Meta’s Massive AI Infrastructure Push (and Reality Labs Cuts) 🏗️
Zuckerberg is trading the Metaverse for Mega-Compute. Meta just announced a colossal new initiative called “Meta Compute,” pledging to invest $600 billion in U.S. AI infrastructure by 2028. This “top-level” push aims to add tens of gigawatts of capacity to ensure Meta never runs out of compute for its future AI models.
Here is the strategic shift:
- The Investment: The $600 billion plan involves building massive new data centers and locking in 20-year nuclear power agreements to feed them.
- The Leadership: Infrastructure chief Santosh Janardhan will co-lead the effort with Daniel Gross, a high-profile hire from AI safety startup SSI. Former Trump official Dina Powell McCormick has been appointed president to handle government deals.
- The Trade-off: To pay for this massive AI bet, Meta is cutting costs elsewhere. Reports confirm layoffs of approximately 1,500 employees (10% of the workforce) from the Reality Labs division, signaling a definitive pivot away from pure VR/Metaverse projects toward AI and wearables.
Why it matters: The “Metaverse” era is officially secondary to the “Intelligence” era. By reallocating billions of dollars and thousands of roles, Zuckerberg is betting the company’s entire future on winning the race to AGI. This infrastructure blitz ensures that if superintelligence is possible, Meta will have the hardware to run it.
UrviumAI Take: This is a classic “Guns vs. Butter” economic decision. You can watch the energy sector. Meta’s 20-year nuclear deals are a signal that “Sovereign AI” (AI powered by its own dedicated energy grid) is the new standard for Big Tech. If you invest, look at utilities and nuclear micro-reactor startups that are partnering with these hyperscalers.
UK Startup ‘Eden’ Learns from Evolution to Cure Diseases 🧬
Nature has been running R&D for billions of years, and now AI is reading the notes. UK startup Basecamp Research, in collaboration with Nvidia, has launched Eden, a revolutionary new family of AI models that learned from the DNA of over 1 million species to design new medicines.
Here is what Eden has already achieved in the lab:
- Learning from Life: Instead of just human data, Eden trained on the evolutionary history of organisms across 28 countries, learning how nature solves biological problems.
- Better than CRISPR? The AI designed a new programmable gene insertion tool that can insert therapeutic DNA without cutting the genome strand—a potentially safer and more precise alternative to CRISPR.
- Killing Superbugs: In tests, Eden created new antibiotic candidates that proved 97% effective against dangerous, drug-resistant “superbugs” that current medicines can’t kill.
- High Success Rate: For genetic diseases like muscular dystrophy, over 63% of the treatments designed by the AI worked functionally in lab tests.
Why it matters: Most AI drug discovery fails because biology is messy. Eden succeeds because it understands the “language” of DNA evolution. By tapping into the biodiversity of the entire planet, this model is finding cures that humans and previous AIs couldn’t see.
UrviumAI Take: The 97% success rate against superbugs is the headline statistic. Watch the regulatory approval process for these AI-designed enzymes. If the FDA creates a fast-track lane for “AI-verified” biological tools (similar to what OpenAI is pushing for), we could see these treatments in hospitals within 3-5 years instead of 10.
McKinsey CEO: We Have 25,000 AI Agents “Employees” 🤖
Your next consultant might be an AI agent. McKinsey & Company CEO Bob Sternfels dropped a stunning statistic recently: the prestigious consulting firm now has a total workforce of roughly 60,000 “employees”, but 25,000 of them are AI agents.
Here is McKinsey’s radical new “25-squared” strategy:
- The Split: The firm employs roughly 40,000 humans and 25,000 AI agents. Sternfels expects these numbers to match (1:1 ratio) by the end of the year.
- The Strategy: Under the “25-squared” model, McKinsey is growing client-facing human roles by 25% while simultaneously shrinking non-client-facing (backend) roles by 25%.
- The Productivity: The AI agents saved the firm 1.5 million hours last year on search and synthesis tasks. In just six months, they generated 2.5 million charts, freeing up human consultants to focus on strategy.
- The Future Hire: Sternfels warned that hiring criteria are changing. The firm is looking less at university pedigrees and more at “demonstrable skills,” like a GitHub portfolio, as AI handles the rote academic work.
Why it matters: McKinsey is the bellwether for the corporate world. If the world’s most expensive strategy firm is actively replacing backend staff with 25,000 AI agents to boost efficiency, every Fortune 500 company will likely follow suit within 18 months. The definition of “headcount” has officially changed forever.
UrviumAI Take: This is the quantification of the “White Collar Shift.” If you are in a “backend” or support role, this is your signal to pivot. The skills McKinsey says AI can’t replicate are “setting aspirations, human judgment, and genuine creativity.” Focus your career development on those three pillars immediately.
Last AI News: Anthropic Blocks xAI, Microsoft’s AI Adoption Gap, and Claude for Healthcare
Jigar Chaudhary is the Editor-in-Chief at UrviumAI, where he oversees coverage of artificial intelligence news, tools, and in-depth studies. With over 5 years of experience analyzing AI and robotics, he focuses on maintaining high editorial standards, accurate reporting, and clear explanations to help readers understand how AI is shaping the future.
