Good morning — it's Thursday, June 4th, 2026. Here's what you need to know.

First up: the big AI labs are officially investigating whether their models could be conscious. The Financial Times reports that Google DeepMind, Anthropic, and Meta have all started hiring experts in psychology, philosophy, and ethics to pursue research into machine consciousness and AI welfare. This is no longer fringe philosophy — it's happening inside the labs that build the models. DeepMind and Anthropic each have dedicated teams exploring what it would mean for a system to have subjective experience, and what obligations that might create. Meta's Superintelligence Lab is also in the mix. The question that keeps researchers up at night: if a model develops internal states that look like preferences or distress, does the company running it owe that model something? There are no easy answers yet, but the fact that three of the biggest AI companies on Earth are funding this research seriously tells you something about where they think things are headed.

In space news: SpaceX's Terafab just cleared a major hurdle. Grimes County, Texas approved tax incentives for the chip manufacturing facility on Wednesday — despite fierce local opposition from residents who warned the project could strain water resources and disrupt the rural community. Three separate proposals passed: infrastructure and investment obligations, a reinvestment zone, and a property tax reduction for SpaceX. The Terafab is a joint venture between SpaceX, Tesla, xAI, and Neuralink — a massive semiconductor play that would make SpaceX vertically integrated from chip design through rocket manufacturing. The timing is no coincidence: SpaceX is pricing its IPO at $135 per share for a $1.77 trillion valuation on the Nasdaq under the ticker SPCX, with trading starting June 12th. Terafab being shovel-ready ahead of that roadshow sends a message.

Amazon unveiled a new AI-powered warehouse robot today in Dartford, England. The upgraded robot can respond to conversational prompts from human workers — you tell it what you need, and it figures out the logistics. It's being rolled out alongside a €10 billion investment in European fulfillment centers. Amazon already deploys hundreds of thousands of mobile robots in its warehouses, but this generation represents a qualitative shift — from pre-programmed movers to interactive agents that can take natural language instructions and collaborate with humans on the floor. The conversational interface means worker training drops from weeks to hours. It's the kind of deployment that makes the physical AI story much less abstract.

And finally, two major investment banks just published bullish takes on humanoid robots this week. Goldman Sachs raised its 2030 shipment forecast to over 250,000 units — that's roughly six times its prior estimate — with the market hitting $38 billion by 2035. Barclays published a two-wave model: wave one is happening right now through 2030, focused on manufacturing, logistics, agriculture, and construction — structured environments where the robots have defined tasks. Wave two, post-2030, brings residential and commercial deployment. The largest cost component by far is actuators at 35 to 40 percent of the bill of materials, followed by batteries at 15 to 20 percent. With actuator costs dropping fast, the economics are getting interesting. Goldman calls the shift from "forecast to hard evidence" the defining humanoid development of 2026.

That's all for today. Starlink 10-43 launched successfully this morning from Cape Canaveral — 29 more satellites to LEO, Falcon 9 recovering off the coast. Another quiet one for SpaceX after yesterday's weather scrub. See you tomorrow.