin to support its customers and they will buy less.
The most exposed are the neoclouds, companies that rent computing capacity and depend on debt to grow. If banks doubt Nvidia's backing, they will lend to them at higher cost. In October, insuring CoreWeave's debt against default already cost more than 11% annually. Large customers, on the other hand, have more options: Google, Amazon, Meta and Microsoft are developing their own chips; OpenAI buys from Google and Anthropic has committed 161 billion to Broadcom. Huang's departure may accelerate those decisions.
Political pressure and dependence on China
Huang had personal access to the president of the United States and obtained permission to sell the H200 to China in exchange for the Government receiving 25% of each sale. However, according to the Commerce Secretary, no H200 had yet reached China and Trump left Huang out of his trip to Beijing in May. With the Democrats controlling Congress since January, Nvidia will have to defend itself with less support and more export controls, which would mean fewer sales in China and more space for Huawei.
What this means for an SME or an IT team
If your company does not buy GPUs directly, you probably will not notice the blow tomorrow. But you will notice it in the medium term through three channels:
- Price and availability of cloud computing. Providers that rent capacity depend on debt and Nvidia's backing. An increase in financing costs may be passed on to GPU instance rates.
- Continuity of AI providers. If you work with services that run on Nvidia infrastructure, it is advisable to review service level clauses and contingency plans. Supplier diversification stops being a luxury.
- Talent and roadmap. The drain of engineers at Nvidia may delay products. If your technology roadmap depends on a specific version of CUDA or a GPU model, ask your provider about timelines.
Our recommendation from ForgeNEX is simple: do not panic, but do review your exposure. Make an inventory of which services in your SME depend on GPU infrastructure, check whether you have alternatives and document a plan B. In a context of political pressure and possible export controls, technological sovereignty stops being a speech and becomes an operational decision. As we already saw with quantum in Spain, depending on a single supplier or a single technology is a risk that is paid for dearly.
What to watch in the coming weeks
- The appointment of a permanent CEO and whether they maintain or expand the cross-investment model.
- The evolution of Nvidia's share price and its effect on index funds: the Swiss central bank had 13 billion invested in June.
- Movements in Congress on export controls and the space Huawei is gaining.
- The reaction of AMD and Broadcom, which were already rising after the news became known.
The element that can contain the blow, according to Maldonado, is that there continues to be more demand for computing than available chips. As long as that happens, Nvidia retains margin. When it reverses, the picture changes for everyone.
Source: Computerworld. Analysis and adaptation: ForgeNEX.
Keep reading
- TypoPrompting: 186 fake AI domains that hijack your browser13 November 2026
- AI agents that break security rules: lessons for SMEs12 November 2026
- Stolen AI keys at Railway: $740,000 in usage charges11 November 2026
- AntiFox: when pentesting AI turns against the cloud11 November 2026
- Sage puts partners at the center of its shift toward AI10 November 2026