This is an excerpt of Sources by Alex Heath, a newsletter about AI and the tech industry, syndicated just for The Verge subscribers once a week.
To understand how Microsoft sees its role in the AI race, look at what was announced at the company’s GitHub Universe developer conference on Tuesday in San Francisco.
The event marked a turning point for one of Microsoft’s most important but often under-analyzed assets. GitHub, which is used by more than 180 million developers, wants to be the central platform for AI coding agents. Its new Agent HQ interface lets outside coding assistants, including OpenAI’s Codex and Anthropic’s Claude Code, plug into the GitHub ecosystem.
The timing is apt. If GitHub doesn’t move fast enough, Microsoft risks being left behind by a wave of agentic coding tools that are quickly redefining what software development looks like. Tools like Cursor, which coincidentally announced a major update this week, are already orchestrating entire workflows for developers rather than simply autocompleting code. While agents aren’t yet making a significant impact in other verticals, in software development, the boundary between helper and co-worker is quickly disappearing.
“We need to have a very arms-wide-open ecosystem.”
Jared Palmer, GitHub’s new senior vice president, joined Microsoft from Vercel just 10 days before we spoke ahead of this week’s Universe conference. “GitHub is in this interesting transition phase,” he said. “It needs to evolve into the platform for not just people, but also agents.”
Palmer said the goal is to keep GitHub “the home where development happens,” no matter which AI tools developers choose. “We need to have a very arms-wide-open ecosystem,” he told me. In addition to OpenAI and Anthropic, Google, Cognition, and xAI have also committed to bringing their coding agents to Agent HQ in the coming months.
Thanks to GitHub’s longtime, essential role in storing code, it has a good shot at becoming the connective layer between agents and codebases. That would mean Microsoft stays at the center of the developer universe. If it doesn’t, developers could move their work and their data elsewhere over time.
Jay Parikh oversees GitHub as Microsoft’s executive vice president of CoreAI. The longtime Facebook CTO and Atlassian board member joined Microsoft nearly a year ago and now leads a group of roughly 10,000 people. His mandate is to rethink the entire Microsoft stack for building AI applications: infrastructure, security, and the tools developers use every day.
“We need to move faster,” Parikh told me in an interview this week. “We need to build better tools, different tools. We need to have that developer ecosystem, that choice, and all the things that support this new way of building software.”
“Humans are going to start spending more time in the specification and creative process and delegate the actual calories they spend to GPUs.”
He sees GitHub’s role as essential to that plan, which explains why Microsoft folded GitHub into his CoreAI group in August. “You really want GitHub to be that place, that toolchain, that community for developers to really unlock this innovation,” he said. “Humans are going to start spending more time in the specification and creative process and delegate the actual calories they spend to GPUs, to coding agents that generate the code.”
As GitHub COO Kyle Daigle put it to me, “A developer shouldn’t have to build up memory and context in every single tool that they’re using in this AI age. You should be able to connect to GitHub and have that context to determine how you want to work, even if you’re using that tool off GitHub’s platform.”
That bet ties directly to Microsoft’s long-running partnership with OpenAI. On the same day as GitHub Universe, CEO Satya Nadella appeared on TBPN to discuss Microsoft’s updated partnership with OpenAI. He said that “GitHub is where that billion to ten billion happened” — a reference to how the early traction of GitHub Copilot, powered by OpenAI’s Codex, convinced him to dramatically expand Microsoft’s initial $1 billion investment in OpenAI.
GitHub has always served two critical roles: it’s a massive repository of open-source code and one of the largest collaboration hubs ever built. Engineers use it not only to store and manage their code, but to find jobs and build community. Microsoft may have sparked the AI coding boom with the original GitHub Copilot, but now it has to prove it can keep up with where the industry is headed.