Is Alphabet a Nvidia/OpenAI killer?

In just a few months, the market has completely changed its mind about Alphabet. In other words, the market got it completely wrong. When a new technology like AI arrives, all sorts of theories of change emerge. That's normal. Some are far-fetched, others highly sophisticated. The most thoroughly argued are not necessarily the ones that ultimately materialize. In the tech world, the cogs don't always turn the way we imagine they will.

At first glance, OpenAI and its peers, with their ambitions to encroach on the turf of the big web players, seem to hold all the cards needed to break some of the existing oligopolies. Although, reality is more complex and the balance of power that appeared after ChatGPT burst onto the global market three years ago is no longer so clear-cut.

The USD 4 trillion mark is getting closer

Initially put under pressure, Alphabet and its Google unit have since restored their reputation by rolling out their plans more discreetly than other tech giants. The group has shown that it is not just a giant of search and the has-been internet. It has become a central player in the global AI infrastructure, to the point that some investors are now wondering whether Google is about to directly threaten two of today's pillars, Nvidia and OpenAI. This new status should also allow the Mountain View group to become the third company to reach a market capitalization of $4 trillion, leapfrogging Microsoft in the process, even though the latter is seen as having a head start in AI thanks to its early partnership with OpenAI.

The best evidence of Alphabet's position is that the new AI stars, particularly OpenAI and Anthropic, are trying to move onto its playing field. "While the media breathlessly reports every step Sam Altman takes to turn OpenAI into a vertically integrated AI giant, Google is already a vertically integrated AI giant," Martin Peers of leading tech publication The Information noted a few days ago. I would add: a vertically integrated AND well-funded AI giant, with obvious synergies already in place between its businesses and an outstanding R&D hub (the Transformer architecture, created at Google in 2017, underpins LLMs. In fact, the T in ChatGPT stands for Transformer).

Gemini 3, Google Cloud and TPU Ironwood

As if to vindicate the market's dramatic about-face on its prospects, Alphabet is currently pushing several pieces forward at once.

The group has just unveiled Gemini 3, the most advanced version of its AI model. Its initial reception last week was excellent. The model is presented as more accurate, more nuanced and less dependent on detailed prompts. Naturally, it is integrated across the entire portfolio of Google tools and solutions, which has some 650 million users worldwide.

Alphabet also moved early onto Nvidia's turf. While Jensen Huang's company plays the GPU ogre thanks to its technological lead, the group has teamed up with Broadcom to develop Ironwood TPUs, AI-dedicated chips that will power the cloud capabilities Google plans to offer its clients from early 2026. Meta, a major consumer of Nvidia GPUs, is reportedly ready to sign a massive deal to secure an alternative to its main supplier as soon as 2027. Google Cloud has even been added to the list of capacity providers by OpenAI and Anthropic. As Nvidia's supply remains constrained by production capacity and soaring costs, the narrative is shifting: Google is no longer just a follower, it is becoming a credible alternative for the AI ecosystem.

OpenAI, too big to fail?

That leaves the answer to the slightly aggressive question in the headline: is this rise in power enough to topple Nvidia and OpenAI? Not so fast. Nvidia is still sitting on USD 500 billion worth of orders for its Blackwell and Rubin architectures, and competition from AMD or Broadcom is only just beginning. OpenAI, despite its colossal funding needs, has secured a network of partners/shareholders/clients that gives it a sort of too big to fail status. Investors also fear a slowdown in capex by hyperscalers and macroeconomic headwinds.

Alphabet is not immune to setbacks either. The fate of oligopolistic leaders is to lose market share. The group still derives 75% of its revenue from advertising, a model under pressure in the face of changing user behavior driven by AI. And the DOJ's sword of Damocles, which could force it to part with Chrome, still hangs overhead. Even if the foray of generative AIs into the world of web browsing has probably rebalanced the arguments.

So, Alphabet the killer? Not necessarily a company killer, but perhaps a killer of naive approaches to AI: the market is not as set in stone as some believed. At 15x (!) 2027 earnings just a few months ago, the stock was clearly undervalued. Now trading at 25x 2027 earnings, it has returned to its sector's norm. But with a level of diversification that is, as we have seen, higher and more coherent.