Breaking RSA-2048: Why the Quantum Threat Isn’t Waiting for Hardware (A Real Analysis)

When the cybersecurity industry discusses "Q-Day" (Quantum Apocalypse), CISOs and security engineers often fixate on hardware manufacturers. The largest experimental quantum arrays have now reached thousands of physical qubits — including a 6,100-neutral-atom array demonstrated by Caltech — but these arrays are not yet fault-tolerant, universal quantum processors at that scale. This creates a false sense of security in many organizations: the assumption that "we need hundreds of thousands or millions of qubits, so we have decades left" is one of the most dangerous assumptions in our sector.

This approach overlooks the other — and much faster evolving — half of the equation: Mathematics and Algorithmic Optimization.

To understand how real the quantum threat is, we must look not at the hardware in laboratories, but at how algorithms are evolving and how much resource they actually require in real-world scenarios.

In this article, we will examine the concrete results of an RSA-2048 cracking simulation I performed in VS Code using Microsoft's Azure Quantum Resource Estimator (QRE), and discuss what these data-driven insights mean for enterprise risk management.

The Threat Has Shrunk 1000x (And It Is No Longer Just a Hardware Problem)

The answer to how many qubits are needed to break RSA-2048 encryption has changed dramatically over the last decade.

Looking at research trends and resource estimates, we can see quantum computing costs are in a free fall:

• 2012 (Fowler et al.): Early surface-code resource estimates put the requirement for RSA-2048 factoring at roughly 1 billion physical qubits under the assumptions used at the time.
• 2019 (Gidney-Ekerå): Algorithmic and arithmetic optimizations reduced the estimate to about 20 million noisy physical qubits for an eight-hour factoring target.
• 2025 (Gidney): A further reduction in algorithmic and error-correction overhead led to an estimate of less than 1 million noisy qubits to factor an RSA-2048 integer in under a week, under explicit hardware and surface-code assumptions.

The historical reduction from roughly 1 billion to below 1 million physical qubits represents about a 1,000-fold (3 orders of magnitude) improvement in resource estimates over 13 years. The reduction was driven primarily by algorithmic, arithmetic, error-correction, and architectural optimization rather than by a corresponding thousand-fold increase in physical hardware. The adversary's theoretical resource requirement is becoming smaller before the required machine has even been built.

The trend has continued. In 2026, newly published architectures have reported substantially lower theoretical requirements — for example, about 381,000 physical qubits with experimentally demonstrated grid coupling, and even lower figures under more speculative long-range-coupling and qLDPC assumptions. Another proposal reports a sub-100,000-qubit scenario under specific QLDPC and hardware assumptions. These are research estimates, not demonstrated capabilities, but they show that the resource boundary is still moving.

A Reality Check in VS Code: Speed vs. Resources

To move beyond the generalized "1 million qubits is enough" narrative, I decided to test the resource requirements myself. Using the Microsoft QDK (Quantum Development Kit) in VS Code, I ran a resource analysis for RSA-2048 using the Ekerå-Håstad factoring algorithm (an optimized variant of Shor's).

The results show that this is not just a "qubit count" problem, but a major engineering and architectural trade-off. These are resource-estimator scenarios under specified assumptions, not demonstrations that RSA-2048 can currently be cracked:

1. The 4-Hour Scenario: Under a nanosecond-scale gate-time configuration with Floquet error correction, the estimator projects an RSA-2048 factoring runtime of about 4 hours, requiring ~5.68 million physical qubits. This is a modeled resource scenario, not an existing capability. (qubit_maj_ns_e6, floquet_code)

2. Resource Optimization (A 2-Day Scenario): If the goal is to reduce qubit count, using Surface Code at the same modeled hardware speed reduces the requirement to ~3.77 million physical qubits. The projected runtime increases from 4 hours to about 2 days. In this model, the trade-off is qubit count versus runtime — not a change in the underlying cryptographic target. (qubit_maj_ns_e6, surface_code)

3. The Slow Hardware Trap (What CISOs Need to Know): One of the most striking findings in the table highlights the critical importance of hardware architecture. Even if enough qubits are produced, slower gate speeds can dominate runtime: in this modeled scenario, microsecond-scale operation leads to roughly 1 year of runtime despite using ~7.8 million physical qubits. (qubit_gate_us_e4, surface_code).

