Cursor in Talks to Raise $2B+ at $50B Valuation
AI-powered code editor Cursor is reportedly in talks to raise over $2 billion at a $50 billion valuation, according to TechCrunch. The round is expected to be led by returning backers Andreessen Horowitz (a16z) and Thrive Capital, signaling continued strong confidence from institutional investors. TechCrunch cites surging enterprise growth as the primary driver behind the elevated valuation. If completed, the round would mark a significant milestone for the developer tooling space and underscore the accelerating demand for AI-assisted software development in enterprise environments.
OpenAI Shuts Down Sora, Folds Science Team in Strategic Pivot
OpenAI has shut down its Sora video generation product and folded its science team as part of a broader organizational shift, TechCrunch reports. The changes come alongside the departures of Chief Product Officer Kevin Weil and researcher Bill Peebles. According to TechCrunch, the moves reflect a deliberate pivot away from what the company has internally described as "side quests" — consumer-facing moonshot projects — and toward a sharper focus on enterprise AI. The restructuring suggests OpenAI is consolidating its product and research efforts around its core business priorities.
World Expands Human Verification Partnerships, Starting with Tinder
World, the anonymous human verification project centered around its iris-scanning Orb device, is moving to broaden its reach through a series of new partnerships, TechCrunch reports. The first announced partnership is with Tinder, positioning World's verification technology as a tool for confirming that users are human on consumer platforms. According to TechCrunch, World has already attracted significant interest alongside scrutiny for its biometric verification approach. The Tinder partnership represents the company's first major step in what appears to be a broader platform expansion strategy.
'Tokenmaxxing' Linked to Reduced Developer Productivity
A new analysis reported by TechCrunch examines the practice of "tokenmaxxing" — a pattern in which developers prompt AI coding tools to generate large volumes of code in order to maximize output — and finds that the approach may be counterproductive. According to TechCrunch, while tokenmaxxing does result in more code being produced, that code tends to be more expensive to run and requires substantially more rewriting downstream. The report suggests that optimizing for token volume, rather than code quality or correctness, may be creating hidden costs for development teams that are not immediately apparent in raw output metrics.
These stories collectively reflect a period of consolidation and recalibration across the AI industry, from organizational restructuring at major labs to evolving best practices in AI-assisted development.