Good morning. Shares in the homebuilder Lennar fell 4 per cent yesterday after the company said it was cutting prices and offering incentives to draw in customers struggling to afford a home. Should the employment picture \u2014 now fine<\/a> but no better than that \u2014 worsen, the housing market could go from weakness to real distress. Email us: unhedged@ft.com<\/a>\u00a0\u00a0<\/p>\n Big Tech\u2019s AI business models \u00a0<\/p>\n We\u2019ve looked at charts like this before<\/a>:\u00a0<\/p>\n That\u2019s capital expenditures at the five Big Tech companies that are making the biggest plays on artificial intelligence (Nvidia, as a chipmaker rather than a service company, is big too, but is in a different category). The pattern of increasing spending looks similar at all of them. But the scale of the increase is not similar. Here is the same chart, with each company\u2019s spending scaled to where it was four years ago:<\/p>\n Microsoft has doubled its capital spending in order to compete in AI. Alphabet, Amazon and Meta have tripled theirs. And Oracle has pushed up spending 11-fold.<\/p>\n This difference in spending has had a correspondingly disparate effect on free cash flow at the five companies:\u00a0<\/p>\n At Microsoft and Alphabet, operating cash flow is growing strongly enough to keep free cash flow more or less steady in the face of higher capital spending. At Meta, cash flow has started to wilt (and is expected to wilt more next year). Amazon \u2014 where cash flow is always volatile \u2014 has seen a sharp downturn, too. Finally, Oracle has begun to burn cash outright.<\/p>\n Unhedged\u2019s argument has<\/a> been<\/a> that the market, far from blindly throwing money at AI and Big Tech, is making distinctions that appear to be based on cash generation. This continues to play out recently<\/a>, with Oracle shares getting badly beaten up and Meta\u2019s struggling. Alphabet\u2019s shares have been relatively resilient partly because they were previously oversold on concerns about the future of the search business. One might be surprised that Amazon\u2019s shares have not been hit harder, given the fall in free cash, but it has long been a company that prioritises reinvestment over profit.\u00a0<\/p>\n The market has lost patience with Oracle and Meta. Might the same happen to the other three before long? In a recent piece<\/a>, Jason Thomas of Carlyle argues that it might, saying that the five companies have moved from an asset-light\/software model that deserves a high valuation to something like an industrial model. But they are still valued the same way:<\/p>\n When these companies were \u201casset-light,\u201d paying 7x their accounting [book] value made a lot of sense; you don\u2019t value a money printing machine based on the cost of the paper or printing press! But at current price-to-book ratios, when they acquire $100mn in data centre assets, shareholders are effectively asked to pay $1bn, on average, for the purchase. Does this make sense?<\/p>\n If we assume that the PP & E on balance sheet should be valued at cost and assign a 10x P\/B multiple to the rest, these companies\u2019 market caps would be roughly half of what they are today.<\/p>\n Perhaps this is too restrictive an assumption\u2009.\u2009.\u2009.\u2009But it does provide a framework to think about valuations in an era where previously \u201cvirtual\u201d companies suddenly need to concern themselves with industrial era minutiae like capacity utilisation and depreciation rates<\/p>\n Harvard Business School professor Andy Wu takes something like the opposite view in a recent interview<\/a>, arguing that, rather than fundamentally changing their business models, they have\u00a0<\/p>\n taken a shrewd and conservative strategy for AI. They positioned themselves well to benefit from the rise of AI, but they don\u2019t stand to lose that much if AI grows slower than anticipated.<\/p>\n Microsoft has mostly outsourced AI to a third party, OpenAI\u2009.\u2009.\u2009. Amazon will support anybody\u2019s AI model\u2009.\u2009.\u2009.\u2009Meta spent billions of dollars building an open-source AI model that they hand out for free to the world\u2009.\u2009.\u2009.\u00a0 these companies don\u2019t really think that core AI technology is a meaningful business in and of itself.<\/p>\n Instead, they\u2019re focused on profiting from all the adjacencies to AI. I often use a gold rush analogy: OpenAI, Anthropic and xAI are out there digging for gold.<\/p>\n Nvidia is the consummate shovel seller, designing the chips needed by the gold diggers. And Meta is the consummate jewellery maker: Meta\u2019s social media, advertising, wearables and metaverse businesses stand to benefit\u2009.\u2009.\u2009.\u2009Microsoft does a bit of shovel selling and jewellery making, but the key thing is they\u2019re not stuck digging for gold.<\/p>\n Certainly, it\u2019s plausible that Amazon and Microsoft and Google might make less money on their cloud computing than they ideally would like if AI growth slows or declines, but they would not end up in financial distress.<\/p>\n We might sum up Wu\u2019s view by saying that while Big Tech\u2019s data centre spending is significant, it is supplemental to the companies\u2019 core business models, which remain \u201cvirtual\u201d or asset-light. They are not making an existential bet on AI; they are spending to make sure that their core businesses can coexist with AI, should they need to. This implies that those core businesses should still be valued on high multiples of cash flows.\u00a0<\/p>\n I\u2019m not sure where I land on this. A crucial question is how long the current highly elevated level of spending persists. Three years? Ten? Forever? As the timeline extends, it will become harder to argue that these are still, at core, virtual businesses (again, this point applies less neatly to Amazon, which has always straddled the industrial\/virtual line). And it\u2019s not clear anyone knows the answer to this question, because it depends on how the technology and the market for it evolves. The market is giving the benefit of the doubt to Big Tech, as Thomas points out. But again, it is not doing so indiscriminately, as the weak performance of Meta and Oracle shows.<\/p>\n One good read<\/p>\n Some nice meals<\/a>.<\/p>\n FT Unhedged podcast Can\u2019t get enough of Unhedged? Listen to our new podcast<\/a>, for a\u00a015-minute dive into the latest markets\u00a0news and financial\u00a0headlines, twice a week.\u00a0Catch up on past editions of the newsletter\u00a0here<\/a>.<\/p>\n Recommended newsletters for you<\/p>\n Due Diligence \u2014 Top stories from the world of corporate finance. Sign up here<\/a><\/p>\n![\"Line](\"https:\/\/www.newsbeep.com\/uk\/wp-content\/uploads\/2025\/12\/https:\/\/d6c748xw2pzm8.cloudfront.net\/prod\/795810b0-db95-11f0-8ed2-bb5d1738e5d9-standard.png\")
<\/p>\n![\"Line](\"https:\/\/www.newsbeep.com\/uk\/wp-content\/uploads\/2025\/12\/https:\/\/d6c748xw2pzm8.cloudfront.net\/prod\/62e2ad60-db98-11f0-a103-9935c8c9af92-standard.png\")
<\/p>\n![\"Line](\"https:\/\/www.newsbeep.com\/uk\/wp-content\/uploads\/2025\/12\/https:\/\/d6c748xw2pzm8.cloudfront.net\/prod\/be5eaac0-dba4-11f0-9f19-9ff02113e0b7-standard.png\")
<\/p>\n![\"Line](\"https:\/\/www.newsbeep.com\/uk\/wp-content\/uploads\/2025\/12\/https:\/\/d6c748xw2pzm8.cloudfront.net\/prod\/6f36d7b0-dba4-11f0-af8c-09374350d1a4-standard.png\")
<\/p>\n![\"\"](\"https:\/\/www.newsbeep.com\/uk\/wp-content\/uploads\/2025\/12\/https:\/\/d1e00ek4ebabms.cloudfront.net\/production\/dfee3b6d-9e31-411d-9bdf-ba4b484346d9.jpg\")
<\/p>\n