This data highlights the real engineering requirement: breaking RSA-2048 requires not just a large number of physical qubits, but sufficiently fast, high-fidelity, well-connected qubits together with effective fault-tolerant error correction.

The AI Multiplier: Why We Should Not Assume Another 13 Years

Looking at the data above and assuming that "it took 13 years to get here, so we can wait another 13 years" is a dangerous strategic mistake. Historical resource estimates have moved faster than a simple linear hardware-progress model would suggest.

A new variable is now entering the equation: advanced AI models.

Frontier AI models are increasingly capable of assisting with code generation, optimization, formal reasoning, and scientific workflows, including work relevant to quantum circuit design and error-correction research. However, the current evidence does not justify a precise claim that AI has made quantum-algorithm improvement exponential, nor does it justify a fixed "less than 3 years" Q-Day prediction. The more defensible conclusion is that AI could compress future optimization cycles and therefore adds uncertainty to an already-moving quantum risk timeline.

Why CISOs Must Act Today

Hardware constraints might temporarily protect us from waking up to a broken internet tomorrow morning. But this does not make us secure.

The immediate threat vector of quantum computing is not real-time attacks, but rather the "Harvest Now, Decrypt Later" (HNDL) strategy.

Encrypted data exfiltrated from your networks today (military communications, critical IP, long-term healthcare data) is being stockpiled by adversarial actors and nation-states. Once sufficient qubit counts, hardware speeds, and AI-backed novel algorithms are achieved, all of today's leaked data will be retrospectively exposed.

At this point, the action plan for CISOs is clear:

1. Acknowledge That the Timeline Is Uncertain: Quantum resource estimates are moving downward, while hardware progress and AI-assisted optimization add uncertainty. Organizations should not wait for a definitive Q-Day forecast before acting.
2. Inventory Your Cryptography: Identify where RSA and ECC are used within your organization, and determine which assets require long-term confidentiality (beyond Forward Secrecy).
3. Initiate Migration to NIST PQC Standards: Transitioning to PQC standards (ML-KEM and ML-DSA) is not a simple software update, but a complex re-architecting process that will take years.

The clearest lesson from my Microsoft QRE analysis is this: the resource boundary for breaking RSA-2048 is not fixed. Mathematics, algorithms, error correction, and architecture continue to move that boundary, while physical hardware remains the major implementation challenge. That uncertainty is exactly why migration should begin before a cryptographically relevant quantum computer exists.

Whether your organization remains secure in a post-quantum world depends not on when hardware engineers finish their jobs, but on when you initiate this migration process.

Selected Sources and Methodological Note

This article combines published quantum resource estimates with a Microsoft Azure Quantum Resource Estimator (QRE) analysis performed by the author. QRE outputs are scenario-based estimates conditional on hardware, gate-speed, error-rate, connectivity, and error-correction assumptions; they are not evidence that RSA-2048 can currently be factored.

Gidney, C. (2025), How to factor 2048 bit RSA integers with less than a million noisy qubits, arXiv:2505.15917.

Mundada et al. (2026), Heterogeneous architectures enable a 138x reduction in physical qubit requirements for fault-tolerant quantum computing under detailed accounting, arXiv:2604.06319.

Webster et al. (2026), The Pinnacle Architecture: Reducing the cost of breaking RSA-2048 to 100,000 physical qubits using quantum LDPC codes, arXiv:2602.11457.

Caltech (2025), Caltech Team Sets Record with 6,100-Qubit Array.

NIST, Post-Quantum Cryptography migration guidance and standards transition materials.

Best Regards.

Hakan Damar | Independent Researcher, 2026