IBM Spins Off Its Quantum Chip Foundry. The $2B CHIPS Act Bet Just Got Real

IBM just took its quantum chip manufacturing arm and made it a separate company. Attached to the move: a $2 billion check from the CHIPS Act. That’s not a research grant — that’s fab money. And it marks the moment quantum computing stops being a “someday” technology and starts looking like an actual supply chain.

Why Spin It Off Now

IBM has run a vertically integrated quantum stack for years. Eagle, Osprey, Condor, and last year’s Heron R2 — every chip designed and fabbed in-house. So why suddenly cut the foundry loose?

One word: customers. IBM isn’t the only company that needs quantum silicon anymore. Google, IonQ, Rigetti, PsiQuantum, plus a wave of Chinese and European startups all have their own chip designs and nowhere viable to manufacture them. TSMC and Samsung don’t do millikelvin processes or Josephson junctions. And no startup is realistically going to spend a decade building its own quantum fab.

The problem for IBM was credibility. No competitor wants to hand its chip designs to a sister division of its biggest rival. Spinning out a neutral foundry solves that — same playbook TSMC ran in the 1990s, now ported to quantum. Pure-play manufacturing, no conflict of interest, take everyone’s orders.

What $2 Billion Actually Buys

Stack this against the CHIPS Act’s other big checks — Intel Ohio, TSMC Arizona, Samsung Texas — and this is the first time quantum has landed in that weight class. The number isn’t symbolic. It’s the cost of actual fab equipment.

The national security logic is unsubtle. Once fault-tolerant quantum machines come online, RSA and ECC become decorative. Whoever fabs those chips controls a strategically critical bottleneck. Washington has decided that quantum supply chains belong inside US borders, and it’s willing to spend like it.

The other tell: this money is going to manufacturing capacity, not papers. Quantum has lived in the research-grant economy for a decade. A CHIPS Act-sized check to build production lines is a different kind of bet entirely.

A Quantum Fab Is Not a Regular Fab

Superconducting quantum chips operate at roughly 15 millikelvin — colder than deep space. The qubits etched onto them are sensitive to magnetic noise, mechanical vibration, even stray cosmic rays. Yield, testing, packaging — every step from the classical playbook has to be rewritten.

That means a quantum foundry isn’t just a smaller-geometry CMOS line. It has to bundle cryogenic test infrastructure, microwave wiring, and shielded packaging into the service. The barrier to entry is brutal, which is exactly why this is a winner-take-most market. IBM is clearly trying to be the one taking most of it.

Who Has To Respond

China is the most obvious loser. Origin Quantum and the Guizhou quantum cluster have been building domestic capability, but a US pure-play foundry with $2B behind it widens the gap fast. Europe’s IQM and Pasqal have credible technology but thin manufacturing infrastructure.

The more interesting question is what happens inside the US. Does Google keep running its own Santa Barbara fab or send some volume to IBM? PsiQuantum’s photonic approach sidesteps the issue, but every superconducting startup now has a real option: stay fabless and just design. We’re about to find out whether quantum follows the ARM-plus-TSMC pattern that reshaped classical chips — and on this timeline, it’s emerging in a single move rather than over two decades.

The Takeaway

This spinoff is the moment quantum computing crosses from lab curiosity into industrial product category. The revenue story is still years out, but the infrastructure bet has started — and infrastructure bets are usually placed before the market is obvious, not after. Two billion dollars sounds like a lot until you consider what the entrance fee will look like if quantum actually arrives. The question worth sitting with: does industrial-scale quantum land inside this decade, or are we watching another fusion-style “ten years away, always”?