NVIDIA invests 40 billion dollars in AI in 2026: Jensen Huang's plan for domination
🔎 When the pickaxe seller buys the gold mine
Jensen Huang is no longer content to just sell the chips that power AI. In 2026, he is buying the entire ecosystem — from model developers to undersea cables, including fiber optic glass and power plants.
Just four months. That's all the time it took for NVIDIA to commit more than 40 billion dollars in equity investments in AI. An unprecedented pace that redefines the role of a semiconductor manufacturer.
Behind the numbers lies a raw vertical strategy: ensuring that every dollar spent on AI infrastructure eventually comes back to fuel GPU sales. But this circular logic is seriously worrying Wall Street.
The key points
- NVIDIA invested over 40 billion dollars in AI equity between January and April 2026, a pace never seen before for a chipmaker.
- The 30-billion-dollar OpenAI deal alone represents about 13% of NVIDIA's projected revenue for fiscal year 2026.
- Vertical investments cover the entire chain: Corning (glass, up to 3.2 billion), IREN (data centers and energy), and several other public companies.
- Bloomberg and Goldman Sachs warn of the risks of circular revenues and massive conflicts of interest.
- NVIDIA became the most valuable company in the world with a market cap of 5,062 billion dollars in April 2026.
Recommended tools
| Hostinger | Web hosting for AI projects | From 2.99 € (May 2026, check on hostinger.com) | Deploy AI demos and apps quickly |
| CNBC — NVIDIA AI Investor | NVIDIA news tracking | Free | Follow results and announcements |
| Let's Data Science — Analyse 40B | Detailed investment analysis | Free | Understand the breakdown of the 40 billion |
| Investing.com — NVIDIA Earnings | Live financial results | Free | Follow Q1 FY2027 on May 20, 2026 |
30 billion in OpenAI: the riskiest bet in tech history
Yes, 30 billion dollars in a single company. That is the amount of NVIDIA's investment in OpenAI, finalized in late February 2026. To put things in perspective, this single deal represents about 13% of NVIDIA's projected revenue for fiscal year 2026, estimated at 272 billion dollars according to the analysis by Let's Data Science.
The logic is simple: OpenAI is the world's largest consumer of GPUs. By taking a massive stake, NVIDIA ensures that its flagship client never turns to a competitor. It is contractual captivity disguised as a strategic investment.
But the risk is proportional. If OpenAI stumbles — and tensions between the two companies have been the subject of public denials — NVIDIA would lose a considerable portion of its order book and the value of its equity. Jensen Huang has publicly denied any friction, but the very fact that he had to do so speaks volumes.
This investment is part of a broader movement where AI giants are creating enterprise joint ventures to lock down their markets. Anthropic and OpenAI each launch their enterprise JV with 10-billion-dollar envelopes, signaling that market consolidation is accelerating.
Corning, IREN and the others: buying the supply chain from A to Z
The OpenAI investment grabs the headlines, but the rest of the 40 billion is just as strategic. NVIDIA isn't just funding AI models — it's funding the pipes that feed them.
Corning: the glass that powers AI
Up to 3.2 billion dollars in Corning, the fiber optic glass specialist. Without quality fiber optics, there is no interconnection between GPUs in data centers. By investing in the glass supplier, NVIDIA secures priority access to a critical resource whose demand is exploding.
It's a vertical supply chain logic reminiscent of what industrial conglomerates did in the last century, but on the scale of cloud computing.
IREN: energy as a bottleneck
Billions invested in IREN, specialized in data centers and energy. The assessment is relentless: models like OpenAI's GPT-5.5 or Google's Gemini 3 Pro Deep Think require colossal computing infrastructure. Energy has become the number one limiting factor.
By funding IREN, NVIDIA is anticipating the next wall: that of the electrical power available to run clusters of hundreds of thousands of GPUs.
