Jim Cramer Says 'This Is Not A Circular Deal' As Amazon, Anthropic Lock In Massive $100 Billion AI Pact: 'Isn't It Possible That Everyone Wins?'

Amazon and Anthropic unveiled a long-term AI partnership centered on more than $100 billion of Anthropic spending on AWS over the next decade, plus an additional $5 billion Amazon investment and up to $20 billion more tied to milestones. The deal also gives Anthropic access to Amazon's upcoming Trainium3 chips later this year. While the agreement underscores massive AI infrastructure demand, it has also revived debate over so-called circular AI deals and competition concerns.

Analysis

The more important read-through is not the headline size of the commitment, but the signal that AI capex is shifting from discretionary experimentation to contracted, multi-year infrastructure demand. That is structurally bullish for AWS utilization and for the suppliers that sit closest to compute bottlenecks, because the market will increasingly value secured demand over near-term unit pricing. The second-order effect is that the largest hyperscalers may get more leverage over model labs by financing their scale-up, which should compress the bargaining power of smaller cloud buyers and independent developers.

The competitive implication for NVDA and AMD is mixed rather than uniformly positive. In the near term, large pre-commitments validate continued AI spending, but the deeper strategic effect is that hyperscalers will push harder to diversify away from merchant GPUs into custom silicon and vertically integrated stacks. That shifts long-run value capture toward cloud platforms and away from pure-play accelerator vendors, especially if Trainium-class adoption proves good enough for training marginal workloads and inference economics.

The contrarian point is that the market may be over-fixated on circularity while underestimating the balance-sheet and antitrust asymmetry. If these arrangements are deemed to create artificial demand concentration, the risk is not immediate revenue erosion but a slower multiple compression on the entire AI infrastructure complex as regulators scrutinize financing-linked purchasing. The catalyst path is months, not days: any evidence that model performance, utilization, or customer growth fails to scale with committed spend would quickly turn this from a growth narrative into a capital-allocation question.

Key Decisions for Investors

• Long AMZN vs. short a basket of AI hardware suppliers over 3-6 months: buy AMZN and short equal-dollar NVDA/AMD on the thesis that AWS captures more durable economics than merchant silicon as custom chips gain share; target 10-15% relative outperformance, stop if NVDA reaccelerates data-center backlog growth.
• Buy AMZN calls or call spreads into the next 1-2 earnings prints: the market should start assigning higher probability to multi-year AI revenue visibility; use defined-risk upside exposure because the core catalyst is contract monetization rather than near-term margin expansion.
• Reduce or hedge NVDA/AMD longs on strength for 1-3 months: the deal supports AI demand, but it also increases the odds of customer vertical integration, capping multiple expansion; use call overwriting or small short-dated puts if the names rally on AI enthusiasm.
• Watch for antitrust headlines and regulatory commentary over the next 6-12 months: if financing-linked purchasing becomes a policy target, expect a sector-wide de-rating; consider a tactical short basket of mega-cap AI infrastructure names versus defensives if scrutiny escalates.