wasl.tech

AMD Buys Into PsiQuantum: The Chip Giant’s Big Bet on Quantum Computing

Picture a computer that can design a new cancer drug in days instead of years. Now picture the machine behind it: not just a quantum processor, but a small data centre of ordinary chips working furiously to keep it on track.

That second machine is the part most people never hear about — and it is exactly what AMD has just bought into.

On Thursday, November 12, 2026, AMD announced it plans to take an equity stake in PsiQuantum, the Palo Alto startup building photonic quantum computers.

The two companies are expanding a partnership to co-design the classical computing infrastructure that utility-scale, fault-tolerant quantum machines will need — and to build applications that mix quantum computing with high-performance computing and AI.

Neither company disclosed the size of the investment. But the shape of the deal says a lot about where quantum computing is heading: the race is no longer just about qubits.

It is about the ordinary silicon that has to surround them, correct their errors in real time, and turn fragile quantum signals into useful answers.

What the AMD PsiQuantum Deal Covers

The partnership is not new, but its scope is. PsiQuantum already uses AMD’s Versal FPGAs — programmable chips that process the stream of measurements coming off its qubits — in its fault-tolerant design. The expanded deal adds AMD Instinct GPUs for large-scale simulation, EPYC server CPUs and the ROCm software stack.

Those pieces will be combined with PsiQuantum’s silicon photonics platform, which builds quantum hardware using chip-manufacturing techniques and light-based connections to move information. The goal, the companies said, is a hybrid computing fabric: quantum, high-performance and AI processing working as one system rather than three separate ones.

AMD will provide the CPUs, GPUs, adaptive computing technology and software. PsiQuantum brings the photonics. Together they will co-design the infrastructure and develop the applications that run on it — a division of labour that plays to each company’s strengths.

The first application target is drug discovery. PsiQuantum says its computational chemistry workflows already run on AMD hardware, and combining those with GPU-driven simulation has helped model covalent inhibitors, a class of compounds used in some oncology research. It is a concrete first job, not a vague promise about the distant future.

Why Quantum Computers Need AMD’s Ordinary Chips

Quantum computers are famously fragile. Their qubits lose their state at the slightest disturbance, so a useful machine must constantly check for errors and correct them — thousands of times a second, across millions of measurements.

That error correction is a classical computing problem, and a huge one. Every measurement coming off the qubits has to be read, decoded and acted on almost instantly, or the calculation falls apart.

The chips doing that work need to be fast, programmable and close to the quantum hardware.

That is where AMD’s FPGAs already earn their place. PsiQuantum’s PRISM decoder, the system that handles error correction, ran on an AMD VP1902 FPGA with 6.59 microseconds of latency, according to a preprint published on November 10.

In quantum terms, six millionths of a second is the difference between a correction that arrives in time and one that arrives too late.

The expanded partnership takes that logic further. Instinct GPUs will handle the large-scale simulation work — modelling molecules and materials before the quantum machine ever touches them — while EPYC CPUs orchestrate the whole system. It is the same hardware AMD sells into AI data centres, pointed at a new kind of problem.

Mark Papermaster, AMD’s chief technology officer, put it plainly: classical computing leadership is essential to building quantum systems at scale. Without it, he argued, the quantum hardware itself cannot deliver on its promise.

Who PsiQuantum Is — and Why AMD Wants In

PsiQuantum was founded in 2016 and is based in Palo Alto, California. It builds photonic quantum computers — machines that use particles of light, rather than superconducting circuits or trapped ions, as their qubits. The approach lets it manufacture quantum hardware with the same chip-making techniques the semiconductor industry has refined for decades.

The company raised $1 billion in September 2025 at a $7 billion valuation, in a round that included NVentures, the venture arm of Nvidia. That detail matters: AMD’s biggest rival already has a financial link to PsiQuantum, and AMD’s equity stake now gives it a seat at the same table.

There is also a personal connection. PsiQuantum’s chief executive, Victor Peng, is a former president of AMD. He knows the company’s chips, its engineers and its roadmap from the inside.

The partnership also covers silicon photonics networking and groundwork for next-generation AI as well as quantum computing itself.

The two companies will also explore a joint presence at the Illinois Quantum and Microelectronics Park in Chicago, where PsiQuantum is building a facility. They plan to present further technical details on their joint roadmap at SC26, the supercomputing conference in Chicago, which runs from November 15 to 20.

A Week That Shows Where AMD Is Heading

The PsiQuantum deal caps a crowded week for AMD. On Tuesday, its new EPYC 6 Venice server processors became generally available, with a flagship 256-core chip aimed squarely at AI workloads. On Thursday, the company refreshed its Instinct lineup, disclosing the MI430X and inference results for the MI455X on Helios systems.

The same week brought a Google Cloud design win: fifth-generation EPYC processors now power the host infrastructure for Google Cloud’s TPU 8tx, which both companies described as the first cloud TPU system with hardware-backed confidential computing. AMD also reported third-quarter data-centre revenue growth of 97% year over year on November 4.

Yet the market did not celebrate. AMD’s shares fell 8.6% on Friday as unease spread through the chip sector, dragging down the Philadelphia Semiconductor Index.

The quantum announcement, the Google win and the new silicon were all buried under the sell-off — a reminder that investors are currently more worried about chip demand than excited about future technology.

That tension is the story of AMD’s week in miniature. The company’s products are landing in the places that matter — AI data centres, cloud infrastructure, and now quantum computing — while its share price reflects a market that wants proof of earnings, not just announcements.

The Government Is Betting on Quantum Too

AMD is not the only one putting money into quantum computing. In May, the US Department of Commerce announced plans to take minority equity stakes in nine quantum companies as part of a $2 billion investment under the CHIPS Act — including a $100 million deal with PsiQuantum.

That government backing is now turning into real awards. Atom Computing finalised its own $100 million CHIPS R&D award on November 13, for fault-tolerant, utility-scale neutral-atom quantum computers. The pattern is clear: Washington sees quantum as strategic infrastructure, and it is willing to pay to keep the work in the United States.

The field is moving fast elsewhere too. IBM unveiled its Quantum Nighthawk r3 earlier this month, a 520-qubit system designed for 10,000 quantum operations, with plans to link two chips into a 1,040-qubit machine. Every major player is racing toward the same milestone: a fault-tolerant machine that can run calculations no classical computer can touch.

That race is why the classical side matters so much. A quantum computer with a million qubits is useless if the ordinary computers around it cannot keep up with the error correction. The company that owns that surrounding infrastructure owns a piece of every quantum machine that gets built.

What Happens Next

The first test of the AMD PsiQuantum partnership will come quickly. The companies present their joint roadmap at SC26 in Chicago this week, where the supercomputing community will get its first detailed look at what the hybrid system is supposed to do — and when.

Drug discovery is the headline application, and it is a demanding one. Modelling a covalent inhibitor well enough to guide real chemistry requires simulation at a scale that strains even today’s largest supercomputers. If the hybrid system can do it faster or more accurately, pharmaceutical companies will notice.

But the honest answer is that nobody knows yet how fast this will move. Fault-tolerant quantum computing has been five years away for a long time, and the history of the field is full of timelines that slipped.

What has changed is the money: governments, chip giants and venture capital are all funding the same bet at the same time.

AMD’s stake in PsiQuantum is a bet that the classical backbone is where the value will sit — that whoever builds the ordinary computers around the quantum ones will be indispensable. It is a quieter bet than building qubits, but it may turn out to be the smarter one.

The quantum future, according to Reuters, will be built on silicon that already exists.

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top