The artificial intelligence trade has faced several headwinds this year, but last week’s sell-off struck at the foundations of the multiyear AI infrastructure boom, revealing a stark shift in investor sentiment as warnings about financing costs and returns become harder to ignore.
The week began with industry leaders highlighting the growing importance of capital availability and the cost of financing AI infrastructure.
Intel and Broadcom executives warned that funding could become a major constraint on the industry’s expansion, reflecting mounting pressure on companies to demonstrate returns on their investments.
“If there’s any challenge in the front, it’s capital and the cost of capital, and then the other part is in the longer run, people want to see the ROI,” said Intel CEO Lip-Bu Tan.
The warnings came as bond yields remained elevated, increasing borrowing costs for companies investing heavily in data centres and computing capacity.
The economics of these investments depend partly on how quickly demand for computing power grows and whether it generates sufficient revenue to justify the spending.
Inflation and competition add to AI concerns
Competition from China also continued to weigh on semiconductor stocks, particularly memory-chip makers, as ChangXin Memory Technologies (CXMT) advances its development of higher-grade memory products.
A hotter-than-expected US consumer price index report added to the pressure by pointing to broader inflationary forces, strengthening concerns that interest rates could remain elevated for longer.
Anthropic’s third-quarter revenue figures provided another reason for investors to reassess expectations.
The AI company reportedly generated $16.8 billion in revenue during the September quarter, an increase of nearly 50% from the previous quarter.
While growth remained substantial, it represented a slowdown from the roughly 140% sequential increase reported between the first and second quarters.
The figures came amid growing questions about whether the revenue generated by AI companies will be sufficient to support the industry’s enormous infrastructure commitments.
Gartner warns of potential infrastructure overspending
A Gartner report added to the concerns, warning that more than $3.5 trillion of AI infrastructure spending could be exposed to impairment by the end of 2030 if providers continue their current pace of investment.
AI providers are projected to spend $7 trillion on infrastructure through 2030, while Gartner estimates they will generate no more than $3.6 trillion in related revenue over the same period.
Neocloud providers, which rent computing capacity to AI developers and other customers, could be particularly vulnerable to the imbalance between spending and revenue.
Gartner predicts that by 2030, cost pressures and a lack of differentiated business models will force more than 80% of neocloud providers to cease independent operations.
The warning follows a report by Bain & Company last month that highlighted the scale of investment required to sustain the AI boom.
By 2031, annual spending on AI infrastructure could reach $1.5 trillion, including new data centres and computing capacity, as well as upgrades to existing GPUs, memory and networking equipment, Bain estimated.
Assuming capital expenditure represents about 25% of industry revenue, sustaining that level of investment would require an AI market approaching $6 trillion annually.
Debt markets signal growing financing pressure
The Gartner report weighed on neocloud companies such as Nebius and CoreWeave, while rising yields on CoreWeave’s outstanding debt reflected investors’ growing demand for compensation for the risks associated with financing the sector.
The yield to maturity on CoreWeave’s 9.625% bond due in 2032 reached 14.8%, while its 9.75% bond due in 2031 yielded around 14.9%.
Financing conditions have also become more challenging across the wider AI ecosystem.
Sources on Wall Street previously told Axios that the AI buildout was less concerning when companies could finance it with cash.
However, around a third of AI capital expenditure is now being funded through debt, the publication reported, while capital markets are showing signs of fatigue.
Global AI-related debt issuance fell to $23 billion in September 2026 from $113 billion in June, the Financial Times reported.
Companies are delaying fundraising plans or paying more to raise capital as investors demand stronger evidence that they can withstand higher interest rates and rising computing costs.
Listings involving Blackstone-backed AirTrunk REIT, Nvidia-backed Firmus and SoftBank-backed SB Energy have been pulled or paused.
“The conversation has shifted from growth at any cost to a much sharper focus on quality,” said Harry Evans, a partner at law firm Freshfields, in a Reuters report.
Evans added that deal activity could remain resilient through next year because of a strong pipeline.
Investors advised to be more selective with the AI trade
The mounting concerns have prompted questions about whether the strongest phase of the AI investment trade has passed.
“As we have highlighted previously, we think the AI thematic is entering a more mature phase, where valuations, funding requirements, competitive threats, regulation, and revenue visibility are coming under much greater scrutiny,” a team of strategists at Barclays led by Emmanuel Cau said, MarketWatch reported.
“It is certainly conceivable that the best of the AI thematic trade may be behind us, and therefore we think investors should be far more selective.”
Barclays continues to recommend balanced sector allocations, no longer viewing AI stocks as a “one-way trade”.
The bank advises investors to retain exposure to major technology companies ahead of further third-quarter earnings reports, while favouring businesses with proven business models, strong balance sheets, visible revenue streams and reasonable valuations.
Meta Platforms and Tesla are scheduled to report results after the market closes on October 21, becoming the first of the Magnificent Seven companies to release earnings this reporting season.
The group also includes Alphabet, Amazon, Apple, Microsoft and Nvidia.
Barclays also sees Europe as a potential source of diversification if the AI trade weakens further, as markets less exposed to the investment theme could provide some protection.
The central question for investors is increasingly whether AI demand can translate into sustainable earnings and cash flow at valuations that justify the capital being deployed.
As financing becomes more expensive and scrutiny of returns intensifies, the market may increasingly reward companies with stronger balance sheets and clearer paths to profitability rather than those simply promising rapid growth.
This post The AI trade is changing: here’s how to play it now may be modified as updates unfold
Please note, this site provides content for entertainment purposes only and does not offer financial advice. Read more here