2026-03-26
By Vadym · Generated with AI, curated by me
The hardware arms race just got personal. Arm is making its own chips, Jensen Huang says AGI already arrived, and OpenAI is raising at a valuation that makes most countries’ GDPs look modest. Meanwhile, the workforce implications are getting harder to ignore.
NVIDIA’s CEO declared that artificial general intelligence has arrived, defining it as systems capable of autonomously building and running billion-dollar companies. The statement came as NVIDIA continues to position itself as the essential infrastructure provider for the AI era, tying its future directly to energy production and compute scale.
Arm has moved beyond licensing chip designs and into making its own silicon, unveiling the AGI CPU purpose-built for AI data center workloads. This marks a strategic pivot from the company’s decades-old licensing model to direct competition in the AI chip market.
OpenAI is pursuing a $10 billion raise at a $730 billion valuation, offering investors a 17.5% guaranteed return to attract private equity. The company has surpassed $25 billion in annualized revenue, while rival Anthropic approaches $19 billion.
A Wall Street Journal survey found that chief financial officers expect AI to reduce company headcount this year, with administrative roles particularly exposed. The survey signals a shift from “AI augments workers” to “AI replaces positions” in executive planning.
Oxford spinout Stateful Robotics raised $4.8M in pre-seed funding led by Amadeus Capital Partners to build long-term memory for physical AI. The startup addresses a core gap: foundation models can perceive the world, but they don’t remember what happened yesterday on a job site or how a specific asset behaves over time.
Today’s stories share a common thread: the AI industry is moving from building models to building moats. Arm is making chips. OpenAI is hoarding capital. CFOs are planning workforce cuts. Even a tiny Oxford startup is building state layers that no foundation model provides out of the box. The model race hasn’t ended, but the winners will be decided by infrastructure, distribution, and execution — not benchmark scores.
— Boba
Curated by Vadym