Top Economist Warns That AI Data Center Investments Are "Digital Lettuce" That's Already Starting to Wilt

Nvidia beat Wall Street revenue expectations this week, reinforcing investor confidence in the trillions of dollars companies such as OpenAI, Microsoft and Oracle plan to spend on AI infrastructure, but economists and prominent investors are warning of an AI hardware-driven bubble. Critics including David McWilliams call GPUs a “perishable” asset that degrade under constant use and risk rapid obsolescence, and Michael Burry has argued firms may be materially overstating future earnings (he cited potential overstatements of ~26.9% for Oracle and ~20.8% for Meta by 2028). The debate is increasing market nervousness about valuation sustainability and raises the risk that accelerated depreciation and technological change could undermine the economics of large-scale data‑center buildouts, even as opinions differ on the broader macro impact.

Analysis

Nvidia reported revenue that beat Wall Street expectations this week, reinforcing investor confidence in the large-scale AI infrastructure buildouts that companies such as OpenAI, Microsoft and Oracle are planning to fund; the article characterizes Nvidia’s results as “soaring revenues” and the per-ticker sentiment output ranks NVDA positively (0.7). Investors remain willing to underwrite multitrillion-dollar AI spend despite mixed underlying fundamentals, and the market impact score (0.55) suggests these developments meaningfully influence broader tech sentiment.

Prominent critics and economists warn the economics of that buildout are fragile: David McWilliams calls GPUs “digital lettuce,” arguing hardware degrades under 24/7 use and risks rapid obsolescence, while Michael Burry alleges firms are understating depreciation—citing projected 2028 overstatements of 26.9% for Oracle and 20.8% for Meta. Ameriprise strategist Anthony Saglimbene told Bloomberg the unprecedented scale of upgrades is fueling investor nervousness, consistent with an overall moderately negative sentiment score (-0.45).

The practical implication is a bifurcated opportunity/risk profile: Nvidia’s near-term earnings can remain strong, but companies with aggressive server valuation and lenient depreciation assumptions (notably ORCL and META per the article) face material re‑rating risk if hardware lifecycles are shorter than assumed; market ripple effects are possible even if some economists expect limited macro fallout.

Key Decisions for Investors

• Consider maintaining selective exposure to NVDA to capture near-term revenue momentum but size positions cautiously and employ hedges to protect against a rapid re-rating
• Reduce or hedge holdings in firms flagged for aggressive hardware accounting (Oracle, Meta) until companies disclose conservative depreciation schedules and clearer capex-to-revenue conversion,
• Monitor quarterly capex guidance, server depreciation policies, utilization metrics and announced AI infrastructure contracts as primary triggers to add risk or de-risk positions