A calculated breakdown
| Company | Estimated amount | Sector | Role in the NVIDIA ecosystem |
|---|---|---|---|
| OpenAI | 30 billion $ | AI models | Main client, guaranteed lock-in |
| Corning | Up to 3.2 billion $ | Glass / Fiber optics | Critical supply chain for interconnection |
| IREN | Billions $ | Data centers / Energy | Hosting infrastructure and power |
| 7 other public companies | Rest of the 40 billion $ | Various AI | Peripheral ecosystem |
Source: Intellectia.ai, OtonTechnology
The circular revenue machine: how money goes around in loops
This is the point that angers Wall Street. The Bloomberg investigation into circular AI deals (May 2026) pointed to a troubling mechanism: NVIDIA invests in an AI company, that same company uses the money to buy NVIDIA GPUs, which inflates NVIDIA's revenues, which then reinvests even more.
The cycle looks like this: NVIDIA lends/invests → OpenAI (or other) buys GPUs → NVIDIA's revenues increase → NVIDIA reinvests. Each dollar can pass through NVIDIA's accounts multiple times, artificially inflating growth.
Goldman Sachs had warned as early as October 2025 that this strategy blurred the line between strategic investment and inflated circular revenue. The warning went unheeded — or was willfully ignored.
VenturesEdge analyzes this "roundtripping" phenomenon between NVIDIA, OpenAI, and Oracle, noting that some analysts see it as pragmatic financing for AI infrastructure, while others perceive growth dependent on an artificial loop.
The nuance is important: AI infrastructure does indeed cost astronomical sums, and someone has to finance it. But the question is whether NVIDIA's revenues reflect organic demand or an accounting mirror effect.
73% growth: the numbers that justify everything (for now)
Despite the warnings, NVIDIA's results are dizzying. Q4 FY2026 posted 68.1 billion dollars in revenue, up 73% year over year. EPS beat expectations by 6.58% to 1.62 dollars according to Investing.com.
These numbers explain why Jensen Huang can afford to bet 40 billion in four months. The machine generates cash at a speed the tech industry had never seen, even at the peak of the smartphone or classic cloud boom.
NVIDIA's market capitalization reached 5,062 billion dollars in April 2026 according to CompaniesMarketCap, making the company the most valuable in the world — ahead of Apple and Microsoft. The Motley Fool notes that the momentum remains huge ahead of the earnings release on May 20, 2026.
But 73% growth is also a problem. This figure creates an expectation that NVIDIA will have to maintain quarter after quarter. However, when the comparison base explodes, maintaining this pace becomes mathematically impossible. This is exactly where the risk of circularity becomes dangerous: if organic growth slows, circular investments become the only way to maintain appearances.
Chinese competition pushes NVIDIA to accelerate
An often underestimated element in this massive investment strategy: the growing pressure from China. Moonshot AI raised 2 billion dollars and its Kimi K2.6 model reaches 88.1 in agentic and 84 in general — scores that place it in the global top 10 in open-weight.
NVIDIA cannot afford to lose market share to a Chinese ecosystem that is progressing rapidly and which, moreover, is forced to develop its own chips due to US export restrictions. By buying the Western AI ecosystem, Jensen Huang is building a wall of dependency that China will not be able to easily cross.
The strategy also aims to lock in clients before they are tempted by alternatives. When OpenAI spends 30 billion dollars of NVIDIA's money on NVIDIA GPUs, there is no longer any possible negotiation on prices or volumes. The client is literally shareholderized by its supplier.
A hedge fund disguised as a chip maker
The phrase is provocative, but it perfectly summarizes what Goldman Sachs had anticipated. NVIDIA no longer looks like a traditional semiconductor manufacturer. It is an investment fund that produces its own underlying assets.
SpazioCrypto summarizes the situation: Jensen Huang is buying the ecosystem, not just chips. The vertical strategy includes the entire supply chain — glass, energy, compute. It is an integrated vision that has no equivalent in the history of tech.
The closest parallel might be Amazon, which built AWS by internalizing the entire chain — from submarine cable to server. Except that Amazon was investing in its own infrastructure. NVIDIA, on the other hand, is investing in its customers' infrastructure.
