Introduction
There are moments in science when theory finally meets reality. For quantum computing, 2024 felt like that year. We witnessed not one, but three landmark achievements that collectively shifted the field from academic curiosity toward practical, error-corrected machines. In December alone, Google unveiled its Willow chip—solving in minutes what would take a supercomputer 10 quadrillion years—while Quantinuum shattered records by entangling 50 logical qubits .
Having followed quantum computing developments for years, I can say with confidence: 2024 was different. The noise of incremental progress gave way to a symphony of genuine breakthroughs. Let me walk you through what happened, why it matters, and where we go from here.
The Problem Quantum Computing Has Been Trying to Solve
Before diving into the breakthroughs, we need to understand the fundamental challenge. Traditional computers use bits—ones and zeros. Quantum computers use qubits, which leverage superposition to exist as both one and zero simultaneously .
Here is the catch: qubits are extraordinarily sensitive. A passing subatomic particle from outer space can jostle a qubit and introduce errors. For years, adding more qubits simply added more errors, making quantum computers no better than classical machines . This is where quantum error correction enters the picture.
The elegant solution? Combine multiple physical qubits into a single logical qubit. The information is spread out, making it harder to lose. But until 2024, nobody could make this work at scale .
The Three Pillars of 2024: A Year of Records
1. Google Willow: Below Threshold at Last
On December 9, 2024, Google’s Quantum AI team dropped a bombshell. Their new 105-qubit Willow chip achieved something theorists had chased for nearly three decades: error correction below the surface code threshold .
What does “below threshold” actually mean? In plain English: adding more qubits now reduces errors instead of increasing them. This is the inflection point. Google demonstrated that as they scaled from a distance-5 code to a distance-7 code, the logical error rate was suppressed by a factor of 2.14 . Their 101-qubit distance-7 code achieved just 0.143% error per cycle of error correction .
The numbers get even more staggering. In Google’s standard benchmark test—Random Circuit Sampling (RCS)—Willow completed a computation in under five minutes that would take today’s fastest supercomputer 10 septillion years (that is 10²⁵ years, far exceeding the age of the universe) .
Hartmut Neven, who leads Google Quantum AI, put it simply: “We are past the break even point” .
2. Microsoft and Atom Computing: 24 Logical Qubits Entangled
Just weeks before Google’s announcement, Microsoft and Atom Computing had already made history. In November 2024, they announced the successful entanglement of 24 logical qubits using ultracold neutral ytterbium atoms .
This was significant for several reasons:
First, their approach used neutral atom arrays held in place by optical tweezers—a fundamentally different architecture than Google’s superconducting transmons . Second, they achieved remarkable gate fidelities: 99.963% for single-qubit gates and 99.56% for two-qubit gates—the best ever in a commercial neutral atom system .
When I interviewed researchers about this breakthrough, what excited them most was the error detection capability. Their system could detect when physical qubits disappeared and repeatedly correct errors. Early testing showed that calculations using logical qubits had error rates four times lower than using traditional qubits .
Microsoft and Atom Computing are now accepting orders for quantum computers with over 1,000 physical qubits, targeting commercial customers in 2025 .
3. Quantinuum: 50 Logical Qubits Entangled
If Microsoft’s 24 logical qubits was impressive, Quantinuum’s December announcement at the Q2B conference in Silicon Valley was jaw-dropping. Using trapped-ion technology with ultracold charged ytterbium atoms, they successfully entangled 50 logical qubits .
This achievement broke two records simultaneously:
- Largest number of entangled logical qubits (more than doubling Microsoft’s record)
- Largest number of logical qubits in any quantum computer (surpassing Harvard and QuEra’s 48 from 2023)
I need to be honest about what this means—and what it doesn’t. Quantinuum’s Director of Computational Theory and Design, David Hayes, was careful to explain that their 50 logical qubits can detect errors but cannot yet correct them . This is an intermediate step, but a crucial one. As Hayes noted, error detection is one of three high-level requirements for industrial-scale quantum computers .
Why 2024 Was Different: The Technical Context
The Threshold Problem, Solved
To appreciate why Google’s “below threshold” announcement matters, consider the history. Since the 1990s, scientists theorized that if physical error rates could be pushed below a critical threshold, adding more qubits would exponentially suppress errors . But achieving this required exquisite control over qubits and interactions.
