The race to build AI that writes code has become one of tech’s most ferocious battlegrounds, and the fight is turning personal. Startups, hyperscalers and the AI labs themselves are now chasing the same scarce engineers, customers and capital, often circling the very same deals. The result is a market where who funds, hires and hosts a company matters as much as whose model writes the cleanest code.
Nowhere is the drama sharper than at Cognition, maker of the AI engineer Devin. Early on, Devin struggled to live up to its hype. Today, Cognition is one of the fastest-growing AI apps, but previously unreported figures suggest that blistering growth is burning through cash at an alarming rate. Behind it sits Zach Frankel, a low-profile investor and board member whom founders reportedly call a “secret cofounder,” credited with helping engineer Cognition’s Windsurf deal.
The giants are muscling in too. On the day SpaceX closed its roughly $60 billion all-stock acquisition of Cursor-maker Anysphere, Elon Musk addressed staff for the first time, folding the startup into his xAI-powered empire. The move vertically integrates Musk’s AI, from chips and models to developer tools, and underscores how the biggest players are simply buying their way to the front.
Even loyal customers are recalibrating. Meta and Microsoft, two of Anthropic’s largest buyers, are reportedly working to reduce how heavily their engineers lean on Claude, a shift that lands awkwardly as Anthropic prepares to go public. Dependence, it turns out, cuts both ways.
The clashes are increasingly bitter. A sales leader’s jump from Factory to rival Cognition triggered duelling public accounts from both CEOs and reportedly split investors inside a single firm. Meanwhile, the ecosystem’s plumbing is consolidating: Vercel has become the default host for code written by Claude Code, Cursor and Codex, while year-old Proximal has emerged from stealth selling automated training data.
So is this all-out war a leap from frying pan to fire? For many founders, yes. Escaping the scramble for talent and funding only lands them in fiercer fights over acquisitions, defections and runaway burn. The heat that once came from building a great product now comes from everywhere at once.
Ultimately, the AI coding boom is proving that surviving the first fire merely earns a ticket to a hotter one. Those who win will be the companies that master not just code, but the brutal economics and alliances surrounding it.
While global AI giants battle for dominance, Delaware-based startup FaceOff Technologies Inc. is quietly developing Detective GPT, an AI-powered investigative intelligence platform. Envisioned to combine multimodal AI, deepfake detection, behavioural intelligence and agentic reasoning, the platform aims to help investigators identify digital deception, analyse suspicious activities and accelerate evidence-based investigations. Its ambition is to transform complex digital signals into actionable intelligence, strengthening trust, cybersecurity and next-generation investigative capabilities.
