According to Morgan Stanley, Anthropic’s expanded collaboration with Google Cloud isn’t just a minor upgrade, it’s a potential game-changer. Analyst Brian Nowak estimates this deal could add a whopping 100 to 900 basis points to Google Cloud’s 2026 revenue. But wait, there’s more! By 2027, this partnership could contribute an estimated $9 billion to $13 billion annually.
What’s fueling this potential revenue surge?
Anthropic gains access to up to 1 million TPUs (Tensor Processing Units) from Google. These powerful processors provide the extra muscle needed to train and deploy the next generation of Claude models. Morgan Stanley notes that this deal, valued in the “tens of billions of dollars,” positions Anthropic to have “well over a gigawatt of capacity online in 2026.” The firm believes Google’s TPUs offer a compelling blend of price-performance and cost efficiency, a critical factor as Anthropic scales its computing capacity to meet growing demand. For more information about TPUs.
This isn’t just about hardware; it’s a testament to the strength of Google’s AI Cloud offering. Morgan Stanley believes the Anthropic deal highlights Google’s comprehensive AI cloud capabilities, encompassing infrastructure, models, and services. While Amazon remains Anthropic’s primary training provider, Google Cloud is poised to handle the inference workloads. With the TPU v7 Ironwood chip, designed specifically for inference and set to ramp up in 2026, Google is positioning itself as a leader in AI inference.
Here’s the kicker
Morgan Stanley anticipates Anthropic spending a staggering $50 billion to $80 billion with Google Cloud over the next six years. With full-year contributions expected in 2027, this deal is more than just a fleeting partnership. It’s a long-term investment that could significantly reshape the competitive landscape of cloud computing. As Morgan Stanley aptly puts it, this deal is “another tailwind to an already accelerating GCP business.” So, keep your eyes on Google Cloud – it looks like they’re just getting started!
