AWS partners with vibe-coding startup Superblocks for private-cloud AI apps
Superblocks has signed a multi-year agreement with AWS to embed its vibe-coding tool in enterprise private clouds, keeping data within customers' AWS environments. The deal reflects a broader push by cloud providers to separate AI models from enterprise AI infrastructure.
Superblocks, a startup in the "vibe coding" space, has announced a multi-year joint marketing agreement with Amazon Web Services. The partnership lets AWS customers embed Superblocks' tool inside their private cloud environments, allowing companies to offer AI-assisted application building to business users while keeping all data inside their own AWS accounts.
According to the company, applications created with Superblocks on AWS will not send data to external model providers or databases. Instead, they will spin up Amazon Aurora databases within the customer's private cloud and integrate with Amazon Bedrock, AWS's AI platform. That means the apps fall under existing IT management and security controls rather than becoming shadow IT.
Superblocks co-founder and CEO Brad Menezes said the key advantage is that data never leaves the customer's AWS environment. Everything is secured with the auditing, encryption, and network controls already in place in the customer's account.
AWS will also help sell Superblocks through its Marketplace, as it does for many partners. An AWS spokesperson said the company supports partners where it sees strong customer demand and alignment with how customers want to build. AWS itself does not yet have a vibe-coding agent aimed at business users; its Kiro product targets developers, and Quick is more like a copilot than a vibe-coding platform.
Beyond the immediate commercial boost for Superblocks — a 50-employee company that raised $60 million as of its May 2025 Series A — the deal illustrates a wider trend. Hyperscale cloud providers are encouraging enterprises to keep AI models separate from the surrounding infrastructure and to buy orchestration, security tools, and agentic apps from the cloud platforms themselves rather than from frontier AI labs.
Microsoft's Satya Nadella has recently made similar arguments, urging customers to adopt multiple models to reduce costs and avoid lock-in, and cautioning that AI labs may use customer data to study a business and compete with it later. Enterprises appear to be moving in that direction, with open models accounting for 29% of all traffic through Vercel's AI gateway last month, according to the article.
Menezes said customer attitudes have shifted dramatically in a short time, and a multi-model strategy is now a must-have for CIOs. He predicted that executives who bet on a single model provider will be fired. Cloud providers, he said, are now bringing vibe coding for business users into secure private clouds, potentially a second wave after AI coding agents for developers.
An AWS spokesperson called the category emerging with momentum and said the company supports that kind of innovation.
