I spent an enormous (but not inordinate) amount of time last month editorializing around what one popular strategist dubbed a grand “reset of the AI hierarchy chess board.”
In two words (or three if you deconstruct the contraction): Google’s coming. For pole position in the AI arms race, and that means trouble for the OpenAI complex, where Sam Altman announced yet another one of his signature deals on Monday, this one even more circular than the rest.
Subject lines from the hodgepodge of daily mailers I receive from The Information give you a sense of the zeitgeist shift. “Google’s vertical AI assault,” one reads. “Google Attacks, Nvidia Reacts,” says another.
For all the plaudits Gemini 3’s garnered, the main takeaway from recent events probably isn’t that Google has it in the bag. Rather, it’s that suddenly, the titans are competing.
“[A] pivotal aspect of [mega-cap tech’s] monopoly status is that they generally have not competed with one another in their primary business lines,” said JonesTrading’s Mike O’Rourke, who’s written brilliantly on this subject. “Now, they are all competing with one another in their most capital-intensive undertaking ever.”
In the same note, O’Rourke pointed out that so far, gains in the Google AI complex have more than made up for losses on the OpenAI side.
The figure above’s a stylized visualization of how this has played out in market cap terms. Broadcom’s a major beneficiary of Google’s TPU coup. The stock’s up 13% since Gemini 3’s launch, and that’s after giving some back in the earliest of this week’s trading.
Headed into Monday, Alphabet and Broadcom had added around $700 billion in market cap between them since Gemini 3’s launch, while Nvidia and Microsoft shed $350 billion or so over the same period.
Commenting further, O’Rourke warned investors against a “winner-take-all” mindset when it comes to AI. “The largest, most well-capitalized organizations in the world are in an arms race with one another — a real competition that should drive multiple compression,” he wrote, adding that “the low-capital, high-multiple platform approach is being replaced.”
If we are headed into an “AI war,” then could these circular-funding schemes become more like “alliances?”
Or firing squads
The success of Google’s TPU chips is really significant. The chips are more efficient for many AI tasks and are more economical (cheaper) than the GPUs from Nvidia.
In time, they (the TPU chips) should challenge the underpinnings of the useful life assumptions underlying most data centers being built with Nvidia GPUs. Especially those funded by borrowed funds.
That is what is rightfully alarming some AI investors more than the success of the Gemini 3 model LLMs versus Chat GPT etc.
They should all just do a big huge merger and become one monolithic Evil Corp. I am sure a feckless US Antitrust would approve the merger.
Agreed, because then I need less apps and passwords.
Fewer
Cyberdyne Systems has a nice ring to it.
H-Man, meanwhile any company that has “AI” in a name, an 8K, a press release is going through the roof. This truly is a bubble with legs and then some.
At what point will the big tech titans, start investing, seriously, in quantum?
Google is one of the leaders currently in the quantum computing field and have set many records
iBM as well
As I follow this evolutionary path, with all the players overdoing their strategies, what my crystal ball is showing is the prospect of massive layoffs as al the excess capacity being created has to unwind. I’m not talking about stock market bubbles; that’s for others to worry over. I’m thinking of the “real” world where business models are crushed from lack of substance and excess capacity. The “dot.com” era was tiny and fragmented by comparison.
What chips are next when the new contest isn’t LLMs. One thing is certain, the rate of change in the IT world is accelerating.