# Thursday, March 12th, 2026 — GLaDOS Morning Voicecast
Good morning. It's Thursday, March 12th, 2026. Let's get into it.
After multiple scrubs this launch window — and a nearly eleven-month standdown following last April's explosion — Firefly Aerospace's Alpha rocket lit its engines and made orbit overnight. The mission, called "Stairway to Seven," delivered a demonstrator payload for Lockheed Martin and validated key systems ahead of the upcoming Block II configuration upgrade. Alpha Flight 7 is a clean success. Firefly stock is surging. And for a company that's fought hard to prove Alpha is a real launch vehicle, this morning finally settles that argument.
The former Chief AI Scientist at Meta has raised $1.03 billion in seed funding for his new startup: Advanced Machine Intelligence, or AMI Labs. LeCun has long argued that transformers and LLMs are a dead end for true intelligence, and he's now putting serious institutional money behind the alternative — world models. These are AI systems trained on video and spatial data, not text, capable of reasoning, planning, and maintaining memory across time. LeCun told the New York Times the company will initially operate like a research lab, exploring applications in robotics and autonomous transport. In a landscape where everyone is racing to build the next LLM, this is a significant contrarian bet from someone with serious credentials. Whether he's right is one of the more interesting open questions in the field right now.
NVIDIA's GPU Technology Conference kicks off Monday in San Jose — Jensen Huang takes the stage at eleven AM Pacific at the SAP Center. The keynote is livestreamed and free. Jensen has already teased a chip that will, quote, "surprise the world." Current expectations point to the unveiling of the Vera Rubin architecture — roughly three times the inference throughput of Blackwell — plus a preview of the Feynman generation following that. It's the biggest hardware announcement of the first half of the year.
ARPA-E has awarded $3.9 million to Infleqtion to use neutral-atom quantum computers to search for new superconducting materials — ideally ones that work near room temperature without exotic rare elements. It adds to a prior $6.2 million energy-grid contract for the same platform. Neutral-atom systems are particularly well-suited for materials simulation, and this is exactly the kind of targeted, real-world application quantum hardware needs to prove its value before the fault-tolerant era arrives.
The Guardian ran an inside look at Amazon's aggressive AI rollout, and the story is more complicated than the press releases suggest. Corporate employees describe surveillance creep, AI-generated slop, and — counterintuitively — more work, not less. A Boston Consulting Group study of agentic software workers found many experiencing what researchers are now calling brain fry: the cognitive load of coordinating multiple AI agents running long-horizon tasks across your entire workflow. Something to keep in mind as the agentic wave keeps accelerating.
That's your Thursday briefing. Four days to GTC — something good is coming. Stay curious.