The global AI surge has become one of the most powerful economic forces of the decade, reshaping technology, finance and geopolitics at a pace few expected even five years ago. At the centre of this transformation stands Google, now investing more than $90bn a year to secure its place in an era its leadership describes as the most “profound technological shift humanity has worked on”. Yet even inside Silicon Valley, the excitement sits alongside a growing contradiction: an industry racing ahead with extraordinary investment while quietly wondering whether the boom rests on fragile ground.
This tension is clear when Google’s CEO Sundar Pichai walks through the company’s vast California campus and proudly points toward a small chip he believes will anchor the future. The Tensor Processing Unit, or TPU, is Google’s answer to the global scramble for chips powerful enough to train and run advanced AI. Though it appears understated, it holds the potential to process billions of queries in the coming years. For Pichai, this custom hardware is not only a technical breakthrough but also a shield against the volatility of the AI economy.
Even so, Pichai admits something that many tech leaders avoid saying publicly: no company is safe if the AI bubble bursts. Global financial institutions, including the Bank of England, have already warned that valuations across major AI firms look “stretched”. Investors whisper about overheated markets. Even OpenAI’s Sam Altman has described parts of the industry as “bubbly”.
If the boom begins to crack, the consequences would be global. The valuation of the world’s most influential tech firms—Google, Apple, Microsoft, Meta, Nvidia, Amazon and Tesla—now makes up one-third of the entire S&P 500. Such concentration means the fate of global markets lies heavily in the hands of a handful of companies and the chips that drive their AI systems.
Despite these warnings, the AI race continues at full speed. The question is why—and what comes next.
The chip war powering AI’s trillion-dollar moment
At the core of today’s AI boom is a fiercely strategic battle for chips and computing power. These chips—GPUs, TPUs and other application-specific processors—now act like the engines of the modern economy. Companies that control them dominate not only search and cloud platforms but also the direction of AI research, defence innovation and global economic strategy.
Inside Google’s heavily restricted TPU lab, the intensity of this competition becomes clear. The room hums with the sound of cooling systems working overtime as racks of chips run calculations at unimaginable speeds. These processors generate enormous heat, but their real power comes from how well they scale across massive data centres.
Across Silicon Valley, similar “AI factories” are being built. Nvidia, the company that pioneered the GPUs now central to AI development, sits at the heart of this rush. Startups and tech giants alike are pushing to secure as many of its chips as possible. Stories of billionaires competing over supply—often over dinner at exclusive hotels—reflect a market where access to hardware now determines who leads and who follows.
The growth is staggering. Nvidia is worth more than $5 trillion. Apple nears $4 trillion. Meta, Microsoft and Google collectively hold trillions more. Their valuations alone have boosted pension funds, supported the US economy through trade shocks and shaped global investment strategies.
But the spending required is even more dramatic. OpenAI, for instance, is preparing for investment commitments worth more than a trillion dollars over the next eight years. It plans to design its own chips and expand its computing power many times over. These ambitions are so large that some observers question whether governments—not just private firms—may eventually need to build and own parts of national AI infrastructure.
Yet signs of strain are emerging. Coreweave, a major AI infrastructure provider, recently lost more than a quarter of its value in a single slide. Venture capital firms have begun asking sharper questions. And the push to scale up chip production is happening at a moment when global energy infrastructure is already under pressure.
Still, the race continues because those involved believe the reward is bigger than the risk: artificial general intelligence, or AGI.
Can the AI giants survive a bubble—if one exists?
The comparison to the dotcom era is impossible to ignore. In the late 1990s, internet companies soared to extraordinary valuations before collapsing in dramatic fashion. Yet the companies that survived—Amazon, Google and Apple—went on to define the global economy for decades.
Many in Silicon Valley believe the same will be true this time. Even if some AI companies fall sharply, the computing infrastructure left behind will shape future scientific, economic and social systems. For Google, the TPU strategy is central to this long-term vision. Custom chips give the company independence from shortages and from rivals’ hardware innovations, helping it weather future downturns.
But challenges remain. AI’s energy demands are rising so quickly that global electricity use for data centres may match the entire consumption of India by 2030. Governments committed to low-carbon energy face a difficult balance: how to support economic growth driven by AI while also meeting climate goals.
There is also the question of reliability. With Google’s Gemini 3.0 now competing directly with ChatGPT, AI companies must show that these models can deliver trustworthy information. If they fail, public confidence in AI could shift sharply—introducing a different kind of risk to the industry’s future.
For now, though, the push continues. Silicon Valley sees AI not only as a technological revolution but also as a geopolitical contest, especially between the United States and China. China backs its AI development with centralised government investment. The US relies on chaotic but often fruitful competition among private companies. In this ecosystem, trial and error—sometimes expensive, sometimes spectacular—drives progress.
This competition, many argue, is why even a bubble might not slow the long-term trajectory of AI.
The bigger picture: power, dominance and the future of AI
The deeper story behind the AI boom is not only about economics or innovation. It is about global power. The countries and companies that control the biggest data centres, the most advanced chips and the strongest AI systems will influence everything from healthcare and finance to national security and scientific research.
Tech giants already understand this. Every billion spent on chips, data centres and research infrastructure is also a positioning move in a larger global contest. Even if markets wobble, the underlying drive to dominate AI ensures investment will continue.
The risk is that smaller companies, and even some large ones, may collapse under the weight of aggressive expansion. But Google, Amazon, Microsoft and Nvidia likely have the resources to survive turbulence. Their investments will shape the physical and digital foundations of a future where AI permeates daily life—how we work, learn and communicate.
The question that remains is whether the AI boom will stabilise into a long-term transformation or burst dramatically before reaching maturity. For now, Silicon Valley is betting big that even if a correction comes, the world has already crossed into an age where AI is too important, too strategic and too powerful to slow down.
What is clear is that the global AI race is reshaping the 21st century. Whether it does so smoothly—or through a dramatic correction—will define how the next decade unfolds.
