Tech • AI • Robotics • Game
Tricks and pressure tactics: Jensen Huang’s method
9.3/10TL;DR
Nvidia is using acquisitions, open-source software, financing deals and customer pressure to defend its dominance in AI chips as Google, Amazon, OpenAI, Chinese groups and AMD try to build credible alternatives.
KEY POINTS
A chip giant with unusual margins
Nvidia has become one of the most profitable companies in technology, with claims of roughly $1 billion in daily revenue and a 75% gross margin. Forecasts cited for 2028 put revenue near $627 billion, a scale comparable to the world’s largest retailers and cloud groups. Those margins have made the company a prime target for rivals seeking to weaken its grip on AI infrastructure.
Competition is closing in
The pressure no longer comes only from traditional semiconductor rivals. Google is expanding its TPU business, with reports that Anthropic could buy billions of dollars of that hardware, while Chinese model makers such as DeepSeek are moving toward domestic chips. OpenAI is also presented as developing its own processor, showing how major AI labs increasingly want alternatives to Nvidia GPUs.
Why Nvidia’s lead still matters
The core advantage remains software maturity and long-term deployment. AI labs have spent more than a decade building on Nvidia’s stack, making migration costly and risky. Earlier in-house challengers, such as Amazon’s Trainium, were initially seen as weaker options, leaving many advanced users to choose mainly between Google TPUs and Nvidia GPUs.
The fight over Hugging Face
A reported $12.9 billion move for Hugging Face is portrayed as strategically important because the platform sits at the center of open-source model distribution and testing. The logic is that whoever controls that ecosystem gains a route to real-world adoption for chips and software. That matters because open platforms can become launchpads for competing hardware if large model developers gain access on their own terms.
OpenAI, pressure, and the Google threat
By late 2025, Google was described as nearing a major leap with Gemini 3, raising fears that the strongest frontier model could be trained and served outside Nvidia’s ecosystem. In that context, OpenAI was reportedly pressed to keep pace or risk broader financing consequences, with figures as large as $100 billion circulating around the relationship. For Nvidia, the strategic danger is clear: if Google proves its chips can train top models at scale, the best advertisement in AI shifts away from Nvidia.
Open source as a competitive weapon
Nvidia is not relying only on premium hardware. It has also pushed open models such as Nemotron, encouraging companies to buy Nvidia systems and fine-tune models themselves. The strategy can expand the overall AI market while reducing the pricing power of closed-model labs such as OpenAI and Anthropic, whose inference margins are described as roughly 70% to 85%.
Tolerance for distillation fits Nvidia’s interests
Distillation, in which smaller models learn from larger ones, is controversial because critics see it as a form of intellectual property extraction. Yet a permissive view benefits Nvidia if it creates more AI developers, more models and more demand for compute. Cheaper competing models may hurt some labs, but they can also deepen dependence on the underlying hardware layer.
Turning into a financier
To stop Amazon and Google from dominating AI compute with their own lower-cost chips, Nvidia has backed NeoClouds such as CoreWeave and Nebius. The model is simple: take equity, require purchases of Nvidia GPUs, and in some cases promise to buy unused computing capacity or extend cash support. That gives these cloud upstarts the credibility to borrow heavily and expand faster than they otherwise could.
A broader financial arms race
The same tactics are spreading across the industry. Google has reportedly structured tens of billions of dollars in commitments around Anthropic and TPU demand, while Broadcom is tied to similar arrangements. AMD has also used aggressive incentives, including equity-linked terms, to persuade major customers such as Meta to adopt its chips. The result is an AI market in which chip sales, cloud contracts, debt financing and equity stakes increasingly move together.
CONCLUSION
The contest over AI chips is no longer just a technology race; it is a battle over financing, ecosystems and customer control. Nvidia remains dominant, but its future depends on whether those maneuvers can outpace the growing push for alternative hardware and vertically integrated AI platforms.
Ask a question