Key Takeaways
- Jim Covello from Goldman Sachs suggests investors pivot from semiconductor firms to cloud hyperscalers including Amazon, Microsoft, and Alphabet
- Cloud platform providers have experienced valuation compression due to investor doubts about artificial intelligence investment returns
- Semiconductor stocks have surged approximately 150% over the last twelve months via the Philadelphia Semiconductor Index, creating stretched valuations
- Covello identifies dual pathways for hyperscaler investment success: demonstrating AI profitability or reducing expenditures to enhance cash generation
- The primary downside scenario involves continued heavy spending by hyperscalers without tangible return evidence
Jim Covello, a senior analyst at Goldman Sachs, believes the investment community has been targeting the incorrect segment of the artificial intelligence opportunity. He contends that cloud hyperscalers currently present superior value compared to semiconductor manufacturers.
Covello serves as co-head of equity research at Goldman Sachs while maintaining coverage of the semiconductor industry. He presented his thesis in a Thursday client communication.
The core of his position revolves around relative valuations. Cloud infrastructure leaders such as Amazon, Microsoft, Alphabet, Meta, and Oracle have experienced price-to-earnings multiple contraction. This compression stems from market uncertainty regarding whether these corporations will generate adequate returns from their substantial AI capital deployments.
Semiconductor manufacturers, in stark contrast, have emerged as Wall Street’s preferred artificial intelligence play. The Philadelphia Semiconductor Index has climbed roughly 150% during the preceding year.
This substantial rally has elevated chip manufacturer valuations beyond their long-term historical norms. Cloud hyperscalers, conversely, continue trading beneath their typical valuation benchmarks.
Dual Scenarios Supporting the Investment Thesis
Covello outlined two potential outcomes where favoring hyperscalers over chip companies could deliver strong results.
Under the first scenario, cloud platforms begin demonstrating favorable returns from their AI investments. Such evidence would alleviate market apprehension and drive their equity valuations upward. Semiconductor stocks would possess limited appreciation potential given their already-rich valuations.
The alternative scenario involves hyperscalers scaling back capital expenditures when investment returns remain disappointing. While this appears negative superficially, Covello maintains it could actually benefit their stock performance through improved free cash flow generation. Simultaneously, this reduction would negatively impact semiconductor firms dependent on that infrastructure spending.
The Primary Downside Scenario
The principal threat Covello identified involves an intermediate outcome. Should hyperscalers maintain elevated spending levels without delivering transparent return evidence, their equity prices could remain suppressed.
This identical situation would sustain robust chip demand, thereby continuing to underpin semiconductor company performance.
Covello’s analysis arrives as major technology corporations face intense investor scrutiny regarding their data center and AI infrastructure capital allocation.
Amazon, Microsoft, and Alphabet have each announced substantial capital expenditure programs extending through 2025 and subsequent years.
Meta has similarly accelerated its AI-related investments, attracting questions about whether this spending will convert into meaningful revenue growth.
Oracle has established itself as a significant participant in AI cloud infrastructure and has reported robust customer demand for its platform offerings.
The semiconductor industry has profited considerably from this infrastructure spending surge, with chip suppliers for AI data centers experiencing healthy order backlogs.
Covello’s recommendation represents a contrarian perspective given recent market dynamics, though it’s grounded in current valuation levels compared to historical averages.
