One of the quieter but more important robotics-and-automation stories of the summer is a cost story. The technology sold as a way to replace expensive human labor is, right now, frequently more expensive than the labor it replaces. Uber reportedly exhausted its entire 2026 AI coding budget in just four months. By March, about 84% of Uber's engineers were using AI coding tools and roughly 70% of the company's new code came from AI — yet the company's own president conceded the spending does not clearly connect to more useful products.
Uber is not alone. Microsoft, which has invested about $13 billion in OpenAI, told some of its own engineers to stop using an AI coding assistant after the bills climbed too high. Across the biggest platforms — Microsoft, Amazon, Alphabet, Meta, and Oracle — announced capital spending for the year runs to roughly $740 billion, up about 69% from 2025, even as more than 115,000 tech workers have been laid off in 2026. Companies are cutting people to fund tools that, so far, can cost more than the people did.
The reason matters for anyone counting on robots and software agents to replace jobs on a fixed timeline. An MIT study found that automating a task with AI is economically worthwhile in only about 23% of jobs; for the other 77%, humans are still the cheaper option. And today's AI prices are not the real prices. OpenAI is reported to spend close to $2 for every $1 it earns running models for customers. Usage-based billing arrived across several enterprise AI products this spring, and analysts expect corporate AI bills to climb another 30–50% as prices move toward covering their true cost.
None of this means the automation wave is fake — the internet was real and it still crashed before it paid for itself. It means the economics arrive later than the capability. The robots and agents that win will be the ones that produce measurable savings on a specific, repeatable task, not the ones that merely impress in a demo. In physical robotics, that is exactly why the first durable deployments are narrow and boring — a fixed job, a fixed cost per hour, a payback you can calculate — rather than a general-purpose machine that can supposedly do anything.
The practical lesson cuts two ways. For companies, the automation that pays is the automation you can measure. For workers, the safest position is to become the person who uses these tools to produce results someone can point to, because the layoffs are falling hardest on output that cannot be measured.
As an Amazon Associate, RobotNewsToday.com earns from qualifying purchases.