TCI's sharp Microsoft cut is not just a hedge fund trade. It is a warning that even the strongest AI platforms still have to prove the spending turns into durable cash flow.
Chris Hohn has spent years owning Microsoft as one of the cleanest ways to back enterprise software, cloud computing and the AI boom. Now his fund has almost walked away from that bet, and the timing matters.
According to a report from the Financial Times, Hohn's TCI Fund Management cut its Microsoft position from about 10% of the portfolio at the end of 2025 to just 1% by the end of March 2026. The stake had been worth roughly $8 billion, and TCI told investors the reason was not routine trimming after a strong run. The concern was that rapid progress in AI could weaken Microsoft's competitive position, especially in Office and, to a lesser extent, Azure.
That is a serious signal because TCI is not known for fast trading around quarterly noise. Hohn runs a concentrated book and typically holds large positions for years. Microsoft had been a major holding for much of the past decade, a period in which the stock rose almost 400%. When a long-term investor sells down one of the market's most accepted AI winners, the question is no longer whether AI is important. It is whether the economics are as attractive as investors assumed.
Microsoft still looks powerful on the surface. In its fiscal third quarter, the company reported revenue of $82.9 billion, up 18%, while Microsoft Cloud revenue reached $54.5 billion. Azure and other cloud services grew 40%, and management said its AI business had passed a $37 billion annual revenue run rate. Those are not weak numbers.
The issue is what Microsoft must spend to keep producing them. Capital expenditures reached $31.9 billion in the quarter, with roughly two-thirds going to short-lived assets such as GPUs and CPUs. Free cash flow fell to $15.8 billion as infrastructure spending absorbed more of the company's operating strength. Microsoft also reported commercial remaining performance obligations of $627 billion, helped by OpenAI commitments, which gives bulls a strong demand argument. Demand is not the same as attractive returns.
This is the hyperscaler problem in plain view. Microsoft, Alphabet, Amazon and Meta are racing to build AI capacity before customers run out of compute. That race supports Nvidia, power providers, data center landlords and networking suppliers, but it can pressure margins at the platforms writing the checks. A software company with extraordinary gross margins becomes a more capital-intensive infrastructure company when it has to keep buying chips, leasing data centers and securing electricity at enormous scale.
For years, Microsoft's Office franchise was the definition of a durable moat. Workers used Word, Excel, PowerPoint, Outlook and Teams because their companies already lived inside that ecosystem. AI changes the interface. If users start completing work inside agent-based systems, chat environments or new productivity layers, the value may shift away from the familiar application bundle. Microsoft is trying to own that transition through Copilot, but TCI's move suggests some investors no longer see that ownership as guaranteed.
Azure brings a different risk. The cloud business is one of Microsoft's great growth engines, and AI demand has made capacity scarce. But scarcity cuts both ways. It can support pricing in the short term while forcing large upfront spending before utilization, pricing and customer retention are fully visible. If Google Cloud, Amazon Web Services, Anthropic, OpenAI or smaller model providers change where workloads sit, Microsoft may still grow quickly while earning less incremental profit than the market once expected.
Investors Are Looking Beyond The Obvious Winners
The interesting part of TCI's repositioning is that it does not read like a rejection of AI. It reads like a change in where the profit pool might settle. The FT reported that TCI increased its Alphabet stake from 3% to 5% of the portfolio, making it the fund's largest technology position. That suggests Hohn is not stepping away from big technology altogether. He is becoming more selective about which dominant platforms have the right balance of valuation, risk and competitive position.
That distinction matters for StartupFortune readers because the AI trade is broadening. The first phase rewarded the obvious names: Microsoft for OpenAI access, Nvidia for GPUs, and the hyperscalers for owning the customer relationships. The next phase may be more demanding. Investors will compare software margins with power constraints, chip supply, data center depreciation, model costs and the risk that AI makes parts of legacy software easier to replace.
This is why capital may keep rotating into less obvious beneficiaries. Semiconductor equipment makers, grid infrastructure companies, cooling specialists, data center operators, energy producers and selected enterprise software firms may offer cleaner exposure to AI buildout than the platforms carrying both the upside and the spending burden. The market is starting to ask who gets paid first, who takes depreciation later, and who has to defend an old franchise while funding a new one.
Microsoft is not suddenly a broken company. It has scale, customers, distribution, cash flow and one of the deepest AI partnerships in the market. But Hohn's sale shows how quickly the narrative can change when the safest winner starts looking more complicated. The next test for Microsoft is not whether it can spend enough to stay in the AI race. It is whether that spending produces the kind of free cash flow investors used to take for granted.
Also read: A smaller Z-Image text encoder lowers the cost of image AI • The fight over vibe coding shows AI software work is growing up • Monk is using rent as a recruiting weapon for AI talent
