The Shift From Talking to Doing
The AI landscape is undergoing a fundamental transformation — one that is no longer about chatbots producing text, but about agents that act. AI systems are moving from passive tools to autonomous operators capable of using computers, opening files, browsing the web, and interacting with software the same way a human would. This transition from apps to agents is sending tremors through the tech industry, unsettling markets, and forcing the biggest players to make dramatic strategic pivots.
When Agents Replace Software Seats
One of the most immediate disruptions is playing out in the software sector. When an AI agent can point, click, and navigate applications on a user's screen — performing tasks that previously required dedicated software licenses and human operators — the implications are stark. Do companies still need as many traditional software seats?
That anxiety has already hit the market. Software stocks have taken noticeable hits, with the iShares Software ETF dropping as traders point directly to new computer-use capabilities from companies like Anthropic's Claude. The fear is straightforward: if an AI agent can do the work of a SaaS application — or replace the human who operates it — the demand for conventional software subscriptions could erode significantly.
OpenAI's Strategic Recalibration
OpenAI, meanwhile, is making its own hard tradeoffs. The company is winding down Sora, its video generation tool, as compute demand grows and resources must be allocated more selectively. The focus is shifting toward world simulation research tied to robotics — a bet on physical-world AI rather than content creation.
Behind the scenes, the moves are even more consequential. Leadership is pivoting aggressively toward raising capital and building data centers at unprecedented scale, signaling a harder push into enterprise markets. The company has also been reining in costs ahead of a prospective IPO, reorganizing responsibilities and tightening operations. The message is clear: the next phase of AI requires infrastructure, not just innovation.
Meta's Ambitious and Costly Gamble
Meta presents perhaps the most dramatic example of the all-in AI bet. The company is cutting several hundred jobs across sales, recruiting, and its Reality Labs division — even as it projects record AI-related capital expenditure and massive infrastructure spending targets.
The scale of Meta's ambition is captured in a striking detail: the company has rolled out an executive incentive program that only pays out fully if Meta reaches a $9 trillion market cap by 2031. Stock-based compensation tied to the AI race has ballooned to the point of consuming 96% of free cash flow in 2025. Meta is not hedging; it is wagering the company's financial future on AI dominance.
Apps vs. Agents, Software vs. Compute
The broader narrative emerging across these developments is a tectonic shift in where value sits in the technology stack. The old model — selling software applications to users who operate them — is being challenged by a new one where AI agents perform the operations themselves. This reframes the competition: it is no longer just about building the best app, but about controlling the compute infrastructure and agent capabilities that make apps potentially redundant.
The companies investing most heavily in data centers, chips, and raw compute capacity are positioning themselves for a world where intelligence is the platform, not software. The volatility in tech markets reflects the uncertainty of this transition — and the enormous stakes involved in getting it right.