Bearish on Anthropic: growing AI demand may not translate into durable profits if model competition weakens pricing power while infrastructure commitments keep costs high.
-
“Good enough” models could erode pricing power. As cheaper alternatives improve, enterprises may reserve Claude’s frontier models for their hardest tasks and route routine work elsewhere. Anthropic could handle more workloads while earning less per task. Staying ahead would require continued research spending, making technical leadership costly to defend.
-
Cloud partners could capture more of the value. Anthropic depends on infrastructure providers that also distribute competing models and develop their own. Their incentive to offer customers interchangeable models could weaken Anthropic’s differentiation and bargaining power. AI adoption could benefit the infrastructure layer more consistently than individual model developers.
-
Flexible revenue sits against committed costs. Customers can reduce consumption or switch workloads faster than Anthropic can unwind infrastructure obligations. If demand growth slows or competition forces prices lower, those commitments could pressure margins and cash flow. Maintaining frontier capabilities adds further spending demands.
Together, these risks challenge the durability of Anthropic’s moat: technical barriers can keep new entrants out without protecting profits from established competitors. My bearish view is that Anthropic may create substantial customer value while retaining too little of it to deliver attractive shareholder returns.
