Exclusive first look at Arm's new CPU

2 min video5 key momentsWatch original

⚡

TL;DR

Arm is abandoning its licensing-only model to manufacture its own chips, starting with the ARM AGI processor designed for power-efficient AI inference with Meta as its first customer.

Key Insights

flipping from licensing to manufacturing — Arm is flipping its entire business model. Instead of just collecting royalties on other companies' chips, it's now manufacturing silicon and going head-to-head with its own customers.

power efficiency advantage — The ARM AGI chip is specifically built for power efficiency in AI workloads. With data centers consuming massive amounts of electricity, Arm's low-power approach solves a real problem that tech companies are losing sleep over.

Meta as launch customer unusual — Meta signed on as the first customer, which is genuinely unusual for a first-generation chip. Getting a hyperscaler to adopt your debut processor is rare and risky.

crowded AI chip competition — The chip exists in a crowded market. Nvidia dominates AI chips, but Arm's expertise in low-power design gives it a different angle than other x86 competitors like Intel and AMD.

supply chain diversity rationale — Supply chain diversity matters to big tech. Meta specifically framed Arm as another option rather than a replacement, showing how customers want alternatives to avoid single-vendor lock-in.

Deep Dive

Arm's Radical Strategy Shift

For 35 years, Arm operated as a design house. It created CPU blueprints and instruction sets, then licensed them to Intel, AMD, Nvidia, Apple, and everyone else, collecting royalties on every chip sold. That model worked beautifully. But now Arm is breaking that playbook entirely. It's manufacturing its own silicon and competing directly against the companies that built their empires on Arm's technology. The new ARM AGI chip targets power-efficient AI inference specifically, launching with Meta as its first customer.

Why Power Efficiency Matters Right Now

Data center power consumption is becoming the limiting factor for AI expansion. One Arm executive put it bluntly: we don't even know how we're going to power the data centers we're building in 2028 and 2029. That's the opening Arm is exploiting. Its chips are ruthlessly optimized for low power consumption without sacrificing performance. It's a different design philosophy than Nvidia's approach, offering tech companies an alternative that addresses their growing electricity bills.

The Betrayal Factor (Sort Of)

Arm's move creates an obvious conflict of interest. It's now a competitor to Meta, Google, Amazon, and Microsoft, companies that helped build Arm's success through massive licensing deals. Arm framed it as supply chain diversity, which is technically true but also a euphemism for not wanting to rely entirely on Nvidia or any single vendor. Whether these customers stay loyal to Arm's licensing model while treating its chip division as competition is the real question.

Why CPUs Are Cool Again

The rise of agentic AI is creating fresh demand for general-purpose CPUs instead of specialized accelerators. Traditional x86 chips from Intel and AMD are seeing renewed interest. Arm's entry into manufacturing positions it to capture part of this renaissance, especially if developers need flexible, power-efficient processors for AI workloads that don't require Nvidia's specialized hardware.

Takeaways

✓Arm is betting that the power consumption crisis in AI data centers is big enough to justify entering manufacturing and competing with its own customers.
✓Meta's adoption of Arm's first-generation chip signals confidence but also underscores how desperate tech companies are for alternatives to vendor lock-in.
✓Watch whether Arm can actually execute manufacturing at scale. Designing chips and making them reliably in factories are two completely different problems.
✓The crowded AI chip market just got more crowded, but Arm's angle on power efficiency is genuinely different from Nvidia's performance-at-all-costs approach.

Key moments

0:15The Business Model Flip

“Now ARM's going to compete with most of these mega customers, too. Joining their ranks as a fabulous chip company because it's now in a whole new business making physical silicon.”

0:25Chip Design Philosophy

“It was really designed for the AGI era in that it's been incredibly low power. It's incredibly high performance and it's sort of ruthlessly optimized.”

0:35Meta as First Customer

“This is ARM's new in-house central processing unit, ARM AGI, built specifically for power efficient inference with Meta as its initial customer.”

0:50Supply Chain Diversity Rationale

“It's just another option in the world for us, right? It's it's all about supply chain diversity.”

1:30The Power Crisis Problem

“We really don't even know how we're going to power data centers that we're building in 2028 and 2029. And ARM is really good at low power at a specific performance level.”

Get AI-powered video digests

Follow your favorite creators and get concise summaries delivered to your dashboard. Save hours every week.

Start for free