The 3 Best Trillion-Dollar Stocks to Buy Now. (Hint: Nvidia Isn't One of Them.

Alphabet is reportedly exploring a multibillion-dollar sale of Broadcom‑co‑designed TPUs to Meta, which would mark a shift from Google Cloud‑only TPU rentals to direct hardware sales and could challenge Nvidia's GPU dominance. Alphabet derives roughly $74B of $102B in Q3 revenue from advertising but would gain a new, not-yet-priced hardware revenue stream; Broadcom already reported $5.2B of $15.9B in fiscal Q3 revenue from AI (management expects $6.2B in fiscal Q4) and stands to earn design/ship payments per TPU; Taiwan Semiconductor, as the primary foundry, would benefit regardless of which accelerator wins and currently trades near 27.5x forward earnings. The article argues these dynamics make Alphabet, Broadcom and TSMC compelling buys if TPU sales to third parties proceed.

Analysis

Market structure: A confirmed Alphabet→Meta TPU sale would re-route a slice of hyperscaler AI spend from Nvidia GPUs to custom ASICs, directly benefiting GOOGL (new hardware revenue), AVGO (chip design fees) and TSM (foundry volumes). Expect pricing power to shift modestly — Nvidia’s ASPs could face 5–15% pressure in targeted large-scale inference workloads over 12–24 months while total AI compute spending grows ~20%+ annually, sustaining TSMC demand.

Risk assessment: Key tail risks are contract failure/regulatory blocks, TSMC capacity or yield setbacks, and slow customer migration due to CUDA lock‑in; any of these can erase near‑term alpha. Immediate moves (days) will be headline‑driven; validate in 2–8 weeks via Q4/earnings commentary and 3–12 month capex/capacity signals before scaling positions.

Trade implications: Tactical allocations: long GOOGL/AVGO/TSM and selective hedges against NVDA; use 9–15 month call spreads to control cost and buy downside protection on NVDA. Cross‑sector: increase exposure to semicap equipment and cloud software (+1–2% portfolio each) if Broadcom/Alphabet disclosure confirms multi‑year TPU revenue >$3–5B.

Contrarian angles: The market underestimates software switching costs — CUDA ecosystem and developer tooling remain a moat, so TPU penetration may stall below 20% share in 18 months. Also, TSMC’s neutrality is priced in but capacity tightness could actually push TSM shares higher if demand proves persistent; conversely Broadcom concentration risk could amplify earnings volatility.

Key Decisions for Investors

• Establish 2–3% long positions in GOOGL, AVGO and TSM each within 2–6 weeks; increase combined exposure to 5–7% if Alphabet confirms TPU sales >$3–5B or multi‑year supply commitments during the next earnings cycle.
• Open a relative value pair: long TSM (3% portfolio) / short NVDA (2%); trim the short if NVDA retains >80% share in large‑scale inference shipments on next quarterly survey, cover if NVDA share stabilizes above that threshold.
• Execute options: buy 9–15 month call spreads on GOOGL and AVGO (buy ATM, sell +20% strike) sized to equal 1–1.5% risk each; separately buy a 12‑month 10% OTM put on NVDA as tail hedge if implied vol >70% or NVDA rallies >20% before confirmation.
• If no TPU contract confirmation in 60 days, reduce GOOGL/AVGO exposure by 50% and reallocate 1–2% to semicap equipment names; stop‑loss on each position at −15% absolute or if public filings show TPU deal value < $1.5B.