A new quantum threat requires 200x fewer qubits than 2019 estimates, meaning RSA-2048 encryption could be cracked in just 9 minutes with near-term hardware.

200x Fewer Qubits: The Quantum Threat Is 9 Minutes Away

by
💡 Key Takeaways

  • The qubits needed to break RSA-2048 dropped from 20 million in 2019 to under 100,000 in 2026 — a 200x reduction driven by three papers in three months
  • Google Quantum AI demonstrated ECDLP-256 (elliptic curve) cracking with fewer than 500,000 qubits in roughly 9 minutes, estimating a 41% chance of deriving a Bitcoin private key before block confirmation
  • Google set a 2029 PQC migration deadline, NSA mandates quantum-safe systems by 2027, and NIST targets full adoption by 2035 — three clocks now ticking simultaneously

Twenty million qubits. That was the estimated cost of breaking RSA-2048 encryption with a quantum computer back in 2019. Three papers, published over three months, just slashed that number by 200x.

In May 2025, Google’s Craig Gidney brought the requirement below one million. By February 2026, startup Iceberg Quantum pushed it under 100,000. Then on March 31, Google Quantum AI published a paper showing elliptic curve cryptography could be cracked with fewer than 500,000 qubits in roughly nine minutes.

The quantum threat to encryption has shifted from a distant theoretical exercise to a timeline with hard deadlines. The FBI, NIST, and CISA designated 2026 as the “Year of Quantum Security” — and the research explains why.

Three Papers, Three Months, 200x Fewer Qubits

How many qubits does a quantum computer need to break modern encryption? The answer to that question has changed dramatically over the past 14 years, and the pace of change is accelerating.

  • 2012 — Fowler estimated hundreds of millions to roughly one billion physical qubits
  • 2019 — Gidney & Ekerå brought the estimate to approximately 20 million qubits
  • May 2025 — Gidney (Google Quantum AI) reduced it to fewer than one million qubits
  • February 2026 — Iceberg Quantum pushed it below 100,000 qubits

The first breakthrough came from Craig Gidney at Google Quantum AI in May 2025. His paper demonstrated that RSA-2048 could be factored using fewer than one million physical qubits with a runtime of less than one week. That represented a 20x reduction from the 2019 estimate of 20 million qubits — a single paper erasing an order of magnitude.

The second paper arrived in February 2026 from Iceberg Quantum. Instead of relying on surface codes — the standard error-correction approach — the team deployed QLDPC codes (Quantum Low-Density Parity-Check), a fundamentally different error-correction architecture. The result: RSA-2048 cracking with fewer than 100,000 physical qubits, a further 10x reduction beyond Gidney’s already dramatic improvement.

Iceberg Quantum closed a $6 million seed round and established partnerships with PsiQuantum, IonQ, and Oxford Ionics. But Gidney himself tempered expectations, stating that “without fundamentally changing the assumptions, another 10x reduction is unlikely.” The implication: the field may be approaching a physical floor for qubit efficiency.

Key point: The qubits required to break RSA-2048 dropped four orders of magnitude in 14 years — from hundreds of millions to under 100,000. Most of that collapse happened in the last three months.
Cinematic 3D render of a quantum computer core shattering digital encryption locks
Three papers in three months collapsed the qubit requirements for quantum hacking by 200x — Generated by OpenAI DALL-E 3

Cracked in 9 Minutes: Google’s ECDLP-256 Bombshell

The third paper hit on March 31, 2026, and it targeted a different class of encryption entirely. Google Quantum AI published an attack on ECDLP-256 — the elliptic curve cryptography that underpins blockchain transactions, digital signatures, and much of the internet’s trust infrastructure. The finding: fewer than 500,000 qubits could crack it in approximately nine minutes, a 20x reduction from the previous estimate of 9 million qubits.

The key here is how Google chose to disclose it. In an unprecedented move, the team used a zero-knowledge proof — a cryptographic technique that lets third parties verify the result without revealing the actual attack circuits. This approach blocks potential misuse while still proving the quantum threat is real and quantifiable.

The author list reads like a cross-section of the institutions most invested in this problem. Ryan Babbush, Google’s Director of Quantum Algorithms Research, and Hartmut Neven, VP of Engineering, were joined by Justin Drake of the Ethereum Foundation and Dan Boneh, a Stanford cryptography professor. Academia, Big Tech, and the blockchain world converging on a single paper to sound the alarm is itself a signal.

Complex internal structure of a quantum computer processor
Google Quantum AI showed that fewer than 500,000 qubits could crack elliptic curve cryptography — Photo by A Chosen Soul on Unsplash
In simpler terms: Cracking a traditional safe with a supercomputer would take decades. The quantum blueprint says it can be done in nine minutes. The quantum computer itself doesn’t exist yet — but the blueprint is getting sharper at an alarming rate.

