Nasdaq just had its worst day in 19 months. Has the AI bubble started bursting?

The artificial intelligence boom on Wall Street is facing a new test. After months of record-breaking gains fuelled by expectations of massive investments in chips, computing power and data centres, investors are beginning to question whether the industry can sustain its spending spree.

The Nasdaq Composite plunged 5.52 per cent on Thursday, its biggest single-day decline in 19 months, as fears of tighter AI regulations and a slowdown in data centre construction triggered a broad sell-off in technology stocks. The index also entered correction territory, falling more than 10 per cent from its October 6 record closing high.

The sell-off wiped out around $3 trillion in market value from S&P 500 companies. Nvidia shares fell more than 9 per cent, while the Philadelphia Semiconductor Index tumbled over 12 per cent. The S&P 500 declined 4.36 per cent and the Dow Jones Industrial Average lost 3.44 per cent.

The immediate trigger was the US midterm election result, which strengthened the Democratic Party’s position in Congress and delivered victories to candidates promising tighter scrutiny of data centre expansion. But the market reaction points to a bigger concern. The enormous investments underpinning the AI boom could face political, regulatory and economic constraints just as investors are demanding clearer evidence of returns.

So, has the AI bubble started bursting? The latest market rout does not establish that. But it exposes several vulnerabilities in the investment story that has driven technology stocks higher.

More from Business

Why did AI stocks suddenly come under pressure?

The political backlash against data centres has emerged as an unexpected risk for the technology industry.

AI models require enormous computing capacity. That means companies such as Microsoft, Alphabet, Amazon and Meta have been investing heavily in data centres, specialised chips, electricity infrastructure and cloud computing capacity to support the growing demand for AI services.

However, the rapid expansion of these facilities has triggered opposition in several American communities. Residents have raised concerns about electricity bills, water consumption, environmental impact, land use and the limited number of permanent jobs created by large facilities.

The November 3 midterm elections brought these concerns into sharper focus. According to The Washington Post, voters in at least 38 jurisdictions approved bans or restrictions on data centres. Democratic candidates who campaigned on tighter oversight also won gubernatorial races in important data centre markets.

Texas is particularly significant. The state accounts for around a quarter of planned US data centre capacity, according to Goldman Sachs estimates cited by Reuters. Governor-elect Gina Hinojosa has pledged to halt approvals for new data centre projects until legislation is introduced to protect residents and electricity ratepayers.

This matters because the AI investment cycle depends not only on companies having the money to build infrastructure but also on securing land, electricity, permits and access to power grids.

A delay in construction can affect the entire supply chain, from chipmakers and server manufacturers to power equipment suppliers and data centre operators.

Why is the AI infrastructure boom so important to Wall Street?

The AI rally has increasingly become an infrastructure investment story.

Nvidia and other semiconductor companies have benefited from demand for advanced processors used to train and operate AI models. Cloud computing providers have also expanded their infrastructure to accommodate customers seeking access to AI computing capacity.

The investment extends beyond technology companies. Utilities, power equipment manufacturers, cooling system suppliers and data centre developers have become beneficiaries of the AI spending cycle.

This explains why Thursday’s sell-off spread well beyond chipmakers.

Data centre companies IREN and Applied Digital fell more than 14 per cent each, while CoreWeave declined almost 10 per cent. GE Vernova, which supplies equipment used in power generation and electricity infrastructure, dropped 7.5 per cent.

Meta fell 4 per cent, while Amazon and Alphabet each lost 2.2 per cent.

The market is effectively reassessing how quickly this infrastructure can be built and whether the expected returns will justify the capital being committed.

The bigger question is whether AI investments can generate enough returns

A central risk for the AI industry is the gap between the scale of infrastructure spending and the ability to monetise that investment.

Building data centres and buying advanced chips require substantial upfront capital. Companies need to generate sufficient revenue from AI subscriptions, cloud services, enterprise software and other applications to justify these investments.

If demand grows as rapidly as technology companies expect, the spending could eventually generate substantial returns. But if adoption slows, customers spend less than anticipated or competing companies build more capacity than the market needs, investors could begin questioning the economics of the entire investment cycle.

Is this another dot-com bubble?

The comparison with the dot-com crash of the early 2000s is difficult to avoid.

During the late 1990s, investors poured money into internet companies amid expectations that the technology would fundamentally change business and consumer behaviour. Many companies attracted enormous valuations despite having limited revenues, weak business models or no clear path to profitability.

When expectations eventually collided with financial realities, technology stocks collapsed.

The internet did transform the global economy, but that did not prevent investors from losing substantial amounts of money in the subsequent crash.

The AI boom has some similarities. Investors have assigned high valuations to companies expected to benefit from a technology that could reshape industries. Capital expenditure has expanded rapidly, and the market has rewarded companies for announcing ambitious AI investment plans.

But there are also important differences. Several of today’s biggest AI infrastructure investors are established companies with substantial existing businesses, revenues and cash flows. AI demand is also translating into actual spending on chips, cloud computing and enterprise services.

What should investors watch next?

Three developments could determine whether Thursday’s sell-off becomes a deeper correction or a temporary repricing of AI-related stocks.

First, investors will watch how technology companies respond to growing political opposition to data centres. Stronger permitting requirements, construction moratoriums and rules requiring developers to bear the full cost of additional electricity infrastructure could increase project expenses and delay capacity expansion.

Second, the focus will shift towards capital expenditure and returns. Investors will want to know whether companies can convert their massive AI investments into sustainable revenue growth and profits. Announcements of bigger spending plans may no longer be enough to reassure markets.

Third, demand for advanced chips and computing capacity will remain crucial. If orders from cloud providers and other large customers remain strong, it would suggest that the underlying AI infrastructure market is still expanding. A slowdown in orders or evidence of excess capacity would strengthen concerns about overinvestment.

Interest rates will also matter. Federal Reserve officials indicated on Thursday that further rate increases could be possible. Higher borrowing costs can make expensive infrastructure projects less attractive and reduce the valuation investors are willing to assign to future earnings.

The AI boom is not over, but the rules of the investment game may be changing

Thursday’s market rout is a warning that the AI trade is becoming more complicated.

For much of the past two years, investors have focused on how quickly companies can develop AI models, secure chips and build computing capacity. Now, the questions are shifting towards who will pay for the infrastructure, where it can be built, how quickly it can become operational and whether the returns will justify the investment.

The political backlash against data centres adds a new dimension to those concerns. Even companies with deep pockets may find that access to land, electricity and regulatory approvals cannot be taken for granted.

The Nasdaq’s 5.52 per cent fall does not prove that the AI bubble has burst. The index remains up 4.6 per cent in 2026 despite Thursday’s decline, according to Reuters.

But investors are beginning to distinguish between the long-term promise of artificial intelligence and the financial assumptions supporting the current investment boom.