2 Stocks to Buy on Overdone AI Infrastructure Spending Fears That Could Rise 30% and 50%, According to One Wall Street Analyst

BNP Paribas analyst Nick Jones says fears that Alphabet and Amazon are overspending on AI infrastructure are "overdone," arguing that rising backlog and improving revenue-per-employee metrics support the higher capex. Alphabet plans $175B-$185B of capex this year versus $91.4B in 2025, while Amazon raised capex to $200B from $131.8B last year. The note highlights structural advantages from proprietary AI chips and internal AI demand, with price targets of $390 for Alphabet and $320 for Amazon.

Analysis

The market is still treating hyperscaler capex as a demand-side gamble, but the more important signal is that these firms are moving from platform expansion to capacity monetization. For AMZN and GOOG, incremental AI spending is less about winning abstract cloud share and more about lowering unit economics through custom silicon, which should compress the payback period versus merchant GPU-dependent peers. That matters because the firms with proprietary chips can price more aggressively, lock in workloads longer, and defend margin even if headline capex looks bloated.

The second-order beneficiary is AVGO: custom ASIC demand is a structural, multi-year revenue stream that scales with hyperscaler internal adoption, not just external cloud demand. The biggest loser is the narrative that GPU scarcity automatically translates into durable pricing power for NVDA; if hyperscalers keep internalizing more workload to TPUs/Trainium, NVDA’s mix could become more exposed to enterprise inference demand and less to the highest-margin proprietary cloud deployments. INTC remains mostly an observer here unless it can re-enter as a foundry beneficiary of the custom silicon cycle, which is a longer-dated and less certain path.

The contrarian point is that the “overspending” debate may be backward-looking: the real risk is not overbuild today, but underbuild leading to product delays, slower model iteration, and lost search/e-commerce monetization. The catalyst path is likely months, not days: if cloud backlog and internal AI feature rollouts continue to accelerate, investors will eventually re-rate capex as maintenance of strategic option value rather than dilution of free cash flow. The key downside trigger is evidence that utilization lags capacity additions for two consecutive quarters, which would pull this trade back into a margin compression story.

Positioning-wise, this is a relative-value story, not a broad beta call. The cleaner expression is long GOOG/AMZN versus short a basket of capex-sensitive software or non-differentiated cloud names, while using AVGO as a secondary long on custom silicon demand spillover. NVDA should not be shorted outright, but it is vulnerable to multiple compression if investors conclude hyperscalers are substituting internal ASICs for purchased accelerator spend faster than expected.

Key Decisions for Investors

• Go long GOOG and AMZN on 1-3 month horizons as a quality-capex compounder pair; target 10-15% upside on multiple re-rating if backlog and AI product cadence hold, with stop-losses if management commentary shifts toward utilization concerns.
• Pair trade: long GOOG/AMZN vs short a basket of high-multiple, capex-insensitive software names over the next 2 quarters; thesis is that capital is being re-allocated toward infrastructure owners with proprietary economics.
• Add AVGO on any post-capex-announcement weakness for a 6-12 month horizon; custom ASIC content should continue to expand, and the risk/reward is attractive if hyperscaler internal chip programs keep scaling.
• Avoid shorting NVDA outright; instead, if positioning is extended, hedge with a modest NVDA call spread sale against a long GOOG/AMZN basket to express relative substitution risk without taking unlimited upside risk.
• Set a watch item for the next two earnings cycles: if utilization and cloud backlog fail to inflect despite rising capex, de-risk the longs quickly because the narrative could flip from strategic investment to margin dilution.