During a tour of Google's California headquarters, Googleplex, CEO Sundar Pichai highlights the company's innovative projects, including a walkway that passes by a giant dinosaur skeleton and a beach volleyball pitch, where dozens of employees are enjoying their lunch under the November sun.

Pichai is particularly enthusiastic about showing me a laboratory located at the back of the campus, hidden behind trees, where a groundbreaking invention is being developed.

This laboratory is the site of Google's secret weapon, a project that has the potential to revolutionize the tech industry.

The Tensor Processing Unit, or TPU, is a small chip that may seem unassuming, but according to Pichai, it will eventually power all AI queries that go through Google, making it a crucial component in the world economy.

Pichai emphasizes that AI is a profound technology with the potential for extraordinary benefits, but also acknowledges the need to address potential societal disruptions.

However, there is uncertainty surrounding the AI hype, with concerns that it may be a bubble that could burst, potentially leading to significant consequences.

The Bank of England has warned of a potential "sudden correction" in global financial markets, citing stretched market valuations for tech AI firms, while OpenAI's Sam Altman has suggested that some areas of AI may be experiencing a bubble.

When asked if Google would be immune to a potential bubble burst, Pichai stated that the company could weather the storm but warned that no company, including Google, would be completely immune.

Despite these concerns, Google is investing over $90 billion annually in AI development, a three-fold increase in just four years, demonstrating the company's commitment to this technology.

The AI surge, in which Google is a key player, represents the largest market boom in terms of cash value that the world has seen.

The market value of Google and four other tech giants, all headquartered within a short drive of each other, exceeds $15 trillion.

Chipmaker Nvidia, now a pioneer in AI systems, is worth over $5 trillion, while Apple's value hovers around $4 trillion, and Meta, previously Facebook, is valued at $1.9 trillion; OpenAI, recently valued at $500 billion, is also a significant player.

The financial implications of this trend are substantial, with significant consequences for the economy.

The value of shares in these companies has helped cushion the US economy from trade wars and supported retirement plans and investments, not only in the US but also globally.

However, this trend also carries a significant risk, as the US stock market's growth is heavily dependent on the performance of a handful of tech giants, known as the Magnificent 7, which comprise one-third of the S&P 500's valuation.

According to the IMF, the market value is now more concentrated in a few firms than it was during the dotcom bubble in 1999.

Pichai notes that every decade brings significant inflection points, such as the personal computer, the internet, mobile, and cloud, and now it is clearly the era of Artificial Intelligence.

The question remains whether the AI hype is a bubble that will burst.

Pichai offers two perspectives on this issue, citing the exciting progress of AI services and the potential for irrationality in investment cycles.

He acknowledges that while there is rational progress, there are also moments of collective overshooting in the industry.

Pichai concludes that the situation is a mix of rational and irrational elements.

A distinction is emerging in the markets between businesses that rely on borrowed money and complex deals to access AI-powered chips and the largest tech companies, such as Google, Microsoft, and Amazon, which can fund their own chip development.

Google's custom silicon chips, known as TPUs, are a key component in this landscape.

The lab where TPUs are tested is a large facility with a mesh of multicolored wires and deep blue blinking lights, surrounded by signs indicating restricted access.

The lab is notable for its high noise level, which is due to the cooling systems required to control the temperature of the chips, which can become extremely hot when processing trillions of calculations.

TPUs are designed to power AI machines and operate differently from other types of chips.

The central processing unit (CPU) is the primary component of a computer, performing most processing and control functions, while graphics processing units (GPUs) execute specialized tasks, including AI processing.

Application-specific integrated circuits (ASICs) are custom-built chips designed for specific purposes, such as AI algorithms, and Google's TPU is a type of ASIC.

The AI boom has been driven in part by the rush to amass high-performance chips and deploy them in data centers.

Nvidia's CEO, Jensen Huang, has referred to these data centers as "AI factories," which are equipped with massive energy and cooling systems.

While some data centers are enormous, such as those described by Mark Zuckerberg as being the size of Manhattan, Google's TPU lab is more modest, focused on testing technology for deployment elsewhere.

There are stories of tech executives, such as Elon Musk and Larry Ellison, attempting to secure large quantities of high-performance chips from manufacturers like Nvidia.

Ellison has described a dinner meeting with Nvidia's Jensen Huang, where he and Musk begged Huang to sell them more GPUs, offering to pay any price.

