Quentir Medicine Monitor
Evidence-based insights for quantum medicine. Published by Quentir Systems LLC · November 14, 2026.

PsiQuantum and AMD said on 12 November 2026 that their first commercial target for a joint quantum-classical computing platform is drug discovery, starting with chemistry that matters for cancer medicines. The two companies announced an expanded partnership to co-design the classical and photonic hardware that a fault-tolerant quantum computer will need, and AMD said it plans to take an equity stake in PsiQuantum.
The medical claim in the announcement is short. According to the joint press release published by PsiQuantum, PsiQuantum's computational chemistry workflows, run at large scale on AMD graphics processors, have "already delivered initial breakthrough results in simulation of covalent inhibitors, a class of molecules relevant in drug discovery for oncology." The release gives no molecule, no target protein, no accuracy figure and no publication. It also gives no amount for the planned AMD investment. For a pharmaceutical or hospital research reader, the useful work is to separate what was shown from what was announced.
Today's chemistry results come from conventional GPUs. The quantum machine they are meant to prepare for does not yet exist. PsiQuantum says its methods improve on current GPU-based approaches today while "preparing critical industries for the arrival of fault-tolerant quantum computing." Only the second of those claims depends on quantum hardware.
What PsiQuantum and AMD announced in Palo Alto and Santa Clara on 12 November 2026
The partnership builds on an earlier choice: PsiQuantum had already selected AMD Versal FPGA chips, programmable processors that can be rewired in hardware, for the control electronics of its photonic quantum computer. The new agreement widens that relationship to AMD Instinct GPUs, AMD EPYC server processors and the AMD ROCm software stack, combined with PsiQuantum's silicon photonics platform. The companies describe the goal as a hybrid computing fabric in which classical supercomputing, AI and photonic quantum processors work together.
Victor Peng, chief executive of PsiQuantum, is quoted listing healthcare first among the fields where he expects quantum computing, working with AI and supercomputers, to solve problems that are out of reach today. Mark Papermaster, AMD's chief technology officer, says that leadership in classical computing is essential to building quantum systems at scale. Both firms say they will explore a footprint at the Illinois Quantum and Microelectronics Park in Chicago, where PsiQuantum is the anchor tenant, and that they will say more at the SC26 supercomputing conference in Chicago from 15 to 20 November 2026.
Quantum pillar: computing. Technology readiness: TRL 2 of 9. In Quentir's assessment, using a fault-tolerant quantum computer to design cancer drugs is still at the concept stage: the chemistry runs on ordinary graphics processors today, and the error-correcting decoder has been tested on a single programmable chip, far from any machine that could serve a drug program.
Why covalent inhibitors for cancer are a demanding test for quantum chemistry
Most drugs hold on to their target protein through weak, reversible contacts. A covalent inhibitor goes one step further: after it settles into the binding pocket, a reactive group on the drug forms a chemical bond with an amino acid on the protein, often a cysteine. Several approved cancer medicines work this way, including osimertinib for EGFR-mutant lung cancer and ibrutinib for some blood cancers. The bond can make a drug more potent and longer-acting, and it also raises a safety question, because the same reactive group can bond with proteins it was never meant to touch.
That is why this chemistry is hard to compute. The conventional fixed-topology force fields used in most computer-aided drug design treat atoms as balls on springs with a fixed set of bonds, so they cannot describe a bond being made or broken. Predicting how fast and how selectively a covalent warhead reacts requires the electronic structure of the reacting atoms, which is the domain of quantum chemistry. Accurate methods for that problem scale very steeply with the number of electrons, and drug companies usually fall back on approximations whose errors are hard to judge in advance. Better reaction energies would help medicinal chemists choose warheads that react with the intended cysteine and leave other proteins alone.
The announcement does not say which of these quantities PsiQuantum computed, how its result compares with an experimental measurement, or how large the claimed improvement over current methods is. Until a paper or a benchmark appears, the covalent-inhibitor work is a company statement about classical simulation on GPUs. Earlier Quentir coverage gives a sense of how far the quantum part still has to go: GSK's Anthony Rowe has put useful quantum chemistry at about 2,000 logical qubits, around 2030, and no machine of that size exists.
What the PRISM decoder paper of 10 November 2026 measured on an AMD FPGA
The one part of the announcement backed by a public technical document is the error-correction work. A fault-tolerant quantum computer constantly checks its own qubits for errors, and a classical decoder must interpret those checks quickly enough for the machine to account for the errors in the operations that follow. In PsiQuantum's photonic design, a dominant source of error is the loss of photons, which the hardware can detect and locate.
The PRISM preprint on arXiv, posted on 10 November 2026 by Daniel Barter, Jacob C. Bridgeman, Chris Dawson and co-authors, describes a transformer-based machine-learning decoder built to use those known loss locations. The authors report that PRISM matches the logical error rates of decoders based on integer programming while running significantly faster, across surface codes, color codes and quantum low-density parity-check codes at distances up to 24. On an AMD VP1902 FPGA, their optimized version decoded logical blocks of distance 8 with a latency of 6.59 microseconds and an amortized decoding time of 1.56 microseconds per block.
The decoder figures say nothing yet about molecules or patients. They do show where AMD's hardware enters the quantum computer itself, and they are specific enough for other groups to test.
How the DOE Quantum Genesis Priority Applications of 8 October 2026 treat biochemistry
The release ties the drug-discovery focus to the U.S. Department of Energy's Quantum Genesis Priority Applications, announced on 8 October 2026. DOE set out eight priority problems in four domains: chemistry, materials, subatomic physics and applied mathematics. It placed biochemistry, together with plasma physics and optimization, among emerging applications where, in DOE's words, the fault-tolerant quantum computing literature "is still early but where future impact is possible." PsiQuantum describes drug discovery as falling within biological modeling, an emerging application area for Quantum Genesis. DOE thus lists biochemistry among the emerging applications it monitors, outside its eight priority applications.
What oncology researchers and hospital research offices can ask before and after SC26 in Chicago, 15 to 20 November 2026
For cancer centers and hospital research offices, the practical question is what would make this announcement useful to a drug program. Four items would settle most of it: the covalent inhibitors and protein targets simulated; the quantity computed, such as a reaction barrier or a binding energy; a comparison with experimental data or a recognized high-accuracy method; and the share of the work that ran on quantum hardware, which on the release's own account is currently none. A roadmap toward large logical-qubit counts matters to a drug program only once those details are public. The companies say they will share more about the partnership at SC26, which gives researchers an early chance to ask for these details; the release does not promise oncology benchmarks there.
Sources
Primary source: joint press release of PsiQuantum and AMD, Palo Alto and Santa Clara, 12 November 2026, quoting Victor Peng (PsiQuantum) and Mark Papermaster (AMD). Also drawn on: the PRISM preprint by Daniel Barter, Jacob C. Bridgeman, Chris Dawson and colleagues (arXiv, 10 November 2026), the U.S. Department of Energy announcement of the Quantum Genesis Priority Applications, and earlier Quentir coverage.