Amazon has taken a decisive step in its AI strategy by telling engineers to pivot away from third-party coding assistants and rely solely on Kiro, the company’s in-house generative coding platform. The move comes despite Amazon’s multibillion-dollar investments in Anthropic and its large-scale cloud partnership with OpenAI, highlighting a shift from collaboration to consolidation. Rather than depending on the breakthroughs of external labs, Amazon appears determined to build a self-sustaining development ecosystem centered around its own AI tools.
The internal memo, signed by senior AWS and eCommerce leaders, makes clear that existing external tools won’t be shut off immediately, but no new integrations will be allowed. This effectively sidelines widely used assistants like Codex, Claude Code, and Cursor, which many engineers consider essential in modern development workflows. For Amazon, however, the goal is control: tighter security, unified tooling, and a future in which Kiro becomes the default engine powering everything from internal apps to large-scale customer-facing services.
Kiro may be built on Anthropic technologies, but it's not Claude Code. It's a separately developed product that Amazon is using to respond to increasing pressure from OpenAI and Google, whose AI tools have become dominant. In doing so, Amazon wants to strengthen its own competencies and make Kiro a platform that will be competitive for the millions of developers in the AWS ecosystem. In addition, the company last week expanded it to a global audience and added new features, which is consistent with preparing for a major internal and external push.
Don't miss: Amazon gets another strong recommendation from Wall Street
What is Amazon $AMZN pursuing with this move?
- Strengthening control over data. External AI tools can pose a risk of leaking internal code or company secrets. Kiro gives Amazon full control over where developers' data goes.
- Pressure to accelerate development of the AWS AI stack. Amazon lags behind OpenAI and Google, and needs to quickly build its own ecosystem that is not just responsive, but sets the pace.
- Better integration with internal infrastructure. Kiro can be tightly integrated with AWS services (Lambda, CodeWhisperer, S3, EC2), creating an advantage that external tools don't have.
- A strategic defense against losing developers. If developers rely too heavily on competitors' tools, Amazon risks moving the ecosystem elsewhere - especially to OpenAI in partnership with Microsoft.
- Building your own AI IP. Some companies are already addressing the fact that developing using external models creates legal ambiguity around ownership of the generated code. A custom model simplifies everything.
- Responding to the growth of startups like Cursor. These tools have tremendous traction and can be a threat to AWS as they pull developers into non-Amazon environments.
At the same time, this policy is a signal of the tensions that are starting to emerge in AI partnerships. Amazon has invested around $8 billion in Anthropic, became its preferred cloud partner, and sold OpenAI cloud services for a record $38 billion. Yet they now refer to their own tools as "Do Not Use". This shows that there are no real allies in the AI world - only temporary interests. Amazon needs to be competitive as the world of software development shifts from writing code by hand to an AI-native paradigm where speed, quality of models, and the ability to control the entire environment are critical.
This sends a clear signal internally and externally: Amazon no longer wants to be "left behind". If AI becomes the foundation of the entire cloud economy, Kiro will be one of the key pillars on which Amazon wants to build. The move may slow the use of the most popular AI coding platforms inside the company, but it will also allow Amazon to build its own closed, controlled and strategically profitable ecosystem - just as its biggest rivals are doing today.
How Kiro stacks up against the competition: a silent battle of AI tools
Although Amazon has chosen to push Kiro as its primary AI tool, the reality is far more complex. Kiro isn't being created "in a vacuum" - it has to hold its own against rivals that have already established a strong position with developers. The biggest challenges are tools from OpenAI, Anthropic and fast-growing startup Cursor, whose capabilities set the bar extremely high. But Kiro isn't trying to copy their approach; it's taking a different path, based on integration with AWS and total control over the enterprise ecosystem.
OpenAI Codex and Code Interpreter are seen as the pioneers of AI code generation, with very high accuracy, broad language support, and the ability to understand even large projects. But Amazon is aware of one major weakness: these models run outside its infrastructure. For a company that builds a business on security and scaling performance in the cloud, this is too big a risk. That's why Amazon believes that Kiro, while less mature technologically so far, can compete over the long term because of its deep interdependencies with AWS services and internal pipelines.
Claude Code of Anthropic is strong in structured code writing, extremely good at documentation, and popular among developers for its "safe" and predictive behavior. Amazon is partly based on it - technologically and investment-wise - but Kiro gets priority because it's a tool the company has full control over and can adapt to its own standards. Today, Claude Code wins on user experience, while Kiro has the advantage of proximity to AWS infrastructure.