Google put agents at the center of I/O, Wall Street put bonds at the center of the question, and workers are discovering that the demo has a payroll department.
A reader wrote us this week with the clean objection that oil prices, growth and the AI economy no longer line up neatly. That is exactly right, and today's news makes the sharper version hard to ignore: the AI boom is no longer a story about clever chat boxes. It is a budget, a labor plan, a power appetite and a credit-market exposure.
At Google I/O, the company moved the pitch from answer engines to acting systems. Its Gemini app now claims more than 900 million monthly users across 230 countries and more than 70 languages, up from 400 million a year ago. Google announced Gemini Spark, described as a 24/7 personal AI agent, alongside Daily Brief, Gemini 3.5 Flash and an agentic push across the app. AP reported that Alphabet's finance chief has said this year's capital expenditures may climb as high as $190 billion. That is not a feature rollout. That is industrial policy wearing a product badge.
The second half of the story is less pretty. AP's survey of recent corporate cuts found more companies using AI as part of the language around layoffs. Cisco plans to cut under 4,000 jobs, about 5% of its workforce, even as it reported record quarterly revenue and demand for AI tools and infrastructure. Block had already moved to cut more than 4,000 jobs while arguing that smaller teams using intelligence tools can do more. Companies still cite restructuring, macro pressure and efficiency. The point is not that AI explains every pink slip. The point is that AI has become a socially acceptable grammar for removing people from the operating model.
Markets see the same thing from the other end. Wednesday morning's Wall Street bounce came because bond pressure eased and oil prices gave back some gains. The 10-year Treasury yield fell to 4.60% from 4.67%, while Brent crude slipped to $106.87 a barrel, still far above its prewar level near $70. The AP market note made the buried connection explicit: high yields could limit companies' borrowing to build the AI data centers that have been supporting U.S. growth.
So the marginal economy is not simply oil, or chips, or labor, or interest rates. It is the conversion rate among them. AI demand needs chips; chips need power; power needs financing; financing gets harder when war lifts oil and bonds revolt; employers then search for payroll they can remove while still telling investors the machine will keep growing.
The important verb in Google's presentation was not answer. It was manage. An assistant that can manage your morning, your inbox, your files and your local machine is asking to become a layer of daily authority. The business model underneath asks something blunter: who pays for the infrastructure, who gets the margin, and who is quietly reclassified from colleague to cost?
That is the AI economy in its honest form. Not a cloud of intelligence floating above the old world, but a new allocator of attention, electricity, capital and blame. The demo is over. The invoice has begun.