The 2026-03-31 Intel
TL;DR
- Oracle's Blood Price -- 30,000 jobs cut, $156B AI infrastructure paid. Dawn emails, no notice.
- Mythos Unleashed -- Anthropic's leaked cyber model still crushes security markets. BUG ETF at 2023 lows.
- Rebellions Rises -- Samsung-backed S. Korean chip play, $400M raise, $2.3B valuation. The Nvidia counter.
- Mistral's Empire Build -- $830M debt for a 13,800-GPU Paris data center. Europe's sovereign compute play.
- DOJ's Countdown -- April 2 deadline to appeal Anthropic injunction. The government's next move.
Lead Story: Oracle's Ritual Sacrifice: 30,000 Jobs for the AI Gods
Oracle began executing what may be the largest layoff in its 48-year history. Termination emails hit US, India, Canada, and Mexico at 6 a.m. local time. From "Oracle Leadership," no warning, no human interface. A cold calculation.
TD Cowen estimates 20,000 to 30,000 people are affected. Roughly 18% of Oracle's 162,000-person global workforce. Oracle remains silent on the numbers. The cost of ambition.
The math is brutal, direct. $156 billion in AI infrastructure spending committed. The cuts free an estimated $8-10 billion. A $2.1 billion restructuring charge was already flagged in its March 10-Q filing. $982 million booked. A zero-sum game.
This isn't weakness. Oracle just posted a 95% jump in net income to $6.13 billion last quarter. Remaining performance obligations stand at $523 billion — up 433% year over year. This isn't a company in distress. It's a company making a capital-intensive AI bet its existing balance sheet can't absorb without gutting payroll. Re-architecture by force.
Oracle joins a pattern. Meta cut 16,000 to fund AI infrastructure. Fortune's CFO survey projected 502,000 AI-driven role eliminations in 2026. The question isn't whether AI reshapes labor markets. It's how quickly, and who bears the cost. The new rule: infrastructure over payroll.
Larry Ellison sits on Trump's PCAST council, alongside Zuckerberg and Huang. His company is a cornerstone of SoftBank's Stargate initiative. None of that insulated 30,000 employees from a dawn email. Not even Ellison's proximity to power offers insulation from the market's new incentives.
In Other News
Anthropic's Mythos leak keeps rippling through markets. The weaponization of AI is no longer hypothetical. Euronews and Axios pushed the Mythos story into its second week because its core implication remains unaddressed: security firms can't secure what they don't understand. Anthropic's internal warnings about 2026 cyberattack likelihood are chillingly prescient. The Global X Cybersecurity ETF (BUG) closed Friday at its lowest since November 2023, down 21% YTD. CrowdStrike, Palo Alto Networks, Zscaler — all bleeding value. Anthropic confirmed the model, blamed human error for the database leak, and in doing so, undermined its very claim to trust, all while winning a landmark court ruling. The irony is sharp.
Rebellions raises $400M at $2.3B, positioning as Asia's Nvidia alternative. The incentive structure for AI chips is diversifying. Samsung-backed South Korean AI chip startup closed a pre-IPO round led by Mirae Asset and the Korea National Growth Fund, bringing total funding to $850M. Rebellions builds for inference, not just training – the burgeoning market where AI actually does things. This is Seoul's direct challenge to Nvidia's dominance, their "K-Nvidia" initiative becoming a tangible reality. With RebelRack and RebelPOD hardware launched, US expansion is next, then IPO. Huawei and Arm are also staking their claims. The chip monopoly is fracturing.
Mistral bets $830M on European sovereign compute. Sovereignty isn't just about data; it's about the silicon and the power grid. The French AI lab raised $830M in debt from a seven-bank consortium to build a 13,800-GPU Nvidia GB300-powered data center near Paris, operational by Q2 2026, adding 44 MW. Their target: 200 MW across Europe by 2027, including a $1.4B Swedish data center. This isn't just growth; it's a strategic move to insulate European governments from American cloud providers. The architecture of trust demands local control.
MIT Technology Review: Pentagon's culture war against Anthropic "has backfired." The tactical misstep of the state. A detailed analysis argues Judge Lin's 43-page ruling exposed a deliberate escalation of a contract dispute through social media and political pressure. Anthropic found strange allies, from former Trump AI policy authors to Catholic ethicists. The White House's demand for ideological loyalty from AI companies reveals the limits of its leverage. The DOJ now faces an approximately April 2 deadline to seek an emergency stay from the Ninth Circuit. A test of power.
X / Social Pulse
Sam Altman, a rare moment of candor, admitted to DC leaders that "AI is not very popular in the US right now." The Oracle layoffs, 30,000 deep, landed the same week, a brutal validation of his point. Senator Mark Warner's projection of 30-35% college graduate unemployment by 2028 no longer sounds hyperbolic. Andrej Karpathy's hope for agents handling full application builds—removing the DevOps friction—underscores a future where Warner's prediction might be conservative. The "6 a.m. email" trending on Monday: the stark human cost of this acceleration. The market demands efficiency; humans are the variable.
One to Watch
The DOJ appeal window. The government has until roughly April 2 to seek an emergency stay of Judge Lin's preliminary injunction against Anthropic. This isn't just procedural; it's a test of wills. If appealed, the Ninth Circuit moves fast. If not, the designation remains frozen, agencies can use Claude, and Anthropic scores a quiet but massive win for enterprise procurement and its IPO. Emil Michael's conflict of interest (holdings in an Anthropic rival) and the D.C. Circuit case add layers of geopolitical strategy to a legal skirmish. The next 48 hours: will the Pentagon concede, or escalate? The incentives are clear.
Quick Hits
- Agent-Infra released AIO Sandbox -- an open-source all-in-one runtime for AI agents. A foundational step for autonomous systems (MarkTechPost).
- SK Networks invested 50B won -- (~$35M) in AI startup Upstage's Series C. The capital flow into emerging AI is relentless (Seoul Economic Daily).
- Physical Intelligence talks for $1B at $11B+ valuation -- doubling in four months. Founders Fund at the helm. Exponential growth isn't slowing (TechCrunch, Bloomberg).
- Enterprise AI governance hitting inflection -- Raconteur, National Law Review, CISO roadmaps. Agentic AI is the new compliance battleground, demanding new business architecture (Raconteur).
- Suno v5.5 now live -- with voice cloning and custom models. 2M paid subs, $300M ARR. AI is composing the future of media consumption.
The first quarter of 2026 closes with a stark lesson: AI's true cost isn't just computational; it's human. Oracle proves that the pursuit of infrastructure mastery devours existing structures. As April dawns, the Ninth Circuit will shape the battle lines between Washington and Silicon Valley, determining who dictates the terms of engagement in this new global architecture. The underlying incentives are clashing.
Sources
Oracle Layoffs: The Next Web | Seeking Alpha | Rolling Out | Fox Business
Anthropic Mythos: Euronews | Axios | Fortune | CNBC
Anthropic-Pentagon: MIT Technology Review | CNBC | The Hill | Defense One
Rebellions: TechCrunch | CNBC | Reuters | Silicon Canals
Mistral: TechCrunch | CNBC | Bloomberg | Data Center Knowledge
Other: MarkTechPost (AIO Sandbox) | Seoul Economic Daily | Raconteur
Lock in. M. mazen@thorterminal.com