Sam Altman warns AI superintelligence will hit harder than expected

Sam Altman is speeding up the timeline.

Speaking at The New York Times’ DealBook Summit in New York, the OpenAI CEO dropped a bold forecast: artificial general intelligence (AGI) could arrive as early as 2025. In Altman’s view, that means systems capable of taking on deeply complex tasks just like a human, orchestrating multiple tools to finish the job.

While early ripples might seem modest, Altman warned that the rise of AGI—and the superintelligence that follows—will ultimately be far more disruptive than people expect, bringing inevitable job displacement in its wake.

Transistors, 300 million users, and a massive hiring spree

Altman compared the current AI wave to the invention of the transistor. Eventually, advanced reasoning engines will become cheap, ubiquitous commodities powering everything.

OpenAI’s own growth numbers underline that transition. Altman shared that ChatGPT now tops 300 million weekly active users. The company has aggressively scaled up, hiring over 1,000 employees since the start of the year to reach a headcount of around 1,700. Following a fresh $6.6 billion funding round in October, the startup sits on a $157 billion valuation.

Altman reiterated that OpenAI’s board is actively working on restructuring the organization into a for-profit entity. Running cutting-edge frontier models requires unthinkable capital that the original non-profit research lab was never built to support. As for OpenAI's complex marriage with Microsoft, Altman acknowledged real tensions and misalignments, but insisted the two remain strategically aligned.

When asked about his personal favorite release, Altman singled out OpenAI’s newly launched web search tool, claiming it completely rewrote how he browses the internet.

Court dates and broken friendships

It wasn't all smooth talking. Host Andrew Ross Sorkin joked about seeing Altman in court over copyright disputes, nodding directly to The New York Times' pending lawsuit. Altman argued that debates over fair use are fundamentally missing the mark, insisting that the tech sector needs entirely new economic models to pay creators.

Pushback is mounting on multiple fronts. OpenAI recently pulled public access to its video generator Sora after artist testers revolted against unpaid R&D and PR exploitation.

Then there is Elon Musk’s aggressive legal offensive. Musk has sued OpenAI and Altman for betraying the original non-profit mission while spinning up xAI to compete head-on. Altman admitted the lawsuit made him tremendously sad, noting he grew up viewing Musk as a personal hero. While he expects xAI to be a formidable challenger, he dismissed concerns that Musk would wield political ties with Donald Trump against competitors, calling that move deeply un-American.

Why it matters

Altman's comments signal a decisive narrative pivot: the frontier is moving away from conversational chatbots and straight into fully agentic systems.

For knowledge workers and builders, this shifts the immediate playbook. If multi-tool execution lands next year, prompt-and-paste workflows will look instantly dated. Teams should look beyond isolated chat interfaces and focus on how autonomous agents chain real-world software together.

Strategically, OpenAI is racing against a ticking clock. If foundational reasoning models become commoditized utilities, proprietary smarts won't be enough to hold a $157 billion valuation. That explains why OpenAI is aggressively pushing into vertical consumer search, securing enterprise distribution, and converting into a conventional corporate structure.

Yet the structural pushback is growing louder. The Sora artist protest proves that creators won't accept token access in exchange for training the machines that replace them. Moving fast and breaking things works in code, but navigating copyright courts and labor fallout will be OpenAI's true stress test.