Hello {{first_name|Motivated and Miffed Community}},

Three things defined AI this week: a company sued the federal government over its right to say no, two tech giants announced they're cutting tens of thousands of jobs to pay for AI infrastructure they haven't built yet, and Nvidia held its annual conference where the theme was less "look at our shiny GPUs" and more "the CPU is back, actually." Heavy week. Let's get into it.

⚖️🤖 Anthropic sued the Pentagon. The hearing is March 24.

🏗️💸 Oracle plans up to 30,000 layoffs to fund data centers it can't yet afford.

🔄⚙️ Nvidia GTC: agentic AI is shifting the compute stack from GPUs to CPUs.

🍎🤝 Apple is replacing Siri's brain with Google Gemini. (Yes, really.)

2026 will redefine IT as a strategic driver of global growth. Automation, AI-driven support, unified platforms, and zero-trust security are becoming standard, especially for distributed teams. This toolkit helps IT and HR leaders assess readiness, define goals, and build a scalable, audit-ready IT strategy for the year ahead. Learn what’s changing and how to prepare.

Download the Toolkit

Anthropic said its AI shouldn't be used for mass surveillance of U.S. citizens or in fully autonomous weapons without human oversight. The Pentagon said it wanted to use Claude for "all lawful purposes" — and when Anthropic wouldn't budge, the Department of Defense designated the company a "supply chain risk," a label that's historically been reserved for foreign adversaries. President Trump then directed all federal agencies to immediately stop using Anthropic's technology. Anthropic's OneGov contract — which had given the government access to Claude across all three branches — was terminated.

So Anthropic sued. Two lawsuits, actually: one in California federal court challenging the designation as unconstitutional retaliation, and one in the D.C. Circuit Court of Appeals seeking an emergency stay. The company's CFO said the government's actions could cost Anthropic hundreds of millions — or potentially multiple billions — in 2026 revenue. A hearing is scheduled for March 24.

What's notable is who's rallying around them. Nearly 150 retired federal judges — appointed by both parties — filed amicus briefs. So did industry groups, former national security officials, and employees from competing labs including OpenAI and Google DeepMind. Even Microsoft weighed in, arguing that using a national security blacklist against a domestic company over a policy dispute sets a precedent that makes the entire procurement system contingent on political favor.

The administration's counterargument is essentially: a private company can't dictate how the military uses its tools. The courts will sort that out. A March 24 hearing will give the first real read on whether the blacklist holds or gets paused.

Why it matters: The outcome of this case will determine whether AI companies can maintain ethical red lines when doing business with the government — or whether "all lawful purposes" becomes the industry default.

Oracle reportedly plans to eliminate between 20,000 and 30,000 positions — potentially 18% of its global workforce — to free up $8 to $10 billion in cash flow. The goal: fund a massive AI data center expansion it committed to as part of a $300 billion cloud deal with OpenAI through the Stargate initiative. Wall Street analysts have said Oracle could run negative on cash flow from this bet for years, with returns not expected until around 2030. (Bloomberg reported this; Oracle hasn't officially confirmed.)

Block (the Jack Dorsey company behind Square and Cash App) also cut roughly 40% of its workforce — around 4,000 jobs — with Dorsey explicitly attributing the cuts to efficiency gains from AI tools. That one was more direct than Oracle's slower-burn infrastructure argument, but both announcements landed in the same week.

The framing debate here is real. Fast Company notes that there's limited evidence Oracle is replacing workers with AI at scale — the cuts look more like a cost reset to cover infrastructure spending that hasn't paid off yet. Dorsey, on the other hand, was unusually explicit: AI made those roles redundant. Both narratives can be true at the same time, and that's the uncomfortable part. Across the industry, more than 55,000 tech workers have already lost jobs in 2026 so far, according to TrueUp.io, with AI cited as a factor in roughly one in five of those cuts.

Why it matters: We're watching companies use AI as both a reason to invest and a reason to cut headcount — sometimes in the same announcement. The bill for the AI buildout is being paid in jobs right now, with the promise of returns that may take years to materialize.

Nvidia held its annual GTC conference in San Jose this week, and the headline shift wasn't a new GPU — it was the argument that the CPU is now the bottleneck. As AI moves from call-and-response chatbots toward agentic systems (AI that spawns other AI agents to complete multi-step tasks), the kind of compute needed changes. GPUs handle parallel operations well. But agentic orchestration — coordinating between agents, managing data transfer, running sequential logic — that's CPU territory.

So Nvidia unveiled its Vera CPU rack, a standalone CPU system going directly after Intel and AMD in the data center. They also released the Groq 3 Language Processing Unit (LPU), built for inference — the compute phase that happens when you actually use a deployed AI model, as opposed to training it. The Groq 3 LPX rack pairs 128 of those chips with Nvidia's Vera Rubin GPU system, and the company says the combination yields 35x higher throughput per megawatt of power.

CEO Jensen Huang projected $1 trillion in purchase orders between Blackwell and Vera Rubin through 2027. CNBC noted that Wall Street was a little underwhelmed — partly because Nvidia previewed Vera Rubin at CES in January rather than saving it for GTC. But the strategic story is coherent: Nvidia isn't just a GPU company anymore. It's building toward owning the full AI compute stack, from training to inference to agentic orchestration.

Why it matters: The shift toward agentic AI isn't just a product trend — it's a hardware story. The companies best positioned for what comes next are the ones redesigning compute from the ground up, not the ones selling faster versions of what already exists.

Apple has officially announced that its redesigned Siri — the one that was supposed to be the company's big AI moment — will run on Google's Gemini model. Not Apple Intelligence. Not some quiet internal model. Google Gemini, a 1.2 trillion parameter system, running on Apple's Private Cloud Compute infrastructure to preserve privacy. The update targets a March 2026 rollout alongside iOS 26.4.

The company spent years and billions positioning itself as the privacy-first AI alternative. And the new Siri is real — context-aware, cross-app, genuinely useful by most accounts. It's just not Apple's AI underneath.

Why it matters: When the most privacy-obsessed consumer tech company decides the pragmatic move is to license its competitor's model, it tells you something about how difficult it is to build frontier AI from scratch — and how the race is increasingly about integration, not invention.

This week rhymed: the government tried to reshape AI through contracts, Oracle and Block tried to reshape it through capital, and Nvidia quietly announced it's building the pipes that all of it runs through. The power to define what AI becomes isn't just in the models — it's in who controls the infrastructure, the courtrooms, and the compute stack.

Stay MOTIVATED,

Gio