For years, the assumption driving IBM’s skilling push — and much of the broader workforce development conversation — was that access was the core problem. Get people the tools, get them the training, and the rest would follow. Nixon-Saintil, IBM’s Vice President and Chief Impact Officer, still believes that. But IBM’s own research is complicating the picture: by 2030, 67% of executives say mindset will matter more than skillset as organizations reinvent for an AI-first economy. Knowing how to use AI, in other words, may be less important than being willing to keep learning as the tools change beneath you.

“What this increasingly requires is a continuous learning mindset,” Nixon-Saintil told Fortune. “Mindset, not just skillset, is going to matter more as organizations reinvent for an AI-first economy.”

That shift is reshaping how IBM thinks about the students it’s trying to reach. On Wednesday, the company announced the AI Builders Challenge — a global competition open to university students across participating countries, built around IBM Bob, IBM’s new AI-powered coding agent. Unlike earlier tools that focused mainly on accelerating code generation, IBM Bob is designed to support the entire software development lifecycle, integrating orchestration, execution, and governance into development workflows. Students will use it to build projects tied to real-world themes — space exploration, the future of creative industries, intelligent work systems — and submit final work through GitHub. The grand prize is $5,000, out of a total prize pool of $15,000, and an invitation to IBM TechXchange, the company’s global developer conference.

The challenge isn’t just about the prize. It’s about forcing students to do something harder than learning a tool: testing it, explaining it, and defending the decisions behind it to a potential employer.

That bar is rising faster than most universities are prepared for. A recent survey from the American Association of Colleges and Universities and Elon University found that 63% of faculty said graduates were not very or not at all prepared to use generative AI in the workplace. Nixon-Saintil was careful not to frame this as employers rejecting graduates outright — but she said she was concerned.

“What worries me most is that students who do not build these skills may face fewer entry points into work and a harder time catching up as expectations continue to change,” she said. “I would describe that as a growing AI readiness gap more than a fixed divide.”

IBM is trying to close it with access at scale. Alongside the challenge, the company announced it is expanding free access to IBM Bob to 20,000 post-secondary institutions worldwide through IBM SkillsBuild, its free learning platform. At the same time, IBM said it is tripling its entry-level U.S. hires in 2026 — and redesigning those roles around AI, using it to automate repetitive tasks and give junior employees more room to focus on higher-value work. Entry-level data analysts can now use AI to identify patterns in large datasets more quickly; junior developers can collaborate with AI to write, test, and document code more efficiently.

The message to students is implicit but clear: the entry-level job you’re walking into looks different than it did two years ago. The AI Builders Challenge is IBM’s attempt to make sure you’re ready for it.

There will be a webinar on June 24, followed by a learning fest in August.

[This story has been updated to clarify that the training of 30 million workers extends beyond AI and the grand prize is $15,000, not $10,000.]