Jensen Huang Leaves Nvidia: What It Means for AI Chip Supply
Jensen Huang's exit from Nvidia raises real questions about chip supply, Taiwan ties, and how OpenAI and Anthropic plan their compute.
What happened with Jensen Huang and Nvidia?
Jensen Huang, who co-founded Nvidia in 1993 and has run it as CEO ever since, is no longer at the helm, departing amid a personal scandal rather than a planned succession. That matters far beyond Nvidia’s stock price. Huang spent three decades turning Nvidia into the dominant supplier of the GPUs that power AI training and inference, and he personally built much of the company’s relationship with Taiwan, home to TSMC, the foundry that manufactures the overwhelming majority of the world’s advanced chips. His exit reopens questions about chip supply stability, the US-Taiwan relationship, and how AI labs like OpenAI and Anthropic plan their compute strategy going forward.
TL;DR
- Jensen Huang’s departure ends a 33-year run at Nvidia and removes the person most closely tied to the company’s relationships with both Wall Street and Taiwan’s semiconductor industry.
- Nvidia’s scale is enormous: the company posted roughly $100 billion in revenue last quarter, with around $90 billion of that tied to data center buildouts for AI.
- OpenAI is almost entirely dependent on Nvidia chips, a concentration that looks riskier now that the company’s leadership and market confidence are in flux.
- Anthropic has spent the past year deliberately diversifying, with large compute commitments split across Nvidia, Google TPUs, and Amazon chips, plus a custom silicon effort of its own.
- Expect more AI labs to pursue vertical integration, building or co-designing their own chips instead of relying on a single external vendor.
- Nvidia’s position as the default chip supplier isn’t guaranteed to hold, especially if reduced market confidence pushes the company to pull back on capital expenditure.
- The era of the single visionary hardware CEO may be fading, as companies look to spread leadership and public reputation across multiple executives rather than one figurehead.
Why does Jensen Huang’s exit matter for AI compute?
Nvidia isn’t just another hardware vendor. It is the primary supplier of GPUs for training and running large language models, and its last quarterly revenue was close to $100 billion, with about $90 billion of that coming from data center infrastructure tied directly to AI buildouts. That scale gave Huang enormous influence, not just over Nvidia’s roadmap but over the broader compute supply chain that every major AI lab depends on.
Huang’s Taiwanese background and decades of relationship-building with TSMC gave Nvidia (and by extension the US) a strategic edge in a supply chain that is geographically concentrated in a region with real geopolitical risk. Roughly 90% of the world’s most advanced semiconductors are manufactured in Taiwan. A CEO who could maintain trust and continuity with that ecosystem was, in effect, a form of soft power. Removing that person, especially under scandal rather than succession planning, injects uncertainty into a relationship that the AI industry has largely taken for granted.
How exposed is OpenAI compared to Anthropic?
This is where the compute allocation story gets concrete. OpenAI’s infrastructure is close to 100% Nvidia. Huang himself said publicly that essentially everything OpenAI runs, runs on Nvidia hardware. That’s a single point of failure: if Nvidia’s capital expenditure slows, if market confidence in the company drops, or if its Taiwan relationships weaken, OpenAI has comparatively little room to maneuver.
Anthropic looks structurally different. Nvidia-powered chips account for only about a third of its compute mix. The rest is split across other providers, including large commitments to Google’s TPUs and Amazon’s custom chips, alongside smaller deals elsewhere. Anthropic has also been building out its own custom silicon team, reportedly using advanced internal versions of its Claude models to help with chip design work behind closed doors. None of this happened by accident. It reflects a deliberate strategy to avoid depending on any one hardware supplier, timed notably as the company moves toward a possible IPO.
Will AI labs start building their own chips?
The direction of travel points that way. Custom silicon isn’t new (Google has used TPUs for years, and Amazon has its own AI chips), but the urgency is increasing. Anthropic’s investment in an internal chip design team is the clearest recent signal that frontier AI labs don’t want to be entirely at the mercy of one vendor’s supply, pricing, or leadership stability.
OpenAI, by contrast, has far more catching up to do if it wants similar optionality. Given its near-total reliance on Nvidia, any disruption, whether from leadership turmoil, a slowdown in Nvidia’s data center capex, or friction in the Taiwan relationship, hits OpenAI harder than it hits a more diversified competitor. Expect pressure on OpenAI and other heavily Nvidia-dependent labs to accelerate their own custom chip efforts or lock in multi-vendor deals the way Anthropic already has.
Is Nvidia’s dominance as a chip supplier at risk?
Not overnight. Demand for AI compute remains enormous, and chip supply chains take years to shift, not quarters. Nvidia doesn’t lose its position just because its founder leaves. But confidence is a real input into capital markets, and Nvidia’s valuation has reportedly already taken a hit, with estimates of roughly $1 trillion wiped off at points following the news. If that reduced sentiment leads to lower capital expenditure from Nvidia itself, that could tighten compute availability industry-wide, which is already the single biggest bottleneck standing between current AI capabilities and the next generation of models.
The practical effect isn’t an immediate collapse of Nvidia’s market share. It’s a gradual erosion of the assumption that Nvidia is the automatic default. Labs that can afford to diversify (financially and technically) will do so faster. Labs that can’t, like a heavily Nvidia-locked OpenAI, face a harder choice between staying the course and scrambling to catch up.
What does this mean for the future of AI leadership?
One underappreciated angle here is reputational risk concentration. Nvidia’s public identity has been deeply fused with Jensen Huang for over three decades. When something damages the individual, it damages the company’s market standing almost immediately, because there was no separation between the two. That’s a structural vulnerability other companies are already moving away from. Anthropic, for instance, hasn’t centered its public presence on a single founder. Dario Amodei regularly appears alongside his sister Daniela Amodei and other executives in major announcements and interviews, spreading reputational exposure across several people rather than one.
The likely trend across the AI industry is less reliance on the “visionary founder-CEO” model and more toward shared leadership structures, where no single person’s personal conduct can single-handedly tank market confidence. It’s a logical response to watching a 33-year tenure unwind that quickly.
Frequently Asked Questions
Why is Jensen Huang’s departure significant for the AI industry?
Because Nvidia supplies the overwhelming majority of GPUs used to train and run AI models, and Huang personally built the company’s relationships with Wall Street and with Taiwan’s semiconductor manufacturing base. His exit introduces uncertainty into both the business and geopolitical sides of the AI compute supply chain.
How dependent is OpenAI on Nvidia chips?
Very dependent. OpenAI’s compute infrastructure runs almost entirely on Nvidia hardware, a concentration Huang himself acknowledged publicly. That leaves OpenAI more exposed to any disruption in Nvidia’s supply, pricing, or leadership stability than competitors with diversified chip strategies.
How is Anthropic different in its chip strategy?
Anthropic has spread its compute across multiple providers, with Nvidia-powered chips making up only around a third of its mix, alongside major commitments to Google TPUs and Amazon’s chips. It has also started building an internal custom silicon team to reduce reliance on any single external vendor.
Does this mean Nvidia will lose its market dominance soon?
Not immediately. Chip supply chains and compute buildouts take years, and demand for AI compute remains very high. The bigger risk is a gradual shift as labs diversify and as reduced market confidence potentially pressures Nvidia’s own capital spending.
What is Taiwan’s role in this story?
Taiwan is home to TSMC, which manufactures roughly 90% of the world’s most advanced semiconductors. Huang’s Taiwanese background and long-standing relationships there have been a quiet but significant factor in maintaining US access to advanced chip manufacturing, making his departure relevant to supply chain and geopolitical stability, not just corporate leadership.