After the tech stock selloff, UBS sees opportunities to buy on dips: The market underestimates the AI supercycle, and 2027 may mark an inflection point for returns.

After the Nasdaq 100 index, known as the "global barometer for AI trades and tech stocks," shed more than US$2 trillion in market value in October alone, Gerry Fowler, Chief European Equity Strategist and Head of Derivatives at UBS Group AG, said in an interview on Monday that after the recent pullback in AI trades, valuations of hyperscale cloud service providers look attractive, because the latest pessimistic pricing reflects an AI supercycle that appears shorter than it may actually last. "When the overall valuation multiples of hyperscalers such as Alphabet, Microsoft Corporation and Amazon.com, Inc. are already below 20x while their revenue is still growing at 10% to 20% a year, that means the market has adopted a rather pessimistic view. So I think the willingness to buy these stocks on dips is quite strong," he said in a media interview on Monday. Fowler said hyperscalers may be approaching the point where their massive AI spending begins to generate returns. He believes 2027 could mark an inflection point, when data center deliveries will drive conversion of backlog orders and revenue will begin to catch up with capital expenditure. He added that capital has not disappeared from the market but has flowed to storage chip makers and other core hardware suppliers of AI computing infrastructure, which are now beginning to return capital to investors. The consensus trillion-dollar-scale capital expenditure (CapEx) forecasts for the large tech giants or cloud computing supergiants through 2030 are mainly for purchases of AI chips, servers and other capital spending needed to build large AI data centers and traditional cloud computing overall; AI financing costs mainly include borrowing interest and financing fees, usually calculated separately from capital expenditure, though a small portion of eligible construction-period interest may have been capitalized and cannot be double-counted in total costs. Alphabet Inc. Class C parent Alphabet reported last week that free cash flow was negative for a second consecutive quarter, intensifying pessimism and anxiety over the global AI trade theme. The Nasdaq 100 index, which is heavily weighted toward AI-themed tech stocks, fell 3.5% on Friday, its biggest single-day drop since June. In Fowler's view, however, the AI trade is beginning to expand beyond the first batch of data center beneficiaries. Many suppliers of cables, power equipment and other infrastructure focused on the construction progress of large AI data centers have sold out capacity through 2027 and even 2028, leaving less room for upward earnings forecast revisions. Because the market is driven by unexpected changes in growth, he believes the next phase of this unprecedented AI bull market trade must find growth drivers in other key areas. As shown in the chart above, Alphabet Inc. Class C parent Alphabet consumed cash for a second consecutive quarterAI computing infrastructure-related spending turned free cash flow negative after remaining positive for the previous two decades. He instead pointed to key global and European industrial companies that help build new plants as worth watching. In the interview he highlighted lithography giant ASML Holding NV ADR (ASML.US) as an example: the company supplies the most critical semiconductor equipmentthe most advanced EUV and even high-NA EUV lithography machinesto storage chip makers such as Samsung Electronics and SK Hynix that are expanding capacity, rather than directly selling AI infrastructure-related hardware products to AI data centers. "So ASML Holding NV ADR suddenly began to receive upward earnings forecast revisions, and now a considerable portion of key industrial companies in Europe and elsewhere are seeing the same," he said. Global tech stocks slump: The market is repricing the waiting period for cash returns Fowler's core bullish judgment at UBS Group AG on the global AI computing theme and broad tech stocks is that the recent market selloff may underestimate the duration of the AI investment cycle. In the view of the chief European equity strategist, the 2027 wave of AI data center deliveries is expected to drive backlog orders into strong revenue, while large-scale capacity expansion by storage chip makers and chip manufacturing leaders such as Taiwan Semiconductor Manufacturing Co., Ltd. Sponsored ADR may in turn transmit earnings growth on a large scale to semiconductor equipment-type industrial system suppliers such as ASML Holding NV ADR, opening a new investment direction for the AI trade theme. On October 29, the Nasdaq 100 index and the Philadelphia Semiconductor Index fell about 1.38% and 1.42%, respectively; in the October 30 trading session they fell further sharply by 3.46% and 7.20%. The Nasdaq 100 unexpectedly posted its biggest single-day drop since June, while Intel Corporation plunged 28.58%, its worst single-day performance since 1974. The credit market explains the source of pressure behind this selloff. On October 15, the spread