📊 Executive Summary

🔬 Technical Deep Dive

Current State

Buried germanium is now the most credible hole-spin platform in existence. A strained Ge quantum well sandwiched in SiGe hosts holes with an effective mass near 0.05 mₑ, no valley degeneracy, strong intrinsic spin-orbit coupling that enables all-electrical spin driving, and mobilities that have climbed past three million cm²/Vs in undoped stacks. Delft, IMEC, Forschungszentrum Julich and Intel have all pushed Ge/SiGe onto 300 mm substrates. Four-qubit processors, 4x4 crossbar arrays and singlet-triplet qubits have all been demonstrated in this material system since 2021.

The missing half of the story is the superconductor. To build Andreev spin qubits, gatemons, superconducting-spin interfaces or planar topological devices, you need a superconductor that makes an electrically transparent, hard-gapped contact to a quantum well sitting tens of nanometres below the surface, and that keeps superconducting while you apply the field needed to Zeeman-split the spins. Those two demands pull in opposite directions. The 2026 PtIrSiGe result is an attempt to satisfy both by alloying iridium into the platinum germanosilicide recipe that Delft introduced in 2023, deliberately shortening the electronic mean free path to push the upper critical field up.

Values above are approximate and drawn from published reports across different device geometries; direct comparison should be treated as indicative rather than exact.

Recent Breakthroughs

The physics behind the improvement is unglamorous and completely sound. In a dirty type-II superconductor the out-of-plane upper critical field scales as Φ0 divided by 2πξ², where ξ is the coherence length. Alloying iridium into the germanosilicide adds scattering centres, collapses the mean free path, and drags ξ down into the sub-20 nm regime. Reaching 1 T out of plane requires ξ around 18 nm or less, which is exactly the range a heavily disordered silicide alloy occupies. Heavy-element spin-orbit scattering also relaxes the Pauli paramagnetic limit, so the in-plane resilience should benefit as well.

What matters more than the field number is how the contact is made. The germanosilicide is not deposited onto the quantum well; it is grown into it. Metal is patterned on the surface, annealed, and consumes the SiGe barrier until the reaction front intersects the buried Ge channel. No trench etch, no regrowth, no ex-situ oxide at the critical interface. That self-aligned, planar geometry is why the PtSiGe predecessor produced a hard superconducting gap in the first place, and why gate-tunable Josephson junctions in this material have shown clean supercurrent modulation.

Second, the thermal budget is compatible with a CMOS back end, and platinum and iridium silicides are already familiar to contact engineers. A superconductor formed by the same class of process step used for source-drain contacts is a much easier sell to a 300 mm line than epitaxial aluminium grown in an MBE chamber. That is the practical argument for taking this preprint seriously.

Third, out-of-plane resilience specifically unlocks planar device geometries. Germanium hole g-factors are strongly anisotropic, and many useful operating points for hole spin qubits and for Andreev states sit at fields perpendicular to the wafer. A superconductor that tolerates perpendicular field without vortex proliferation lets designers place resonators, flux lines and spin qubits on the same die without contorting the magnet geometry around the material's weakness.

Remaining Challenges

The trade-off is written into the physics. Adding disorder to raise the critical field also suppresses the transition temperature and degrades the superfluid density. A superconductor with Tc in the few-hundred-millikelvin band forces operation well below 100 mK for a clean induced gap, which constrains the cooling power budget per chip and rules out the emerging class of above-1 K spin qubit demonstrations that Basel and others have chased. Higher field tolerance bought with a lower gap is a real cost, not a free lunch.

Interface control is the second problem. Diffusion-formed contacts are stochastic at the nanometre scale. The lateral position of the germanosilicide edge, and therefore the junction length, is set by anneal time, temperature ramp and local strain. Wafer-to-wafer and die-to-die spread in junction transparency is the metric that will decide whether this becomes a manufacturable process or a heroic single-device result. The preprint reports device-level physics; it does not yet report yield statistics across a wafer.

Third, vortex dynamics. High Bc2 does not mean vortex-free. Above the first critical field, flux enters, pins, and generates low-frequency noise that couples straight into spin coherence and resonator quality factors. Pinning-site engineering, hole arrays and moat structures will all need to be characterised before anyone claims that 1 T is a usable operating field rather than a survivable one.

Finally, the honest limitation: an out-of-plane critical field above 1 T is a materials milestone, not a qubit milestone. No coherence numbers for an Andreev spin qubit built on PtIrSiGe exist yet, and the gap between a hard-gapped junction and a long-lived, gate-controlled quantum state has swallowed several promising material systems already.

Expert Perspectives

The hole-spin community has converged on a shared view over the last two years: germanium's advantage is process compatibility, not raw coherence. Groups at QuTech and IMEC have repeatedly framed Ge/SiGe as the only spin platform where the qubit, the superconducting element and the control transistor could plausibly come off the same fab line. This preprint fits that narrative precisely.

