Clouds are becoming agent platforms

Cloud vendors are turning artificial intelligence agents into a platform feature, not just a model add-on. GitLab, Cadence and Technogym each announced new Google Cloud tie-ins on April 14 and April 15 that package agents inside existing cloud buying and deployment paths. (about.gitlab.com) (cadence.com) (technogym.com) An agent is software that can carry out a chain of tasks, like a digital worker that plans, calls tools and hands work to other systems. In GitLab’s case, the company said customers can now run its artificial intelligence gateway on Google Cloud and use Vertex AI models already approved inside their organizations. (about.gitlab.com) GitLab said customers do not need separate artificial intelligence infrastructure and can count Duo Agent Platform usage toward existing Google Cloud commitments. The company announced the expanded collaboration on April 14, one day before investor coverage pushed the news into markets. (about.gitlab.com) (fool.com) Cadence made a similar move on April 15 in chip design, a field where engineers use software to place, test and verify billions of transistors before a factory ever builds a chip. Cadence said it is optimizing its ChipStack AI Super Agent with Gemini on Google Cloud and making the product available now through Google Cloud Marketplace as a click-to-deploy service. (cadence.com) Technogym’s April 14 deal shows the same pattern in a different industry. The fitness equipment company said its multi-year collaboration with Google Cloud will feed Google’s generative artificial intelligence into the Technogym ecosystem, including an AI Coach for personalized training and an AI Assistant for business operations. (technogym.com) The common thread is distribution. These companies are not only adding models; they are plugging task-running software into cloud contracts, marketplaces and governance systems that customers already use to buy computing and control data access. (about.gitlab.com) (cadence.com) (cloud.google.com) Google Cloud has been arguing for that model in public. In a March 26, 2026 post, Google Cloud said companies are moving from chatbots to agents that run “entire workflows from start to finish,” and in a separate report Google pointed to cross-platform agent systems built around the Agent2Agent protocol. (cloud.google.com) (blog.google) (developers.googleblog.com) That is changing the cloud sales pitch. Instead of selling raw computing, storage and model access by themselves, cloud providers are increasingly selling pre-wired agents for software development, semiconductor design, fitness coaching and other industry workflows. (about.gitlab.com) (cadence.com) (technogym.com) There are still limits. None of the three announcements gave customer adoption numbers, pricing detail beyond existing cloud-commitment language, or independent data showing that the new agents cut costs or improve output at scale. (about.gitlab.com) (cadence.com) (technogym.com) But the direction is now visible in product form: agents are being wrapped inside cloud infrastructure, sold through cloud channels and tied to cloud spending. That makes the cloud look less like rented servers and more like a shelf for specialized digital workers. (about.gitlab.com) (cadence.com) (technogym.com)