This distinction is crucial. When you own your infrastructure, the risk is controlled. When you own your customers' infrastructure, you are exposed to their failures, their strategic pivots, and their own competitive pressures. If one day a client like OpenAI decides to diversify its chip suppliers — a distant hypothesis but not impossible — NVIDIA will end up with devalued equity and a lost client.
AIToolInsight highlights this paradox: the OpenAI deal alone could represent 13% of NVIDIA's projected 2026 revenue, creating an unprecedented mutual dependency between a hardware provider and a software developer.
What this means for the AI market
For users and companies deploying AI, this vertical concentration has concrete consequences. When a single player controls the chips, finances the models, owns the data centers, and provides the energy, the price of AI will never be dictated by the free market.
The most performant current models — GPT-5.5 at 98.2 in agentic, Claude Opus 4.7 at 94.3, or Gemini 3 Pro Deep Think at 95.4 — cost a fortune to train. These costs are directly linked to the price of NVIDIA GPUs and the infrastructure surrounding them.
If NVIDIA controls the entire value chain, it implicitly sets the price of every API request, every fine-tuning, every deployment. Companies building on these models are actually building on an infrastructure controlled by a single player.
Zoom Assurance's French-language analysis confirms this reading: NVIDIA intends to consolidate its position as an essential provider by ensuring its technologies power OpenAI's advances. The keyword is "essential." That is exactly the risk for the market.
❌ Common mistakes
Mistake 1: Confusing strategic investment with AI philanthropy
NVIDIA is not investing $40 billion to "advance humanity." Every dollar has an expected return, direct or indirect. Failing to see the lock-in behind the checkbook is an analytical error. The solution: always ask "who is the end customer of this investment?".
Mistake 2: Ignoring the risk of circularity in financial analysis
Many analysts take NVIDIA's revenues at face value without adjusting for potential circular revenues. It's like counting the same euro twice. The solution: follow Goldman Sachs alerts and Bloomberg investigations, and look at organic growth excluding equity investments.
Mistake 3: Thinking the strategy is solely American
NVIDIA's massive investment is also a response to the rise of China in AI. Ignoring geopolitics means understanding only half the picture. The solution: integrate the competition from Moonshot AI, DeepSeek V4 Pro, and other Chinese players into any analysis of NVIDIA's strategy.
Mistake 4: Believing the $40 billion is a one-shot
This is not an exceptional investment. It is a new business model. NVIDIA has become a permanent investor in the AI ecosystem. The solution: anticipate additional investments and not treat the $40 billion as an isolated event.
❓ Frequently asked questions
Why is NVIDIA investing in glass (Corning) and not just in AI?
Because without high-quality optical fiber, GPUs cannot communicate with each other in data centers. Glass is a hidden bottleneck in AI infrastructure. By controlling it, NVIDIA controls the entire chain.
Are circular revenues illegal?
No, but they pose a financial transparency problem. If NVIDIA's money comes back to NVIDIA via purchases from its investments, the apparent growth is inflated. It is a risk for investors, not a crime.
What is the maximum risk for NVIDIA?
A simultaneous crash in the valuation of its AI stakes AND a drop in organic demand for GPUs. In this scenario, NVIDIA would find itself with a devalued stock portfolio and revenues in free fall. The scenario is extreme but not impossible.
How are competitors reacting?
AMD is trying to break through with its MI300 chips, but does not have the cash to replicate NVIDIA's equity investment strategy. Google and Amazon are internalizing their chips (TPU, Trainium), reducing their dependency. The real alternative comes from players who control their own stack.
Do the $40 billion only include equity investments?
Yes, these are specifically equity bets, not supply contracts or loans. NVIDIA is taking equity positions, which makes the strategy even bolder because it exposes the company to the stock market fluctuations of its partners.
✅ Conclusion
NVIDIA in 2026 is no longer a chip company — it is a vertical investment fund that manufactures its own underlying assets. With $40 billion committed in four months, Jensen Huang is building the most sophisticated monopoly in tech history: one where every link in the chain belongs to the same owner. The question is no longer whether the strategy is ambitious, but whether it is sustainable when growth slows. Follow the results of May 20, 2026, on Investing.com to get the first answer.