AI Pulled the Trigger on Q-Day

The compression of the quantum threat timeline didn’t happen in a vacuum. Artificial intelligence played a confirmed role in accelerating the research. Researchers at Oratomic, a quantum computing startup, stated explicitly: “There is no question that we used AI to accelerate this development.”

The theoretical groundwork stacked up in a rapid sequence. In 2023, NYU’s Oded Regev published the first improvement to Shor’s algorithm — the foundational quantum factoring algorithm — in 30 years.

Follow-up research at the CRYPTO 2024 conference resolved key bottlenecks in Regev’s approach. Then in December 2024, Google’s Willow chip crossed the quantum error correction threshold, proving that adding more qubits could actually reduce error rates rather than increase them.

Why this matters: these breakthroughs aren’t isolated events. They form a reinforcing loop — algorithmic improvements enable AI-accelerated circuit design, and optimized circuits get validated by improving hardware. The speed at which Q-Day arrives — the day a quantum computer actually breaks production encryption — depends not on any single technology, but on how fast all three accelerate together.

💡 Related Posts
#1 on FLEURS at Half the Cost: Microsoft MAI Takes On OpenAI

Bitcoin’s 41% Problem: Cracked Before the Block Confirms

Google’s ECDLP-256 paper delivers its most visceral impact in the cryptocurrency space. Bitcoin’s average block confirmation time is 10 minutes — and the estimated quantum cracking time for a private key is roughly nine minutes.

The paper calculates a 41% probability that a primed quantum computer could derive a private key before a Bitcoin transaction is confirmed. That one-minute gap between quantum attack speed and blockchain confirmation speed is where the entire security model breaks down.

The research involved collaboration with Coinbase, the Stanford Blockchain Research Center, and the Ethereum Foundation. Justin Drake of the Ethereum Foundation called it “a monumentous day for quantum computing and cryptography.”

The reaction from blockchain investors was stark. Haseeb Qureshi of Dragonfly Capital warned that all blockchains need transition plans now, and Nic Carter of Castle Island Ventures called the findings very sobering — the industry is beginning to treat quantum risk as an operational reality, not a hypothetical.

A golden padlock on a keyboard symbolizing encryption security in the quantum threat era
With quantum cracking faster than Bitcoin’s block confirmation, cryptocurrency security faces its most concrete quantum threat yet — Photo by Towfiqu barbhuiya on Unsplash
Key point: The quantum cracking window is now shorter than Bitcoin’s block confirmation window. That gap transforms quantum risk from a theoretical concern into a design-level vulnerability for every blockchain relying on elliptic curve cryptography.

‘Harvest Now, Decrypt Later’ — The Threat Already Underway

Even though no quantum computer can break encryption today, the threat is already operational. Nation-state actors are collecting encrypted communications and data in bulk, stockpiling it for the day quantum decryption becomes feasible. Any data that needs to remain confidential into the 2030s is exposed to this “harvest now, decrypt later” strategy right now.

The urgency is reflected at the highest levels of government. The FBI, NIST, and CISA jointly designated 2026 as the “Year of Quantum Security.” The takeaway: waiting until quantum computers are operational to begin transitioning means the already-harvested data becomes retroactively compromised — and that window cannot be closed after the fact.

Google 2029, NSA 2027 — The Migration Deadlines

Google announced on March 25, 2026, through a statement by Heather Adkins, that it will migrate all systems — Android, Chrome, and Cloud — to post-quantum cryptography (PQC) by 2029. That deadline is six years ahead of NIST’s 2035 target, making it the most aggressive public commitment by any major technology company.

The U.S. government’s timeline is even tighter in some areas. Under the NSA’s CNSA 2.0 standard, all new national security system acquisitions must be quantum-safe by January 2027, with full migration across all existing systems required by 2033. NIST finalized the first three PQC standards in August 2024 — ML-KEM, ML-DSA, and SLH-DSA — and set a global full-adoption target of 2035.

Put simply, the attack timeline and the defense timeline are compressing simultaneously. Three deadlines now define the pace of the transition: the NSA’s January 2027 mandate for new systems, Google’s self-imposed 2029 conversion, and NIST’s 2035 global standard.

Key point: NSA 2027 (national security), Google 2029 (consumer services), NIST 2035 (global standard) — three post-quantum cryptography deadlines are now active, and the gap between the quantum threat and the defense buildout is narrowing fast.
💡 Related Posts
$30K Chip Prices: TSMC’s 4-Year Hike Ends Cheap Silicon Era

References

Stay Ahead of the Curve

Get the latest insights on AI, tech, and markets — delivered in both Korean and English.

Explore More Posts