Big Tech companies are now racing to see who can build the best AI coworker

For the past few years, the AI boom has been defined by experimentation: launch everything, chase each use case, and figure out the business later.

That phase is winding down.

Over the span of a few days, Alibaba, Microsoft, and OpenAI each made moves that reflect the same shift: AI is no longer an experiment. It’s the core business.

Their focus is increasingly on tools that can do real work, not just answer questions. In other words, the race is shifting from who has the smartest model to who can build the most useful coworker.

Microsoft is combining its consumer and commercial Copilot efforts into a single system, aiming to reduce “manual coordination,” or human intervention, and make its AI more useful for actual work. Alibaba is consolidating its AI efforts into a new unit led by CEO Eddie Wu, who framed the shift around capturing a “historic opportunity” as AI agents take on a growing share of business tasks. And OpenAI is pulling back from its sprawl, with leadership warning it can’t afford “side quests” as it doubles down on coding and enterprise productivity.

All of these companies are spending heavily on AI infrastructure — and now they need to show that it pays off. That means focusing on the use cases that generate real revenue, especially in enterprise environments.

That includes software that can handle multistep tasks: writing code, managing workflows, responding to customers, and completing transactions. Alibaba is building assistants designed to help users shop and complete tasks across its platforms. Microsoft is pushing deeper into “agentic” features inside Office to help complete more complicated tasks. And OpenAI is refocusing on coding and business users as it tries to keep up with Anthropic, whose tools have emerged as early leaders in enterprise use.

With each release — spanning legal and finance, coding, and office work — Anthropic has put pressure on traditional software vendors, showing how AI can take over tasks that previously required dedicated tools or teams. It’s part of the reason why the center of gravity is shifting toward enterprise. Consumers may drive attention, but businesses pay the bills — and they are far more willing to spend on tools that save time or reduce labor. What these tech companies are really hoping to sell is not software, but labor in software form.

Meta, which lacks a comparable enterprise software business, is taking a different approach. To justify its AI spending, the company has focused on how the technology can boost its existing revenue streams. It has also been cutting costs, including through layoffs, while pushing remaining employees to increase output using AI tools. Over the weekend, Reuters reported that Meta is planning to cut roughly 20% of its workforce to reach those ends.

The question for these companies has shifted from what AI can do to whether it can justify its cost.