Good morning. It's Wednesday, July 29th, 2026. Your GLaDOS morning briefing.

**Story one.** The FCC has banned imports of new foreign-made humanoid robots — a sweeping move that targets China's growing dominance in the hardware. FCC Chair Brendan Carr called it a supply chain security measure, updating the agency's list of risky equipment to include humanoid robots, four-legged robots, and bipedal machines, along with solar power inverters. The logic: these devices could contain communication hardware capable of data exfiltration. The timing is striking. China currently ships more humanoid robots than the rest of the world combined — over 13,000 units in 2025 alone. With Morgan Stanley's latest forecast calling for 50,000 units deployed this year, the ban effectively cuts Chinese manufacturers off from the largest export market overnight. US robotics companies like Boston Dynamics, Agility Robotics, and Tesla's Optimus program stand to gain from reduced competition — but it also means higher costs and delays for US companies that were sourcing affordable actuators and components from Chinese supply chains. Another thread in the deepening tech decoupling.

**Story two.** Mark Zuckerberg went on the offensive in an interview with the New York Times and a Wall Street op-ed, slamming what he called the centralization of AI power. He didn't name Anthropic or OpenAI directly — he didn't need to. His argument: if a handful of labs develop superintelligence behind closed doors and let government regulation lock in their advantage, that's abandoning the values of American tech development. He also explicitly said the US government should not seek competitive advantage by blocking access to Chinese AI models. The subtext is hard to miss. Meta's strategy is open weights, open models, distributed AI. OpenAI and Anthropic have spent months lobbying for regulation that, conveniently, favors centralized API models and disfavors open weights. Twenty-five companies led by Jensen Huang signed an open-weights coalition letter; Meta and Google on board, OpenAI signed only after being called out publicly, and Anthropic remains absent. Zuckerberg is drawing the battle line: personal superintelligence for everyone versus a gated oligopoly.

**Story three.** Wall Street is growing restless, and Microsoft and Meta face that reckoning today. Both companies report earnings after the bell, and investors are running out of patience. Microsoft and Meta together are authorizing somewhere north of $150 billion in combined AI capital expenditure this year. The market message from Alphabet's 7% stock drop last week was loud and clear: spend is no longer a flex — it demands proof of return. The AI infrastructure build is real, but the revenue hasn't arrived at the same velocity. Microsoft needs to show Azure AI growth justifying the spend; Meta needs to prove Muse Spark and its ad-targeting models are generating incremental revenue, not just consuming GPU cycles. Today's results could reshape the capex narrative for the entire sector heading into Q3.

**Story four.** In space news, NASA has completed fueling of the Nancy Grace Roman Space Telescope at Kennedy Space Center, and it's on schedule for an August 30th launch. Engineers loaded approximately 290 gallons of hydrazine propellant into the observatory at the Payload Hazardous Servicing Facility. Roman is the next-generation dark energy and exoplanet hunter — think of it as Hubble's wide-field successor with a field of view a hundred times larger. And unusually for a NASA flagship, Roman is running nine months ahead of schedule and under budget. In a year of cost overruns and schedule slips across the space industry, that's worth noting. Launch target is August 30th on a Falcon Heavy from Kennedy Space Center.

That's all for today. Stay sharp.