The AI stack is becoming the cloud battleground
A subtle shift is taking place in how cloud providers win usage among AI startups. For years, competition focused on compute availability, enterprise relationships, and pricing. But as AI teams increasingly begin with model experimentation and development tools, the AI … The post The AI stack is becoming the cloud battleground appeared first on CB Insights Research .
The AI stack is becoming the cloud battleground as cloud providers vie for the attention of AI startups. For years, competition among cloud providers focused on factors such as compute availability, enterprise relationships, and pricing. However, as AI teams increasingly begin their workflows with model experimentation and development tools, the AI stack they choose first is shaping where they run and scale their workloads. This subtle shift is reshaping how startups make cloud decisions, raising questions about what's driving cloud adoption among AI builders today and where providers are gaining or losing traction.
Google's AI-native stack is converting early experimentation into long-term cloud usage. The company's integrated Gemini ecosystem, which includes Google AI Studio, Vertex, and its expanding infrastructure footprint through Cipher Mining, is pulling AI teams into Google Cloud Platform (GCP) earlier in their workflow. This strategy is proving effective, as cloud backlog surged 46% quarter-over-quarter to $155 billion in Q3'25, with AI revenue as a key driver. Startups that begin experimenting in this environment often scale directly onto GCP, contributing to Google Cloud's 4-point rise to 38% and its faster year-over-year growth. As these teams move from experimentation to production, cloud spend compounds inside Google's ecosystem.
AWS's neutrality remains a competitive position, even as its share dips from 33% to 30%. The "Switzerland" stance of AWS still appeals to AI startups navigating an ever-expanding model landscape. Without a proprietary large language model (LLM) anchoring its stack, AWS attracts teams seeking broader compute availability, large credit packages, and interoperability with multiple third-party models. This neutrality may become more valuable as founders try to avoid lock-in and as model providers diversify.
Microsoft Azure's enterprise-first partnerships aren't driving startup adoption as effectively as its competitors. While Azure offers a robust set of AI services, its focus on enterprise clients may be limiting its appeal to AI startups. However, Azure's integration with Microsoft's broader suite of tools and services could still make it an attractive option for startups looking to leverage a cohesive ecosystem.
The GPU scarcity and evolving model ecosystems are further complicating the cloud landscape. Hyperscalers are adopting different approaches, ranging from integrated AI stacks to model neutrality to enterprise partnerships. These varying strategies are forcing AI startups to carefully consider their cloud provider choices based on their specific needs and long-term goals.
In conclusion, the AI stack is becoming a critical factor in the cloud battleground. Providers like Google and AWS are adapting their strategies to attract AI startups, while Microsoft is focusing on enterprise partnerships. The shifting landscape is driven by factors such as GPU scarcity, evolving model ecosystems, and the need for startups to avoid lock-in. As AI teams prioritize experimentation and development tools, the choice of AI stack will increasingly determine where they run and scale their workloads, ultimately shaping the trajectory of their cloud adoption.
