McKinsey's widely cited $600 billion quantum computing forecast for finance contains arithmetic errors, fails to separate quantum benefits from artificial intelligence gains, and compares future quantum systems against frozen 2026 technology rather than realistic 2035 alternatives. The consulting firm published these projections in its April 2026 Quantum Technology Monitor, and the numbers have since circulated through executive presentations and banking strategy documents without scrutiny of the underlying model.

What's Wrong With McKinsey's $600 Billion Number?

The core problem emerges immediately when you try to reproduce McKinsey's math. The firm displays seven financial sub-sectors with revenue baselines, impact percentages, and "affected share" labels that supposedly show what portion of each sector quantum computing would influence. However, when you multiply the baselines by the impact percentages, the arithmetic works out. But when you apply the affected-share percentages as multipliers, the total plummets to roughly $175 to $270 billion, less than half the headline figure.

McKinsey never explains which interpretation is correct. The affected-share percentages either are already embedded in the impact hypotheses (making them redundant), serve as decorative labels with no computational role, or represent a missing step in the published model. This reproducibility failure alone should give financial executives pause before committing nine-figure budgets based on the projection.

How to Evaluate Quantum Computing Claims for Your Organization?

• Demand Reproducible Math: Any projection worth acting on should allow you to verify the arithmetic from published inputs. If multiplying the displayed numbers produces a different result than the headline, the model is not transparent enough for strategic decisions.
• Separate Quantum From AI: Ask whether value estimates combine quantum computing with artificial intelligence improvements. McKinsey's finance chapter attributes the entire $400-600 billion to quantum in its headline, but a footnote reveals the estimates measure "combined impact of quantum computing and AI." The quantum-specific share remains unknown.
• Check the Baseline Comparison: Verify what technology the projections compare against. A 2035 quantum benefit measured against 2026 technology ignores nine years of GPU advances, AI surrogate models, and classical optimization improvements that will narrow any quantum advantage.
• Review Peer-Reviewed Evidence: Look for citations to published research on hardware requirements and algorithmic performance. McKinsey's finance chapter includes no peer-reviewed resource estimates, hardware benchmarks, or algorithmic papers in its source lines.

Where Does the AI-Quantum Confusion Come From?

The entanglement of quantum and AI benefits runs through every use case McKinsey describes. Corporate banking value comes from improving default risk prediction for real-time product decisions. Retail banking value comes from faster credit scoring at scale and personalized finance recommendations. Asset and wealth management value includes personalized pricing and recommendations explicitly tied to AI.

All of these workloads already run on classical machine learning systems in production today. Quantum machine learning research exists, but no peer-reviewed result establishes a scalable quantum advantage for any of these tasks. McKinsey provides no allocation of the projected value between quantum computing, AI, and ordinary technology modernization, leaving the quantum-specific share unknowable from the report alone.

The Baseline Problem: Comparing Against Yesterday's Technology

McKinsey's only disclosed statement about the classical comparator appears in a single footnote: "Savings are vs situation today, not vs future technology alternatives." This caveat applies only to the savings lines, which represent just $1 to $2 billion of the total. The other six segments, comprising more than 99 percent of the $400-600 billion, are labeled as revenue increases with no comparator disclosed at all.

Where a baseline is stated, it is frozen at 2026 technology. A 2035 benefit measured against today's Monte Carlo stack assumes away nine years of GPU generations, AI surrogate models, and variance-reduction advances. This "moving classical baseline" erodes quadratic quantum speedups faster than any other factor; a modest advantage over today's technology can become no advantage at all over the 2035 stack.

Where the baseline is unstated, which is nearly everywhere in the finance chapter, readers cannot evaluate the claim at all. An impact percentage with no counterfactual is an opinion with a decimal point, not a forecast grounded in evidence.

What McKinsey Got Right (And What It Didn't)

To be fair, McKinsey does label its value estimates as approximate rather than definitive. The firm also includes a caveat that quantum computing's incremental impact overlaps with generative AI, so the totals are not fully additive. These caveats are accurate and important. The problem is that they appear in small type while the $600 billion figure dominates headlines and client presentations without them.

The Monitor's broader market data and industry overview contain useful material. But the finance chapter deserves separate scrutiny because finance is the sector where the distance between headline and evidence is widest, and because financial institutions are the audience most likely to misallocate real budgets in response to an unexamined projection.

For CISOs, CFOs, and technology leaders evaluating quantum computing investments, the lesson is clear: demand transparency. McKinsey's model should be published in full, with every assumption and calculation step disclosed. Until then, the $400-600 billion range should be treated as a starting point for conversation, not a forecast ready for budget allocation.