{"text":[[{"start":6.9,"text":"The data centre infrastructure boom fuelled the stock market’s AI euphoria: is it also now about to destroy it?"}],[{"start":13.7,"text":"Unease about huge — and growing — capital spending has turned into a serious overhang for some of the biggest tech shares. That was clear last week, when the stock market chose to focus on how Alphabet had burnt through cash for the first time in its history as a public company, rather than the surprising leap in cloud revenue and profitability that the spending had made possible."}],[{"start":36.5,"text":"This week, Meta has been put through the wringer for vowing to press on with its massive investments. At the same time, it dashed hopes that it would relieve the financial pressure by reselling some of its surplus data centre capacity, as Elon Musk’s SpaceX has already done."}],[{"start":52.1,"text":"Microsoft bucked the trend with the news that it is finally seeing serious revenue acceleration from its AI spending. Its shares, though, are still down about 15 per cent from last year’s peak."}],[{"start":63.95,"text":"The unease over data centre spending comes at the end of a month that has also brought a sharp drop in semiconductor shares. The question now is whether this is just a temporary summer tantrum or a serious change of heart about the AI trade that has underpinned the stock market."}],[{"start":82.1,"text":"One possibility is that Wall Street believes the peak in capex is close, even if spending still looks robust in the short term. Neither the numbers nor the rhetoric coming out of the tech companies give any reason to think this. Rather, the mood was summed up by the studied understatement of Microsoft chief financial officer Amy Hood, who described the present shortage of capacity as a “relatively extreme moment”."}],[{"start":107.5,"text":"Some cyclicality in spending is inevitable, even in the midst of a secular boom. The memory chip shortage will run until at least the end of 2027, to judge from recent reports from Micron and others, but new capacity due in 2028 could change the picture. It seems early to anticipate a turn, though."}],[{"start":127.2,"text":"Another possibility is investors are worried that while the revenue the tech companies have been able to generate from all their new AI capacity is starting to rise fast, it is still modest given the scale of spending."}],[{"start":140.05,"text":"The order books of the big cloud companies provide some encouragement. Microsoft and Google’s combined backlog jumped to $1.2tn at the end of last month, compared with less than $500bn a year ago. But there is a clear concentration of risk here, with a massive dependence on OpenAI and Anthropic, which have yet to prove they will need all that capacity to service their own customers."}],[{"start":164.85000000000002,"text":"In a rapidly evolving market, it is also hard to tell when the AI companies will reach anything like sustained profitability. Anthropic provided a shot to the market with a growth surge earlier this year that led it to predict its first profitable quarter. But that was followed soon after by reports of companies cracking down on excessive use of AI services by their workers, known as tokenmaxxing. More recently, inroads made by Chinese models have raised the spectre of greater pressure on pricing."}],[{"start":194.65000000000003,"text":"A third explanation for the unease may be the sheer scale the capital spending has reached relative to the industry’s overall finances. Meta’s generation of positive free cash flow tumbled in its latest quarter, while Microsoft said it would stay positive for its next fiscal year, which has just started. But this is scant comfort coming from companies which churned out more than $110bn spare cash between them in their last fiscal years. "}],[{"start":221.60000000000002,"text":"This points to a deeper structural shift in the industry’s finances. Alphabet, besides burning cash, has lifted its long-term debt to nearly $100bn, up from $11bn a year ago, while surprising the market recently by raising $85bn in fresh equity. Credit rating agency Moody’s summed it up in a report last week: As the big tech companies turn into “asset-heavy” businesses, their “relationship with capital, risk and credit [are] being redefined.”"}],[{"start":253.10000000000002,"text":"Until now, it has been possible to view Alphabet as a hugely profitable search engine company that chose to invest its spare cash in AI. It is quickly turning into an AI company with a very different financial structure — and, possibly, business model — to what came before."}],[{"start":268.90000000000003,"text":"None of these explanations, on their own, provide a particularly persuasive case for why this would be the moment for investors to get cold feet about the data centre boom. Taken together, though, they explain why the anxiety level is rising."}],[{"start":289.40000000000003,"text":""}]],"url":"https://audio.ftcn.net.cn/album/a_1785510383_3868.mp3"}