Vendor strategy shifts

Software teams buying artificial-intelligence models are increasingly choosing vendors that can be deployed across more than one cloud, and Anthropic now sells Claude through its own API, Amazon Bedrock, Google Cloud Vertex AI, and Microsoft Foundry. (anthropic.com) That footprint is visible in current documentation from all three hyperscalers. Amazon Web Services lists Anthropic Claude in Bedrock, Google documents Claude as a partner model on Vertex AI, and Microsoft now has setup guides for Claude in Foundry. (aws.amazon.com) (cloud.google.com) (learn.microsoft.com) Anthropic’s own developer materials also steer builders toward those channels. Its “Build with Claude” guide links to courses for Amazon Bedrock and Google Cloud Vertex AI, while the Claude Opus 4.7 product page says the model is available natively and through Bedrock, Vertex AI, and Microsoft Foundry. (anthropic.com 1) (anthropic.com 2) A cloud marketplace listing is not the same thing as model quality, but it changes how companies buy software. Procurement teams that already have spending commitments with Amazon, Microsoft, or Google can often purchase usage through those existing accounts instead of adding a new standalone vendor. (cloud.google.com) (learn.microsoft.com) (aws.amazon.com) The contrast with OpenAI is sharper in 2026 because Microsoft and OpenAI reaffirmed their platform ties in February. Their joint statement said Azure “remains the exclusive cloud provider of stateless OpenAI APIs,” even as OpenAI kept selling APIs directly to customers. (openai.com) (blogs.microsoft.com) That arrangement has loosened at the edges, but not at the API layer most software teams buy. OpenAI and Microsoft said in October 2025 that some non-API products could be served on other cloud providers, while API products developed with third parties would remain exclusive to Azure. (openai.com) (blogs.microsoft.com) Google’s Gemini strategy is different again: Google sells Gemini through its own Gemini API and through Vertex AI, both inside Google’s ecosystem. Google’s current documentation positions Vertex AI as its managed platform for building with Gemini and other foundation models. (ai.google.dev) (cloud.google.com) That leaves buyers comparing two separate questions at once: which model performs best on coding, reasoning, or support tasks, and which vendor creates the fewest infrastructure constraints. Microsoft pitches Azure’s security and enterprise controls for OpenAI, while Google and Amazon pitch managed access to multiple model families on their own platforms. (learn.microsoft.com) (cloud.google.com) (aws.amazon.com) Anthropic has not publicly claimed to be the default choice for software teams, and “default” depends on price, latency, region, and internal policy as much as model rankings. But as of April 2026, Claude is one of the few frontier model lines that can be bought natively and through all three major United States cloud platforms, which is why the procurement debate has shifted from model scores to portability. (anthropic.com) (cloud.google.com) (learn.microsoft.com)