Jensen Huang, NVIDIA CEO resigned today. While AI infrastructure implications and potential regulatory scrutiny will gain the media attention, technology leaders shouldn’t concern themselves with the personal dealings of Jensen Huang. He may have triggered this crisis but this is about what it means for NVIDIA as the world’s most important supplier of AI chips. Instead, take a hard look at AI concentration risk and your exposure to it.

Today, NVIDIA is deeply embedded in the infrastructure, economics, and direction of the AI boom. Its influence isn’t confined to chips. Its power extends upward through the AI technology stack and into models and the AI computing ecosystem. That means a leadership change can affect broader technology markets, financing relationships, enterprise AI plans, and the wider ecosystem. Our AI Bubble model shows that NVIDIA’s $7.4 billion of capital spending over the past year is about 1% of what the nine companies we track spent. Its exposure runs through its enterprise customers. NVIDIA is party to 15 of the circular-financing loops we track, more than any other company, including a guarantee of up to $105 billion on data center leases for an OpenAI affiliate and a $30 billion commitment to OpenAI itself. The guarantee cost impacts NVIDIA only if those customers can’t pay, and Kress, who oversaw these deals as CFO, keeps that job. Keep a close eye on whether financing relationships, investment commitments, and ecosystem deals change following this leadership transition.

Forrester’s AI Bubble Model says a burst is not imminent as most major players financing the build-out remain financially healthy. But the model shows extraordinary supply pressure and warns that circular financing can inflate demand. If NVIDIA stabilizes quickly, the AI boom will show resilience. If financial or ecosystem stress persists, fragility will grow.

Tech Leaders: Discuss Six Questions In Your Next Leadership Meeting

Tech leaders shouldn’t overreact, but they should treat this as a prompt to reassess concentration risk across their AI infrastructure, suppliers, and investments. Ask these six questions in your next leadership meeting:

  1. Where are we dependent on NVIDIA? Even if you are not exposed to NVIDIA directly, you likely are dependent on them through your existing tech relationships. Look beyond GPUs to CUDA, architecture, models, cloud services, partners, and contractual commitments. identify which dependencies would materially disrupt AI services if NVIDIA capacity, pricing, support, or product roadmaps changed. The goal is to find hidden lock-in, particularly where applications have been optimized so heavily for NVIDIA that switching hardware would require significant reengineering. In fact, well before this development unfolded, Forrester predicted that a quarter of AI infrastructure RFPs will require non-NVIDIA accelerator options. Rather than replacing NVIDIA, we believe that organizations will segment workloads across accelerator types based on model size, latency, data sensitivity, and deployment location.
  2. Where is there a credible alternative? Separate theoretical supplier diversity from options that can actually deliver the required capacity and performance. AMD’s tech stack or custom accelerators from AWS, Google, Microsoft, and others may provide alternatives for particular workloads, but availability alone doesn’t make them substitutes. Test alternatives against performance, capacity, compatibility, and cost. Identify which workloads can move in weeks or months, which require significant reengineering, and which effectively remain locked in.
  3. How exposed are our investments to ecosystem and financing changes? Run your AI infrastructure plans against different financial, supply, and partner scenarios. This includes higher GPU or cloud prices, tighter credit, delayed data centers, capacity constraints, partner failures, and your organization’s changing demand. Pay particular attention to multiyear capacity commitments based on aggressive utilization or ROI assumptions. The question isn’t simply whether a supplier survives, but whether changing economics could leave your organization paying for capacity it no longer needs or unable to obtain capacity it does.
  4. Have we sacrificed confidence in a leader for resilience of an ecosystem? Apply the lesson beyond NVIDIA. “Key-person” risk belongs in technology concentration assessments just as supplier, architecture, geographic, and financial risk do. Ask what changes if an influential founder or CEO suddenly leaves, including relationships with governments, customers, investors, partners, and employees.
  5. How else can we reduce our AI concentration risk? Even before this crisis, tech leaders should be asking themselves, how do we scale AI without allowing its economics to scale uncontrollably? Forrester has always recommended that tech leaders deploy AI workloads across endpoints, edge environments, private infrastructure, and public cloud based on cost, availability performance, latency, security, privacy, sovereignty, and data gravity. This not only optimizes costs but builds resilience and drastically reduces concentration risk.
  6. How sovereign is our AI strategy? It’s not clear yet what IP may have been leaked or if controlled NVIDIA chips were sold because of Jensen’s dealings with an alleged spy. However, tech leaders can be sure it’s going to lead to intense scrutiny and calls for more export controls and sovereignty. Tech leaders must be prepared to test whether export controls, sanctions, data residency mandates, or geopolitical conflict could restrict access, and determine where sovereign or regionally controlled alternatives are necessary to maintain operational resilience.

Jensen Huang’s resignation doesn’t guarantee an AI boom collapse or even the end of NVIDIA’s market leadership. NVIDIA has 38,000 employees who design and engineer some of the worlds’ leading AI infrastructure. It also takes years to design today’s AI chips, let alone the years it takes to then manufacture them and bring them to market. That expertise and capability can’t be scaled overnight. However, there is a real risk over time from reinvigorated competitors, a brain drain, and slowing of innovation. The lesson for technology leaders isn’t to bet against NVIDIA. It’s to stop treating concentration risk as merely a chip sourcing problem.

Schedule a guidance session or inquiry call with us to discuss your broader AI strategy.

Share