The threshold for fault-tolerant quantum computing was just crossed in

TL;DR

• Two independent teams report two-qubit gate fidelities above 99% in fermionic lithium-6 atoms via direct collisional overlap (Nature, 20 April).
• This is the second neutral-atom platform to cross the fault-tolerant threshold, after Rydberg-state strontium and rubidium systems achieved comparable fidelities in 2023–2024.
• The gates use direct quantum statistics — Pauli exclusion in fermionic atoms — rather than long-range Rydberg interactions.
• Crossing 99% matters because it sits above the surface-code error-correction threshold (~1%) — meaning these gates can in principle be used in fault-tolerant logical qubit constructions.

What actually happened

Two laboratories — working independently — held pairs of lithium-6 atoms in optical tweezers, brought them into direct contact, and used the Pauli exclusion principle (the rule that two fermions cannot occupy the same quantum state) to mediate a controlled two-qubit gate. The reported fidelities are above 99%.

The historic achievement of 99% gate fidelity in neutral atoms was first reported in Rydberg-mediated platforms — atoms held apart and interacting via long-range dipole forces between highly excited electronic states. Those systems have driven most of the headlines about neutral-atom quantum computing for the past three years.

The April result is the first time the same threshold has been reached using the direct quantum statistics of the atoms themselves, with no Rydberg excitation required.

Why the second platform matters

Reaching the error-correction threshold in one hardware family is a milestone. Reaching it in a second, independent family is a different kind of milestone. It says the achievement is not specific to Rydberg physics. It says fault-tolerant quantum computing has at least two viable hardware paths, and they have different engineering trade-offs.

Rydberg-based gates are fast but require precise control of high-lying electronic states. Collisional fermionic gates are slower but use the cleaner physics of bare ground-state atoms. They have different error sources, different dominant decoherence channels, and different scalability profiles.

When two independent platforms can do the same thing through different physics, the field has stopped being dependent on a single technical assumption. That is the structural news here.

What's actually new

The methodological step is "no Rydberg required." For most of the past five years, the working assumption in neutral-atom quantum computing has been that high-fidelity two-qubit gates required Rydberg-state excitation. The April result invalidates that assumption.

It also opens the door to platforms that use direct on-site fermionic interaction — relevant for quantum simulation of strongly-correlated materials, where the natural physics is exactly the physics of interacting fermions. A neutral-atom quantum computer that natively simulates fermions at high fidelity is a different machine than one that emulates fermions on top of qubit-like states.

What this isn't

Not a fault-tolerant quantum computer. The gates have crossed the threshold; the machines have not. Logical qubits — the error-corrected units actually used in fault-tolerant computation — require thousands of physical qubits operated above threshold for sustained periods. That engineering still has years of work ahead of it.

Not the end of superconducting or trapped-ion quantum computing. Both architectures continue to improve. The story is not "neutral atoms won"; it is "the field is now genuinely multi-platform."

Not a commercial breakthrough. There is no near-term commercial product downstream of this result. The translation from hero experiment to fielded device is the multi-year engineering programme that quantum-computing companies are now running.

Stakeholder landscape

• Quantum-hardware physicists — the result expands the design space significantly; expect a wave of new architectures over the next 18 months.
• Quantum-error-correction theorists — every threshold crossing tightens the resource estimates for fault-tolerant computation. The new platform's noise profile feeds directly into that work.
• Quantum-computing companies — neutral-atom firms (Atom Computing, QuEra, Pasqal, Infleqtion) gain a second technical pathway; superconducting and trapped-ion firms are not displaced.
• National-security funders — fault-tolerant quantum computing's timeline is relevant to post-quantum cryptography migration policy. A second platform crossing threshold slightly compresses that timeline; nobody serious thinks it compresses it dramatically.

Cross-layer implications

• Cryptography policy — the NIST post-quantum migration deadlines (currently 2030–2035 for federal systems) are not changed by this result. They were set with multi-platform progress already assumed. But the result is the kind of evidence that justifies sticking to those deadlines, not extending them.
• Quantum simulation — the strongest near-term scientific application of these systems is not breaking encryption. It is simulating fermionic many-body physics — high-temperature superconductors, nuclear matter, strongly-correlated electron systems. Lithium-6 collisional platforms are unusually well suited for this.
• Workforce — the discipline now needs experimentalists who can work across at least three hardware families. Graduate programmes will have to broaden.

What this means for you

• If you build with quantum-secure cryptography — the 2030–2035 NIST migration deadlines remain the operating assumption. Audit your cryptographic inventory now if you have not.
• If you fund early-stage quantum — neutral-atom quantum computing is no longer a single-bet category. Diversification within the platform is now a real strategy.
• If you teach physics — the lithium-6 result is the cleanest pedagogical example available of "Pauli exclusion as a computational resource." Worth a lecture.

Uncertainty ledger

• The 99% figure is a single-gate fidelity. Sustained multi-gate sequences typically degrade significantly; the published experiments do not yet demonstrate long-circuit fidelity above threshold.
• The path to scaled-up logical qubits in fermionic-atom platforms is not yet engineered. Five-year horizon at minimum.
• Whether collisional gates can match Rydberg gates on speed at scale is open. Speed-quality trade-offs will determine which platform owns which application class.

Bottom Line

Two independent teams crossed the fault-tolerant quantum computing threshold in April using physics that nobody knew, five years ago, would get there. That is the core story. The deeper story is that quantum computing's hardware question is now plural — at least three platform families (superconducting, trapped-ion, neutral-atom in two flavours) are above threshold, and the field's progress no longer depends on any one of them succeeding. That is the structural condition every mature technology eventually reaches. Quantum computing has reached it in 2026.

Sources

• Nature, two-qubit fermionic-atom gate fidelity papers (20 April 2026) — Tier 1
• ScienceDaily, "Fermionic-atom quantum gates pass 99%" (April 2026) — Tier 2
• Phys.org, summary coverage and physicist commentary (April 2026) — Tier 2
• NIST, post-quantum cryptography migration timeline — Tier 1 (referenced)