Elon Musk Says He Was "Clearly Wrong" About Anthropic's Artificial Intelligence (AI) Models. Here's Why That's Outstanding News for Amazon and Alphabet Investors.

Musk says he was wrong about Anthropic and now views it as the clear AI leader, citing Mythos/Fable as highest-quality available. The article argues Anthropic’s scaling uses more custom silicon (Amazon Trainium/Inferentia; Google Cloud TPUs) and increases AWS/GCP utilization, supporting cloud growth and operating leverage, noting AWS revenue +28% YoY and operating margin up to 38% alongside GCP sales +63% YoY with operating margins >30%. It also claims Alphabet and Amazon have upside as forward P/E compression has not fully priced in Anthropic’s next model compute demands (e.g., Mythos 2).

Analysis

The market mechanism here is not “AI sentiment” but cloud utilization leverage: if Anthropic is increasingly perceived as the frontier leader, AWS and GCP are likely to see incremental workloads first in inference, then in training bursts as new model generations ship. That is a cleaner near-term earnings tailwind for AMZN and GOOGL than for the app-layer AI names, because the monetization is tied to metered compute and custom silicon saturation, not to eventual consumer adoption.

The second-order risk is that Anthropic’s multicloud setup weakens the exclusivity value of any single provider and may keep pricing discipline tighter than bulls expect. Also, the more Anthropic shifts volume onto Trainium, Inferentia, and TPU stacks, the more upside accrues to hyperscaler margins and the less pass-through there is to NVDA’s share of frontier-AI capex. Over 6-18 months, this could become a subtle relative-value trade: hyperscalers capture operating leverage while semiconductor beta broadens less than headline AI demand suggests.

Consensus may be overrating the signaling value of a public endorsement and underrating execution risk: model leadership does not automatically convert into durable revenue unless Anthropic keeps shipping and customers actually scale spend. The thesis is falsified if next cloud prints show decelerating AI attach rates, if Anthropic signs a materially more favorable exclusive deal with a rival, or if custom silicon utilization lags and compute demand leaks to competitors. Near term, the setup is more about sentiment confirmation; the real catalyst path is the next 1-3 cloud earnings calls and any update on Anthropic workload growth.

Key Decisions for Investors

• Add modestly to AMZN and GOOGL on 1-2 day pullbacks, targeting a 1-3 month trade into cloud earnings; asymmetry is favorable if AI-related utilization keeps reaccelerating, but size should be small because the marginal information content is limited.
• Pair trade: long AMZN / short NVDA for 1-3 months if you believe frontier-AI demand is shifting toward hyperscaler-owned silicon; this expresses compute growth while hedging against a slower-than-expected mix benefit to merchant GPU demand.
• For a cleaner relative-value expression, long GOOGL vs short MSFT over 1-3 months if you think Anthropic’s multicloud footprint and TPU leverage are underappreciated; risk is that Azure/OpenAI momentum reasserts itself and overwhelms the signal.
• Set an alert on AWS and GCP commentary around AI capacity, custom chip utilization, and margin expansion at the next earnings prints; if AI capex intensity fails to translate into revenue acceleration, take profits quickly because the endorsement trade will fade.
• No options recommendation unless implied vol compresses meaningfully; this is better treated as a fundamental long-only add or a small relative-value pair than a directional gamma trade.