The topological-device community is more cautious. After Microsoft's contested 2025 Majorana claims and the long history of soft gaps and trivial zero-bias peaks in InAs and InSb, reviewers now demand hard-gap evidence, multi-terminal correlations and disorder characterisation before entertaining topological interpretations. Nobody serious is calling a high-field germanosilicide a topological result. What it is, credibly, is an enabling material for Andreev spin qubits and for superconducting-spin coupling experiments that need field headroom.

Materials engineers at foundry-adjacent institutes tend to ask a different question first: what is the anneal window, and does it collide with the gate stack? Platinum-based silicides typically form in the 300-500 C range, which sits uncomfortably close to the thermal limits of high-quality gate dielectrics on SiGe. Expect the next round of papers to be about process integration order, not about physics.

🏢 Market Landscape

Key Players

On the research side, the centre of gravity is European. QuTech and TU Delft, working with IMEC, have driven the Ge/SiGe hole-spin programme from single dots in 2018 to 4x4 crossbar arrays and gate-tunable Josephson junctions. Forschungszentrum Julich supplies some of the highest-mobility Ge heterostructures in the world. CEA-Leti and Quobly in Grenoble are pursuing FD-SOI silicon spin qubits with a similar foundry-first thesis. Intel's Components Research group has shipped Tunnel Falls silicon spin chips to academic partners, establishing the template of a 300 mm line producing quantum test vehicles at scale.

Among startups, Diraq (Australia, Si MOS), Quantum Motion (UK, CMOS silicon), Equal1 (Ireland), SemiQon (Finland) and Photonic Inc (Canada, silicon T centres) all compete for the semiconductor-qubit narrative, though none has publicly committed to germanium as the primary channel. Microsoft remains the most visible advocate for superconductor-semiconductor hybrids through its InAs/Al topological programme. Nokia Bell Labs and IBM maintain hybrid device research without productising it.

The equipment and materials layer is where the commercial exposure is most legible today: Applied Materials, Lam Research and ASM International for epitaxy and contact metallisation; Oxford Instruments and Bluefors for dilution refrigeration and high-field magnets; Keysight, Zurich Instruments, Quantum Machines and FormFactor for cryogenic measurement and probing. Every high-field hybrid experiment consumes magnet time and cryostat capacity, and those vendors get paid regardless of which qubit modality wins.

Investment Trends

Private quantum funding set records in 2025, with PsiQuantum's roughly $1B round at a reported $7B valuation anchoring the year and total private investment across the sector estimated in the $2-3B range. Public commitments dwarf that: the EU Quantum Flagship at €1B over a decade, Quantum Delta NL at €615M, the UK's National Quantum Strategy at £2.5B over ten years, Japan's Moonshot programme, and Chinese state investment widely estimated above $15B. Germanium and silicon spin work is disproportionately funded by these public programmes because the payoff horizon exceeds venture patience.

Notably, semiconductor spin qubits attract a small slice of private capital relative to superconducting transmon and neutral-atom companies, despite a stronger long-term scaling argument. That asymmetry is the investment thesis and the risk at the same time: cheap exposure to a modality that could leapfrog, or a modality that stays perpetually five years out.

Competitive Dynamics

The competitive question is not germanium versus silicon. It is whether any semiconductor spin platform can close the gap on qubit count before superconducting and neutral-atom systems reach useful error-corrected operation. Transmon processors are at the thousand-qubit scale with demonstrated below-threshold error correction; spin qubits are in the tens. The counterargument, which the PtIrSiGe work supports, is that spin qubits sit at 100 nm pitch on existing 300 mm infrastructure, so their scaling curve, once started, is steeper.

Within hybrids, the InAs/Al ecosystem retains a head start in device sophistication but carries the reputational overhang of contested topological claims and a low field ceiling. A germanium platform with tesla-class field tolerance and CMOS-compatible contacts is a genuine second entrant, and one that does not require III-V materials in a silicon fab.

Market Projections

Consensus places 2025 quantum computing revenue near $1.6B, rising to roughly $8-12B by 2030 depending on how aggressively cloud access and simulation services are counted. McKinsey's longer-range framing puts quantum technology value creation at $28-72B by 2035 and far higher by 2040. Semiconductor spin qubits capture essentially zero of today's revenue. Their plausible share arrives after 2032, and only if a fab-manufactured array of thousands of qubits with acceptable uniformity is demonstrated. The nearer-term monetisable market touched by this research is cryogenic infrastructure and high-field magnet systems, a segment growing at a healthy double-digit rate with real order books today.

📅 Timeline & Milestones

💰 Investment Perspective

💡 Key Takeaways

💡 Lab Test Report

📖 Sources & References