BTC 
ETH 
XRP
SOL 
DOGE - This topic is empty.
-
Posts
-
What happened
Goldman Sachs strategists led by Ryan Hammond have put hard numbers on what the AI buildout must earn to justify itself. The note stress-tests the 2026–2027 compute investments of six hyperscalers — Alphabet, Amazon, Microsoft, Meta, Oracle and SpaceX — and finds they would need to generate about $1.42 trillion in cumulative revenue between 2028 and 2030 to earn a 15% annualized return on invested capital, as reported by Bloomberg and corroborating outlets.
The per-unit math: roughly $42 billion of upfront investment per gigawatt of computing capacity, and about $11.6 billion of annual revenue per gigawatt to clear the hurdle. The trajectory: roughly $405 billion in 2025, nearly doubling to about $750 billion this year, then jumping to $1.2 trillion in 2027 — more than 50% above this year’s level and above the roughly $1.1 trillion Wall Street consensus — plus another 12% to about $1.4 trillion in 2028. Total 2023–2030 capex approaches $6.2 trillion, and the note frames 2027 spending as a larger share of GDP than any tech investment cycle since the late-1800s railroad buildout.
The breakeven bar: hyperscalers need roughly $300 billion in annual AI revenue to cover their outlays. Cloud revenue across the group is annualizing about $70 billion above the pre-AI trend line as of Q2 2026 — leaving a gap of more than $200 billion. One telling contrast: Apple, with capex under $15 billion a year, has been the best-performing Magnificent Seven member in 2026. Goldman’s overall read stays constructive — token demand growth is explosive and the firm still expects AI capital invested over the next 18 months to earn a good overall return.
Why it matters
This gives the capex-scrutiny theme a number investors can track. The $200-billion-plus gap between today’s AI revenue run-rate and the $300 billion breakeven turns every earnings call into a return-on-spend audit. The math is toughest for Oracle and Amazon; Microsoft, Alphabet and Meta face more forgiving hurdles. In 2026, the market rewards capital efficiency over spending scale.
What to watch
The next verdict comes with Q3 earnings. Watch 2027 capex guidance from each hyperscaler, commentary on token and compute demand on the calls, and whether anyone trims their buildout plans. Key data points: cloud revenue acceleration against that $70 billion above-trend run-rate, and whether the $300 billion breakeven gap starts to narrow.
Related reading
NVDA vs AMD: Which Semiconductor Stock Fits Your Portfolio?
Intel Says 14A Will Land Within 5% of TSMC’s A14Discussion: Do you think hyperscaler AI capex will earn the roughly $300 billion in annual revenue Goldman’s note says the buildout needs — or will the 2027 spending guides get cut when the numbers don’t close?
Sources
https://startupfortune.com/goldman-sachs-says-hyperscalers-will-spend-12-trillion-on-ai-in-2027/
https://finance.biggo.com/news/00d0de5b-5dfe-461d-969b-d3161848ec79
https://www.cointime.ai/flash-news/goldman-sachs-41128Auto-posted by the Bull Lounge news desk.
Not financial advice.Log in to vote