The competition to access high-performance chips and scale them up in massive data centers is driving the AI boom, with a perception that the only way to win is to continue spending heavily.

The Rosewood Sand Hill hotel, a hub for Silicon Valley deal-making, is where major transactions are negotiated, with its proximity to Stanford University and Meta's HQ, as well as major venture capital firms.

There are rumors about which companies will next announce customized AI chips, or ASICs, to compete with Google and Nvidia.

Before my visit, there was speculation about OpenAI's investment plans, which Elon Musk co-founded, and its efforts to establish a commercial structure.

OpenAI has experienced phenomenal user growth, particularly with its chatbot, ChatGPT, but there are questions about its spending commitments and revenue.

Some have speculated that OpenAI may need government support to design its own custom AI chips, despite its ambitions to do so.

In a recent podcast, an OpenAI investor questioned the company's spending commitments, prompting co-founder Sam Altman to respond, challenging the revenue figures and offering to find a buyer for the investor's shares.

In a recent X post, he outlined OpenAI's plans, including commitments of approximately $1.4 trillion over the next eight years, and made the case for investing in scaling up their technology.

When asked about government involvement, he stated that he does not believe governments should provide insurance for AI companies.

However, he suggested that governments could play a role in developing and owning their own AI infrastructure, which could be beneficial.

Meanwhile, the share price of Coreweave, a start-up that supplies OpenAI, dropped by 26% earlier this month, reflecting the recent volatility in AI infrastructure companies.

There have also been market reactions to perceived credit risks among other firms, although most tech share prices have generally increased over the course of 2025, with a slight decline in recent days.

The excitement surrounding AI's potential remains unabated, with Google's recent launch of Gemini 3.0, a consumer AI model, poised to compete with OpenAI's ChatGPT for market share.

A question remains as to whether this investment will lead to more reliable information, or if it will exacerbate issues like chatbots providing inaccurate recommendations.

In response to this concern, he noted that relying solely on standalone AI systems could lead to unreliable information, emphasizing the need for a more comprehensive information ecosystem.

When pressed on the importance of truth, he concisely replied, "truth matters."

Another significant challenge facing the tech industry is the question of how to power AI technologies, which is not dampening enthusiasm for their development.

By 2030, global data centers are projected to consume as much electricity as India did in 2023, according to the IMF, highlighting the tension between energy demands and climate change targets.

This issue was raised with Google's leader, who was asked if it is feasible for a government to aim for 95% low-carbon electricity by 2030 while also aspiring to be an AI leader.

He replied that it is possible, but emphasized the need for governments to scale up infrastructure, including energy, to support their ambitions.

He cautioned against constraining economic growth due to energy limitations, warning of potential consequences.

Drawing from experience covering the 2000 dotcom bubble, it is clear that even in the face of significant challenges, catastrophe is not inevitable for all companies.

Having interviewed key figures like Steve Jobs and Larry Page, and witnessed the collapse of companies like WorldOfFruit.com, it is evident that some companies can weather the storm and thrive.

A key lesson from this period is that even in the worst-case scenarios, not all companies will fail, and some will emerge stronger.

Amazon, for example, saw its share price plummet to $6 and its market capitalization fall to $4bn during the crash, yet today the company is worth $2.4 trillion.

Similarly, companies affected by a potential AI bubble burst may also recover and continue to grow.

One factor driving the push for AI development, despite potential risks, is the prospect of achieving artificial general intelligence (AGI), a goal that many believe is within reach.

The pursuit of AGI, and potentially artificial super-intelligence (ASI), is a key driver of innovation, with many companies and governments investing heavily in this area.

A Silicon Valley figure noted that the question of whether there is an AI bubble is less important than the broader context of a global battle for AI supremacy, with the US and China at the forefront.

This competition is characterized by a mix of government funding and private investment, with the US relying on a free market approach, while China takes a more centralized approach.

Currently, the US has an advantage in terms of silicon technology, with companies like Nvidia and Google leading the way, but this landscape is likely to evolve.

As the pursuit of AI supremacy continues, some companies will fail, while others will succeed, leaving behind a significant computing infrastructure that will shape the economy and potentially transform the way we work and learn.

The outcome of this competition will have far-reaching consequences, influencing not only the tech industry but also the global economy and the balance of power in the 21st century.