AI is likely to prove a transformative technology in the long term, but predicting its short-term impact on share prices is by nature speculative and subject to swings in sentiment. (UBS)

Meanwhile, Alphabet has said it is rolling out generative AI search tools in India and Japan, matching features in the US, and boosting AI chip purchases from Nvidia to offer more AI training model capacity to its customers.


The surge in AI applications and investments has created a powerful new narrative for the broader tech sector. In July, we raised our long-term AI end-demand forecasts from 20% compound annual growth rate (CAGR) during 2020–25 to 61% during 2022–27. As a result, we now expect global AI demand to grow from USD 28bn in 2022 to USD 300bn in 2027. Second-quarter tech results, including the reacceleration in datacenter capital spending seen at top hyperscaler companies (large-scale cloud service providers) and Nvidia’s strong guidance on AI chip delivery, suggest our USD 300bn forecast may be too conservative.


But rich valuations may limit near-term upside. We suggest investors maintain exposure but balance near-term optimism against other portfolio considerations:


  • Within equities, the overall US tech sector’s price/earnings ratio relative to the S&P 500 stands at 1.4x, its highest level since the global financial crisis. We continue to favor laggards whose valuations are lower and have scope for a catch-up—including some smaller AI beneficiaries.

  • Within tech, we think software and internet stocks are best positioned to ride the next wave of the technology cycle and the broadening of AI demand as companies try to monetize AI applications. We expect the applications and models segment of AI to deliver USD 170bn in revenues (139% CAGR), versus USD 130bn for the semiconductor and hardware-centric infrastructure segment (38% CAGR) in 2027. This is consistent with our tech playbook, which calls for a switch into mid-cycle segments like software and internet, and away from semiconductors and hardware, which have already performed well in 2023.

  • Given this year’s sharp rally in mega-tech names, we keep our preference for equal-weighted US indexes compared to capitalization-weighted ones. Investor flows support this view. Between the beginning of June and 25 August, the Invesco S&P 500 equal-weighted ETF, which mirrors the performance of the S&P 500 equal-weight index, attracted inflows of USD 8.8 billion. By contrast, the SPDR S&P 500 ETF Trust, which tracks moves in the market cap weighted S&P 500, suffered outflows of USD 10.9 billion over the same period, according to Refinitiv Lipper data.

AI is likely to prove a transformative technology in the long term, but predicting its short-term impact on share prices is by nature speculative and subject to swings in sentiment. So we retain a selective approach toward the tech sector.


Main contributors - Solita Marcelli, Mark Haefele, Vincent Heaney, Jon Gordon, Sundeep Gantori


Original report - Take a selective approach to AI exposure, 31 August 2023.