on CoreWeave's five-year credit default swaps (CDS) broke through 1,100 basis points, and the yield on its 2031 bonds reached about 16% at one point; default protection costs for Alphabet, Amazon.com, Inc. and Meta also hit new highs, but the increases were significantly smaller than for CoreWeave. Capital is raising its demands on the speed of AI investment payback and debt-servicing capacity. Alphabet's third-quarter free cash flow was negative US$9.065 billion, negative for two consecutive quarters; Amazon.com, Inc.'s free cash flow for the trailing 12 months ended September was a net outflow of about US$42.4 billion. Over the same period, Alphabet Inc. Class C cloud revenue grew 103% and AWS grew 39.3%, which is enough to show that the market debate over the AI trade centers on whether strong demand can be converted into cash returns covering investment in a timely manner. It should be noted, however, that negative free cash flow increases external financing needs but cannot be directly equated with operating losses or credit deterioration on the same scale as new cloud companies. From data center deliveries to storage capacity expansion: AI market seeks the next round of earnings growth The 2027 inflection point proposed by Fowler has clear industrial logic: only after data centers complete power supply, server deployment and acceptance can signed orders be converted into billable services; only when utilization rises is there a chance to improve returns on capital. Another key growth transmission path is storage demand driving record wafer fab capacity expansion, which in turn increases purchases of lithography, deposition and etching equipment. The investment focus is on finding forecast trends that have not yet fully priced in earnings optimism and continued expansion of growth pace, rather than inferring that share prices still have equivalent upside simply because order books are full. From the perspective of massive AI inference workloads, the continuous background execution of the globally popular AI agent Muse and the improved ability of frontier AI large model Astra on complex computer tasks have expanded the scope of work AI can handle; multi-step tasks, tool calls and parallel agents may in turn increase model invocations and context processing per user. Anthropic has observed in its research systems that token usage for multi-agent tasks is about 15 times that of ordinary chat. It can be inferred from this that with the full penetration of frontier AI agents such as Muse, AI computing demand will spread along GPU computing, HBM and DRAM capacity and bandwidth, KV cache and SSD storage, high-performance CPU tool execution and high-speed data transmission systems based on optical interconnect chips; total resource demand ultimately depends on the combined effect of task volume growth and per-task efficiency improvement. From an engineering logic perspective, agents expand a single Q&A into planning, retrieval, tool calls, code execution and result validation, and multi-round reasoning increases computing demand. Management of U.S. semiconductor equipment giant Applied Materials previously explicitly stated that agent applications urgently need massive AI accelerators such as AI GPUs and more CPU-intensive computing architectures, and will significantly increase DRAM and NAND demand over the long term, providing additional growth momentum for wafer fabrication equipment. This is also why "the world's chip foundry king" Taiwan Semiconductor Manufacturing Co., Ltd. Sponsored ADR reported third-quarter dollar revenue of US$46.72 billion, up 41.2% year over year, with a gross margin of 66.4%, and expects total chip manufacturing equipment-related investment from 2026 to 2028 to significantly exceed that of the previous three years. Therefore, the semiconductor equipment segment can be said to benefit from both "capacity expansion demand" and "process upgrades." The gate-all-around structure of advanced logic chips, high-layer-count 3D NAND, and HBM stacking and advanced packaging have raised technical requirements for EUV and even High-NA EUV lithography, precision deposition, selective etching, chemical mechanical polishing and defect control. Continued demand growth may drive both new production lines and upgrade demand for existing lines. Morgan Stanley's Wilson supplemented this judgment from the perspective of market rhythm: midterm elections, AI financing pressure and liquidity constraints may first push the S&P 500 down to 7,100 points, but after risks are digested he still looks toward 8,000 points. He believes companies adjusting between equity and debt financing will help them adapt to changes in the cost of capital; high-quality companies are better able to withstand this environment. Kobeissi's statistics showing AI stocks fell more than 13% in October while its "ex-AI" portfolio rose 8% also show that capital is rotating. UBS Group AG focuses on the AI-related investment cycle being longer than initial consensus expectations, while Wilson looks at entry opportunities after adjustment; both focus on whether earnings can weather current financing pressure.