AI app builders and the new era of cloud infrastructure deals
AI app builders are specialized platforms, tools, and services that help teams design, generate, and deploy AI-powered applications, and they are becoming some of the most demanding and long-term customers for cloud infrastructure as every prompt, model call, and deployment pipeline consumes large amounts of compute. The shift is visible in how companies that began as lightweight tools are now signing large-scale cloud infrastructure deals and embedding deeply with a single provider’s stack. Instead of running quick trials of AI app development, enterprises are committing to multiyear agreements that cover training, inference, security, and deployment at scale. These arrangements are less about basic hosting and more about controlling the next layer of software creation: AI-driven code generation, agents, and full application workflows. Cloud providers are responding with tailored partnerships that bundle AI models, specialized chips, marketplaces, and governance features, locking in these builders before they grow into even larger customers.
Lovable’s fivefold Google Cloud partnership and AI app development scale
Lovable’s expanded Google Cloud partnership shows how quickly an AI app builder can become a heavyweight infrastructure client. The company agreed a multiyear expansion that increases its cloud footprint fivefold, with significantly greater AI usage tied to both Anthropic’s Claude models and Google’s Gemini models for coding tasks. Lovable reports more than 25 million projects created in its first year and over one million new projects processed each week, while Lovable-built apps attract 600 million visits per month. This growth is backed by serious revenue. According to TechCrunch, Lovable crossed USD 400 million (approx. RM1,840 million) in annual recurring revenue with 146 employees, after raising a USD 330 million (approx. RM1,518 million) Series B at a USD 6.6 billion (approx. RM30,360 million) valuation. The Google Cloud partnership deepens access to Gemini models, places Lovable Agent in the Gemini Enterprise Agent Gallery, and adds Marketplace distribution plus Wiz-powered security, turning a “prompt-to-app” tool into a full-stack AI app development platform with enterprise-grade infrastructure.
Pinterest’s USD 4 billion AWS enterprise commitment to AI discovery
At the application layer, Pinterest is showing how a mature consumer platform can transform into a long-term cloud infrastructure anchor tenant for AI. The company has announced a planned USD 4 billion (approx. RM18,400 million) commitment to Amazon Web Services through 2031, its largest infrastructure investment so far, to support the next phase of AI and machine learning growth. Serving more than 600 million monthly users, Pinterest depends on AI models for visual search, recommendations, its Taste Graph, multimodal systems, and the Pinterest Assistant. Under the expanded agreement, Pinterest plans to use AWS Trainium to train and run large language and vision-language models, and to increase its use of AWS Graviton processors, which already power about one-third of its computing infrastructure. The company is also shifting from traditional EC2-based environments to a Kubernetes architecture on Amazon EKS, aiming for better efficiency, reliability, and developer productivity while it scales AI-driven discovery and advertising features.
From experiments to long-term cloud infrastructure customers
Lovable and Pinterest together show that AI app development is no longer a side experiment for cloud providers. Lovable’s rapid move from buzzy startup tool to infrastructure-heavy platform, and Pinterest’s multibillion-dollar AWS enterprise commitment, both signal that AI application builders expect to run sustained, intensive workloads over many years. Their usage looks less like one-off productivity boosts and more like recurring, infrastructure-heavy workflows. For Lovable, every generated app, security scan, deployment workflow, and agentic task consumes compute inside Google Cloud, while Pinterest’s AI recommendation and discovery systems continually train and serve models on AWS. These patterns turn AI builders into predictable, high-volume customers. Instead of trying different clouds for each project, these companies are aligning their future roadmaps with specific providers, from AI models and chips to Kubernetes stacks and enterprise marketplaces. The result is a new class of cloud infrastructure deals anchored in AI application lifecycles rather than generic hosting contracts.
How cloud providers are locking in AI builders for the full lifecycle
Cloud providers are shaping AI-focused deals that span the entire lifecycle of AI app development: building, securing, deploying, and operating applications at scale. In Lovable’s case, Google Cloud is not only supplying AI-optimized infrastructure and access to Gemini, but also offering Marketplace distribution, simplified procurement, and integrations like Wiz to scan and fix vulnerabilities in AI-generated code. That gives enterprise buyers governance, billing, and security features they need before letting AI-generated software touch customer data or regulated workflows. Pinterest’s AWS agreement shows a similar pattern at a different tier. The company is standardizing on Trainium and Graviton hardware, moving to Amazon EKS for container orchestration, and relying on AWS to support both consumer experiences and advertising performance. These specialized partnerships give AI builders a consistent stack from experimentation to production, while tying their success to a particular cloud ecosystem. In turn, cloud giants secure loyalty from customers whose AI usage is likely to rise over time.
