Hedge funds start to trim crowded AI bets, UBS says

The top five US hyperscalers are projecting a combined capital expenditure of $660 billion to $725 billion for 2026, nearly double their 2025 outlays, as the AI build-out shifts from software to physical infrastructure. Microsoft is guiding for roughly $175 billion in adjusted capital expenditure for both FY2026 and FY2027, with two-thirds of quarterly spend going to short-lived assets like CPUs and GPUs and the rest to long-lived data center infrastructure, and it added 1 gigawatt of capacity in Q3 FY2026, doubling its global footprint in two years. Amazon AWS has raised its 2026 capex guidance to approximately $220 billion, with CEO Andy Jassy saying AI capacity is expected to remain constrained through 2027 and contracted demand extending into 2028. Meta saw profit drop 14% in Q2 2026 despite a 28% revenue increase as its build-out, including a 1 gigawatt data center in Ohio and a Louisiana facility that could scale to 5 gigawatts, compressed margins, while Alphabet raised its 2026 capex guidance to as much as $205 billion and its Google Cloud backlog more than doubled year-over-year to $240 billion. The Stargate joint venture involving Oracle, OpenAI and others targets up to $500 billion in infrastructure investment by 2029, and Oracle's FY2026 capex reached $55.7 billion, more than doubling from the previous year.

Tigress Financial reiterated its Buy rating on Intel and raised its price target to $145 from $118 on September 15, citing the company's Terafab partnership with companies associated with Elon Musk, including SpaceX, Tesla, and xAI, as a strategic boost to its AI foundry ambitions. The initiative is part of an ambitious semiconductor manufacturing project focused on producing advanced chips for artificial intelligence, robotics, autonomous vehicles, and other compute-intensive applications. Intel said in its second-quarter earnings report that it remains well positioned to pursue sustainable growth through advanced packaging and its wafer foundry network, pointing to momentum in Physical AI and robotics with more than 130 customers testing its Core Ultra Series 3 and Core Series 3 processors for edge AI and robotics applications. The partnership's commercial terms, production volumes, and potential revenue contribution have not been fully disclosed, and the principal risk is that it could strengthen Intel's strategic position in foundry manufacturing without generating meaningful financial returns for several years. According to the Insider Monkey database, 138 hedge funds held stakes in Intel at the end of the second quarter, up from 112 funds in the first quarter, with SoftBank Group Corp. and Coatue Management among the notable institutional holders at approximately $12.14 billion and $1.68 billion, respectively, while roughly 152.25 million Intel shares were sold short as of August 31, representing about 3.02% of the public float.

UBS now expects artificial intelligence capital expenditure to reach nearly $1 trillion in 2026 before climbing to around $1.4 trillion in 2027, with surging memory costs accounting for most of the increase. The bank's updated estimates put total AI capex at $998 billion this year, almost double the $506 billion recorded in 2025, and forecast spending of $1.447 trillion next year. Memory is emerging as the main source of that growth, with UBS estimating memory spending will climb from $71 billion in 2025 to $367 billion this year and $923 billion in 2027, while other AI-related costs are estimated at $631 billion in 2026 before declining to $525 billion next year. That means higher memory costs will account for about 60% of the increase in AI capex this year and more than the entire net increase in 2027, and across the two years UBS calculates that roughly 90% of the nearly $1 trillion increase in AI capital expenditure between 2025 and 2027 will come from higher memory spending. Memory represented about 14% of total AI capex in 2025, a share the bank estimates will rise to 37% this year and reach 64% in 2027, and UBS said price-driven increases would add relatively little to real U.S. gross domestic product, instead representing a transfer of income and profits toward memory producers in Asia.