Powell: AI Investment Is No Bubble

Federal Reserve Chair Jerome Powell has drawn a clear line between the current boom in artificial intelligence and the excesses of the dot-com era, arguing that today’s technology leaders are profitable, capital-intensive businesses driving real growth rather than speculative hype.

Speaking after the Federal Open Market Committee meeting in Washington, Powell said he saw no evidence that AI spending was being inflated by cheap money or loose financial conditions. Instead, he described a broad-based corporate expansion that could lift US productivity over the coming decade.

“I will not name individual companies,” he told reporters, “but they actually have earnings. These are businesses with viable models and profits. It is a very different situation from the late-1990s technology bubble.”

Real profits, not paper dreams

Powell’s remarks mark one of his most direct acknowledgements that AI has become a meaningful driver of the US economy. Tens of billions of dollars are being channelled into data centres, semiconductor fabrication plants, and cloud-computing infrastructure as companies race to meet surging demand for generative AI and automation tools.

Unlike the dot-com boom, where valuations ran far ahead of revenues, the companies leading today’s AI surge, such as Nvidia, Microsoft and Alphabet, are posting record profits. Nvidia’s revenue is expected to exceed $120 billion this fiscal year, while Microsoft and Alphabet have each announced capital-expenditure plans running into hundreds of billions to expand their data and compute capacity.

“These firms are investing from strong balance sheets,” said Powell. “It is not a credit-fuelled expansion. It reflects long-term confidence that these technologies will raise output and efficiency.”

Monetary policy not the driver

The Fed chair dismissed the suggestion that high equity valuations or rapid investment in AI infrastructure were a by-product of interest-rate expectations. “I do not think rates are an important part of the AI or data-centre story,” he said. “The investment decisions are based on long-run views that this is an area of sustained opportunity.”

That view challenges analysts who see parallels with past bubbles inflated by easy liquidity. Instead, Powell presented the AI build-out as structural, driven by fundamental demand for computing power and automation rather than speculative borrowing.

Goldman Sachs shares that view. In a recent note titled The AI Spending Boom Is Not Too Big, chief US economist Joseph Briggs said investment levels were “sustainable” and pointed to potential productivity gains worth up to $19 trillion in cumulative output. AI investment currently represents less than 1 per cent of US GDP, far below the 2–5 per cent share reached in earlier technology cycles such as the Internet and telecom build-outs of the 1990s.

“We are not concerned about the overall scale of AI spending,” Briggs wrote. “If anything, the sector has capacity for further expansion as deployment spreads across industries.”

Productivity promise

The Federal Reserve’s own data shows business investment in equipment and software rising faster than in any period since 2011. The biggest contributors are firms building AI infrastructure: chipmakers, data-centre operators, and cloud-service providers.

Powell said these outlays were already visible in broader output figures. “We are seeing strong growth in capital spending on information processing and electrical equipment,” he said. “That is feeding through to employment, construction, and utilities. It is clearly one of the big sources of growth in the economy.”

Economists at Morgan Stanley and Bank of America have estimated that AI-related investment could add between 0.3 and 0.5 percentage points to annual US GDP growth over the next three years. The Atlanta Fed’s latest regional survey found that 37 per cent of manufacturers expect to adopt generative AI tools within 18 months, citing gains in product design, logistics, and maintenance.

Avoiding the bubble trap

The dot-com crash of 2000 remains a reference point for central bankers wary of asset bubbles. Powell, who joined the Fed’s Board in 2012, has long emphasised the need to separate financial-market exuberance from real-economy fundamentals. In his latest remarks, he argued that AI investment appears to be translating into measurable productivity rather than speculative asset inflation.

“Equity valuations have risen in parts of the technology sector, but overall financing conditions remain balanced,” he said. “Corporate debt levels are not out of line with earnings. We do not see the same leverage or concentration risks that existed 25 years ago.”

The cautionary tone reflects the Fed’s dual challenge of supporting innovation while preventing financial instability. Officials are wary that a surge in investor enthusiasm could push valuations beyond reasonable assumptions about future profits. But so far, Powell indicated, that threshold has not been crossed.

The corporate build-out

The scale of investment underscores the point. Microsoft has committed more than $100 billion to AI and cloud infrastructure over the next five years, Alphabet plans a similar figure, and Amazon Web Services is expanding data-centre capacity across North America and Europe. Nvidia’s dominance in high-end chips has made it the most valuable semiconductor company in history, with market capitalisation exceeding $3 trillion.

Intel, AMD and Taiwan Semiconductor Manufacturing Company are also spending heavily to capture demand for AI-grade processors. The Biden administration’s CHIPS and Science Act, passed before Trump’s return to office, continues to channel public subsidies into domestic semiconductor plants, further anchoring the investment cycle.

Construction data backs up the narrative. Spending on computer, electronic and electrical manufacturing facilities has risen nearly 400 per cent since 2022, according to the US Census Bureau. Employment in data-centre operations has expanded by roughly 25 per cent over the same period.

Risks remain

Not all economists share Powell’s confidence. Some warn that corporate investment cycles can still overshoot, particularly when technological returns are uncertain. The pace of infrastructure build-out, they argue, risks creating short-term overcapacity if demand for generative AI proves slower than expected.

“The productivity benefits are real, but the timing is unclear,” said Diane Swonk, chief economist at KPMG US. “We could see a plateau where spending outpaces adoption before settling into a sustainable trend.”

Others point to concentration risk: much of the capital spending is flowing to a handful of dominant firms, potentially amplifying systemic exposure if one falters. Yet Powell’s comments suggest the Fed views these risks as manageable within the broader health of the corporate sector.

A different kind of boom

Powell’s distinction between the dot-com era and the AI surge is significant. In the late 1990s, much of the market capitalisation was tied to start-ups with little revenue and unsustainable valuations. Today, the AI leaders are profitable giants funding expansion largely from cash flow.

“This is not a speculative mania built on borrowed money,” said Bill Adams, chief economist at Comerica Bank. “It is more like a digital industrial revolution, where heavy upfront spending creates the infrastructure for future productivity.”

That assessment aligns with the Fed’s broader message: innovation can coexist with stability if anchored in real profits and tangible output. For now, Powell seems convinced that the AI revolution belongs firmly in that camp.

“The data support the view that this is an authentic driver of growth,” he said. “We will continue to monitor valuations, but the underlying story is one of genuine productivity gains, not just market euphoria.”