Many of these larger customers “are not locked into long-term contracts with us,” Anthropic writes. That means they could reduce their spending on Anthropic’s models at any time, whether due to a shift toward competitive models, less reliance on AI, or if they “develop their own solutions that compete with our models.” That last risk is particularly acute, Anthropic notes, for “software as a service and other software companies, whose offerings may incorporate AI functionality similar to or dependent on the capabilities we offer.”
What’s more, Anthropic notes that many of its large clients “may rely on venture capital or other external financing to fund their operations.” That kind of spending on AI services could drop quickly if the venture capital spigot starts to dry up.
Collaborators or competitors?
Anthropic is in an awkward position: Companies like Amazon, Google and Microsoft are its biggest cloud computing providers and some of its biggest competitors in AI research. That situation “creates complex dynamics that could give rise to conflicts of interest and negatively affect our access to computing,” as Anthropic says.
These companies’ position as Anthropic’s cloud providers gives them the unique opportunity to increase Anthropic’s prices, terminate contracts with Anthropic, or establish more favorable relationships with competitors, “for competitive or other reasons,” the company writes. Their position also gives these cloud providers a competitive advantage through “visibility of [Anthropic’s] prices and commercial terms on the platforms they operate.”
Anthropic is working to secure direct data center leases to avoid purchasing on-demand cloud computing resources from its own competitors. But the $125 billion in data center commitments Anthropic has made come with their own problems. As Anthropic writes, “if our computing requirements or technology needs change, we may find ourselves with leased or dedicated infrastructure that is not a good fit for our actual needs, and technological changes in AI hardware or training methodologies could make such infrastructure less efficient than newer alternatives.”