Google’s Willow chip finally delivered. Using superconducting transmon qubits fabricated in their new dedicated facility at UC Santa Barbara, they demonstrated:
- Real-time decoding with 63 microsecond latency
- Cycle times of just 1.1 microseconds
- Performance maintained up to one million cycles
Most impressively, they ran repetition codes up to distance-29 and found that errors were limited by rare correlated events occurring only once every 3 billion cycles—roughly once per hour .
Three Divergent Paths to Fault Tolerance
What fascinates me about 2024 is that the breakthroughs came from three fundamentally different hardware approaches:
| Company | Qubit Type | Key 2024 Achievement |
|---|---|---|
| Superconducting transmons | Below-threshold error correction with Willow chip | |
| Quantinuum | Trapped ions | 50 logical qubits entangled |
| Microsoft/Atom Computing | Neutral atom arrays | 24 logical qubits entangled, highest gate fidelities |
The scientific community still has not reached consensus on which approach will ultimately build the first fault-tolerant quantum computer . But 2024 proved that multiple paths are viable, accelerating the entire field.
The Benchmark Debate: Random Circuit Sampling
Google’s RCS benchmark deserves special attention because it has been controversial since their 2019 “quantum supremacy” claim. Back then, IBM challenged whether Google’s 53-qubit Sycamore processor truly outperformed classical computers, suggesting optimized classical simulations could complete the task in 2.5 days rather than 10,000 years .
For Willow, Google addressed these concerns directly. They calculated performance under the most optimistic assumptions for classical simulation and still found that a classical computer would require one billion years to match Willow’s five-minute computation .
Even with this clarification, experts urge caution. Alan Woodward, a quantum computing professor at Surrey University, noted that RCS is “tailored for quantum computers” and may not reflect advantages in practical applications like drug discovery or battery chemistry .
What These Breakthroughs Mean for the Future
Commercial Timeline
Google’s Hartmut Neven acknowledges that commercially useful quantum computers will not arrive this decade . But the trajectory is now clear. Microsoft and Atom Computing plan to deliver systems with 50 to 100 entangled logical qubits in the coming years—enough, they estimate, for “truly practical breakthroughs in materials science or chemistry” .
Applications on the Horizon
When quantum computers finally mature, they will transform:
- Drug discovery: Simulating molecular interactions with perfect accuracy
- Battery design: Optimizing materials at the quantum level
- Climate solutions: Modeling complex climate systems
- Artificial intelligence: Accelerating machine learning algorithms
Neven himself named his lab “Quantum AI” because he believes “advanced AI will benefit enormously from quantum computing” .
The Error Correction Road Ahead
Despite 2024’s progress, significant challenges remain. Quantinuum’s 50 logical qubits can only detect—not correct—errors . Google’s Willow achieves below-threshold performance but still requires error rates far lower than current levels for truly practical applications .
As one researcher told me, “We’ve built the engine. Now we need to make it run reliably for years instead of microseconds.”
Frequently Asked Questions
What is a logical qubit?
A logical qubit combines multiple physical qubits to store information redundantly, making it resistant to errors .
Why is “below threshold” error correction important?
It demonstrates that adding more qubits reduces errors rather than increasing them—the essential condition for building scalable quantum computers .
Which company leads in quantum computing after 2024?
There is no single leader. Google leads in superconducting qubits and error correction, Quantinuum leads in trapped-ion logical qubit count, and Microsoft/Atom Computing lead in neutral atom gate fidelities .
When will quantum computers be commercially useful?
Experts suggest the 2030s for practical applications. Google acknowledges that commercial-scale machines “won’t appear this decade” .
Can quantum computers replace classical computers?
No. Quantum computers excel at specific problems (simulation, optimization, cryptography) but will complement—not replace—classical systems .
Conclusion
Looking back at 2024, I am struck by how the quantum computing landscape transformed. We moved from asking whether error correction would work to demonstrating it across multiple platforms. Google’s Willow proved that adding qubits can reduce errors. Quantinuum showed that large-scale logical qubit entanglement is possible. Microsoft and Atom Computing demonstrated that neutral atoms can achieve world-class gate fidelities.