Sent to 17 readers
Race to AGI Daily Digest - Monday, March 9, 2026
Share:
TLDR
Frontier labs cross the defense line as OpenAI and Anthropic ink a Pentagon deal and face internal revolt.
OpenAI’s robotics chief resigns over AI surveillance concerns, signaling real internal limits on military work.
Nvidia reports 70% of healthcare providers now use AI, showing how fast practical deployment is moving.
Block workers and Oracle rumors show the AI ‘efficiency’ story is colliding with job security on the ground.
AI chip leaders from ASML to AMD sold off today, even as long-term AI infrastructure bets like Nvidia–Groq stay in focus.
The Full Story
Last week the big story was Nvidia’s $20B bet on Groq, a move toward highly specialized AI hardware and talent that confirmed what we’ve been saying for a while: the race to AGI is becoming a race to own the stack. If you want to keep tracking that thread, the Nvidia–Groq deal is now its own evolving storyline. Follow that narrative here ->
Now the spotlight swings from chips to power. Over the weekend, OpenAI and Anthropic stepped into a formal partnership with the U.S. Department of Defense, sparking resignations from OpenAI’s robotics lead and a broader "QuitGPT" revolt inside the labs. Catch the Pentagon deal coverage -> This is the moment a lot of people were waiting for (or dreading): when frontier model companies stop just talking about safety and start fielding systems for intelligence, targeting, and surveillance.
Here’s the thing: this is not happening in a vacuum. Workers at Block are already pushing back on the AI-as-efficiency story after 4,000 layoffs tied to automation narratives. Read the worker pushback story -> On Hacker News, people are gaming out Oracle-sized job cuts to fund new AI data centers. The social mood is shifting from "AI will make us more productive" to "productive for whom?"
At the same time, Nvidia says 70% of healthcare providers are already using AI. That’s a real deployment curve, not just slideware. Dive into the healthcare survey -> And if you scroll Reddit, you see the other side of that stat: engineers still wrestling with how to actually ship finetuned models in production. See one of those deployment threads ->
Meanwhile, AI hardware darlings had a rough day: ASML, Intel, ARM, TSM, and AMD all sold off hard after weeks of euphoria. That doesn’t undo last week’s optimism, but it does remind us that “infinite demand for compute” still passes through jittery public markets. You can explore the companies behind those tickers in our live directory. Browse all AI-linked companies ->
So for this week, we’re watching three threads: defense and surveillance becoming a first-class AI use case, workers testing how far "AI efficiency" can go before people rebel, and the widening gap between lofty AGI timelines and the messy reality of deploying these systems at scale.
Get This Delivered Daily
Join thousands of AI professionals who start their day with Race to AGI.