Who is more vulnerable to Google's challenge, Nvidia or OpenAI?

Author: Ye Huiwen

Source: Hard AI

On Monday, December 1, veteran technology analyst Ben Thompson published an article on his widely followed blog Stratechery, diving deep into the current landscape of competition in the AI domain. He likened the development of the entire AI industry to a classic “hero's journey” epic.

Thompson, known for his incisive business strategy analyses, believes that OpenAI and Nvidia are undeniably the two protagonists in the narrative of AI over the recent years.

One has leapt from a startup to a phenomenal consumer technology hero; the other has transformed from a gaming chip maker into the cornerstone “arsenal” of the AI revolution. However, just as the hero Luke in the movie “Star Wars” must enter the Death Star and face challenges, these two companies now face their own “Empire Strikes Back”—with Google launching a fierce and comprehensive counterattack.

As Google, the former overlord, mounts a strong comeback, which of the two AI sector leaders in recent years—hardware giant Nvidia or model leader OpenAI—is in a more dangerous position?

Thompson’s analysis goes straight to the foundations of the two companies’ business models. He believes that while OpenAI continues to “burn cash” and Nvidia is “printing money,” the fundamental attributes of the moat show that OpenAI’s advantages may be even more robust than Nvidia’s.

However, he also points out incisively that OpenAI itself seems not fully aware of this, and is even actively undermining its greatest advantage, which plants a significant risk for its future.

Google’s Counterattack: A Dual-Front Assault

First, let’s look at how Google is “counterattacking.” Thompson notes Google’s first move was releasing its powerful Gemini 3 model. This model overtook OpenAI’s state-of-the-art models in multiple benchmarks, proving that Google still possesses unfathomable prowess in technological R&D. This directly shakes the foundation of OpenAI as the “best model provider.”

More crucial, though, is Google’s second move, which targets Nvidia’s lifeblood.

In the past, it was commonly believed that Google’s self-developed TPU chips, though superior in performance, were only for internal use. But circumstances have changed: Google has begun to sell TPU chips on the market as alternatives to Nvidia GPUs, and has already reached cooperation intentions with giants like Anthropic and Meta. This is equivalent to a formidable competitor suddenly appearing in Nvidia’s lucrative backyard.

As Thompson observes, this “puts Nvidia squarely in the spotlight, with fresh doubts raised over the sustainability of its long-term growth, especially its sky-high profit margins.”

Nvidia’s Moat: Seemingly Solid, Yet Secretly Cracked

So, is Nvidia’s moat deep enough? Thompson analyzes that Nvidia’s advantages mainly stem from three points: superior performance, greater versatility (GPU is more flexible than TPU), and a powerful developer ecosystem built on the CUDA software platform. Yet, when the performance of Google’s TPU matches—or surpasses—Nvidia, its first advantage is diminished.

The deeper risk lies in the CUDA ecosystem. Thompson uses a clever analogy: how AMD once overturned Intel’s dominance in the data center market. He points out that hyperscale cloud service providers like Google and Microsoft found it worthwhile to invest resources in rewriting underlying software to support both AMD and Intel chips, thereby breaking Intel’s monopoly.

Now, Nvidia faces the same problem. Its customer base is highly concentrated among a few big players, all of whom have the motivation and resources to “tear down the wall of CUDA.” Thompson cites his past article: “The pressure and possibility of escaping CUDA is higher than ever before.” Nvidia’s position may be unshakeable in the near term, but the long-term risk of eroded profit margins has clearly emerged.

OpenAI’s Trump Card: The Choice of 800 Million Users

How does OpenAI’s situation compare? Though it appears more financially fragile, Thompson argues its moat has a fundamental difference. Nvidia’s customers are a handful of large corporations, while ChatGPT’s stronghold is a vast consumer market with over 800 million weekly active users.

Here, Thompson makes a simple yet profound point: “The strength of the moat is proportional to the number of independent users.”

Why is that? He explains: persuading a big company’s CEO to switch tech stacks may only take a few meetings and a flashy presentation, but changing the daily habits of 800 million users is a “street fight,” where every user must be won over individually.

This network effect formed from voluntary consumer choice is the hardest fortress to break. Thompson astutely points out: “Changing the habits of over 800 million weekly ChatGPT users is a battle waged only one-on-one. This is ChatGPT’s real difference from Nvidia in fighting Google.”

Biggest Worry: Holding a Trump Card But Not Knowing How to Play It

Given such a massive user base, why is OpenAI still considered at risk? This leads to Thompson’s central criticism in the article: he believes OpenAI has made a huge business mistake: it still hasn’t launched an advertising model.

He sharply points out that for an aggregator platform with a vast user base, advertising is not only the best monetization method, but also the catalyst that can make the product better.

Ads can attract more free users, bring more usage data and feedback to optimize the model; at the same time, by capturing users’ purchase intentions, it enables deeper understanding and more personalized services.

As another expert Eric Seufert notes, Google launched advertising less than two years after rolling out search, and it’s that ad revenue that funded all subsequent innovation. Thompson bluntly states that OpenAI still relies on subscriptions after three years, which is “a commercial failure,” especially when it has committed to purchasing trillions of dollars in compute. This approach is tantamount to handing the huge free user market over to Google, a master at this game.

The Ultimate Test of Business Models

Lastly, Thompson elevates this competition to the ultimate test of his own “Aggregation Theory.” He has always believed that in the internet world, whoever controls user demand holds real power. The rise of ChatGPT perfectly illustrates this.

The question now is: can an established aggregator (Google) use its overwhelming resources to defeat an emerging challenger (OpenAI), which hasn’t fully leveraged its aggregation advantage (that is, the advertising model)? Thompson admits he is both nervous and excited about this.

In summary, Thompson’s core viewpoint is: Nvidia’s vulnerability stems from its business model—its high profits depend on a few big customers who could easily betray it; OpenAI’s advantage is its huge user base, but this advantage is being eroded by shortsighted business strategies. The final outcome of this clash of giants will not only decide the fate of these companies, but possibly redefine the fundamental rules of technology platform competition: Which matters more—the vastness of resources, or the ultimate control over user demand? This is undoubtedly one of the most important issues to watch in the tech sector in the coming years.

This article comes from WeChat official account “Hard AI”. For more AI frontier news, click here

Risk Notice and DisclaimerThe market has risks; investments must be cautious. This article does not constitute individual investment advice, nor does it consider any user’s specific investment objectives, financial status or needs. Users should consider whether any opinions, viewpoints, or conclusions in this article suit their particular circumstances. Investments made based on this are at users’ own risk.