From Oracle to Broadcom, the Concerns About Artificial Intelligence Stocks Are Starting to Pile Up

Investor skepticism is rising around richly valued AI stocks as the sector shifts from growth euphoria to scrutiny over whether massive infrastructure spending will deliver acceptable returns; companies are committing tens-to-hundreds of billions to data centers, power and cooling, and investors are increasingly questioning the economics. Oracle, after reporting strong demand and more than $450 billion of remaining performance obligations, has faced reports it may need to raise $38 billion of debt, disclosed AI data-center margins of roughly 10–20%, narrowly missed revenue (~$16.1 billion) in its most recent quarter, raised full-year capex guidance from $35 billion to $50 billion and posted about negative $10 billion of free cash flow—moves that pushed five-year CDS on its debt to record highs. Broadcom beat near-term numbers but warned of declining gross margins and an understated AI backlog, while IBM CEO Arvind Krishna warned the math may not work—estimating ~$80 billion to build a 1‑gigawatt data center and implying ~$1.5 trillion for 20–30 GW—leaving returns and valuations for AI investments unclear and prompting investors to reassess and potentially trim exposure.

Analysis

Investor sentiment toward large AI-related technology names has shifted from enthusiasm to scrutiny as valuations climb and capital intensity becomes apparent. The article highlights that major players are committing “tens of billions, if not hundreds of billions” to AI infrastructure and that the market is questioning whether such spending will produce acceptable returns, given data-center, power and water constraints.

Oracle illustrates the tension: the company reported more than $450 billion of remaining performance obligations driven by its AI cloud services yet in its most recent fiscal-quarter revenue of nearly $16.1 billion narrowly missed consensus, it raised full-year capex guidance from $35 billion to $50 billion, reported roughly negative $10 billion of free cash flow for the quarter, and five-year CDS on its debt hit record highs; media reports also suggested a potential $38 billion debt raise and AI data-center margins of only 10%–20%.

Broadcom beat near-term consensus but warned of declining gross margins and an understated AI backlog, while IBM’s CEO argued the math on hyperscaler returns is dubious (citing ~$80 billion to build a 1‑GW data center and ~$1.5 trillion for 20–30 GW). These developments imply elevated execution, credit and resource risks for highly leveraged, capex-intensive AI strategies and justify re‑pricing and reassessment of growth assumptions across the sector.

Key Decisions for Investors

• Reassess allocations to high-valuation AI infrastructure names and consider trimming positions where valuation relies on aggressive margin or scale assumptions
• Monitor company-level signals — specifically capex guidance revisions, free cash flow trends, reported AI data-center margins, and five-year CDS spreads — as primary indicators of funding and credit stress
• Avoid adding leverage to portfolios concentrated in capex-intensive AI plays; prefer firms with stronger near-term free cash flow or diversified revenue streams
• Consider hedging concentrated exposure if credit-market indicators (CDS) and capex increases continue to deteriorate, while waiting for clearer margin expansion or evidence of sustainable returns
• Favor companies that demonstrate lower incremental capex per unit of AI capacity or materially higher reported AI margins before increasing exposure