Quantum Computing in the Global Energy Landscape

A four-month technoeconomic brief for Wood Mackenzie's research desk, asking a single question — where in the energy sector will quantum computing actually move the needle, and on what timeline? Commissioned as the SIPA 2025–26 Capstone and delivered as a final presentation at Wood Mackenzie's New York office on April 24, 2026.

The team

Six MPA / MIA candidates, advised by Prof. Jeanne Fox (SIPA), working with Ben on Wood Mackenzie's research side and Trent Yang (client lead — sustainability VC). The six of us:

• Akshay — project lead; hardware modalities, AI data centers, batteries
• Vincent Saputo — methodology and the technoeconomic matrix (he built the scoring spreadsheet the team scored every application against), enhanced solar / singlet fission
• Ruby Wu — project overview, nuclear, the global-policy and stakeholders chapter
• Terry Zhang — quantum-computing value chain, the case-studies survey, grid optimization
• Coco Guo — covers and table of contents, the QC 101 chapter, quantum sensing
• Jialong Wang — the Monte Carlo explainer, energy × finance — derivatives pricing and the application-matrix summary

The report is the team's work, and whatever is good in it is the team's. Credit is shared; mistakes are mine.

The question, narrowly

Quantum computing sits in an uncomfortable place in 2026. Fault-tolerant hardware is still five to ten years out. The news cycle oscillates between breakthrough and bubble. Wood Mackenzie's research clients — utilities, IOCs, developers, institutional investors — want a view that is neither.

So the brief was narrow:

For energy, which quantum applications have a credible technical case, a credible timeline, and a credible economic payoff — and which are still stories?

We scored each candidate application on a three-axis matrix:

1. Time to value — when will hardware support this? (1–5, from theoretical to within five years)
2. Quantum fit — does the problem's mathematical structure actually reward a quantum approach? (1–5, from no advantage to exponential speedup)
3. Cost-benefit of implementation — hardware access, integration effort, specialized software, licensing (1–5)

The score is a composite. The framework itself is the product: a reader can drop any new application into the rubric and get a defensible comparative read.

Six applications, one sentence each

Grid optimization — Unit Commitment. The most underrated near-term win. A hybrid quantum-classical schedule for 26 generators over 24 hours was solved on IonQ's 36-qubit Forte in 2025. Unit Commitment is run continuously, so even small-percentage gains compound; classical solvers just run combinatorics and accept the gap.

Battery chemistry. The clearest long-term case and the one the literature loves most — VQE-family algorithms map molecular Hamiltonians directly onto qubits without the DFT approximations that classical methods are stuck with. Ten to twenty years out for real materials; near-term value is in hybrid methods like QC-AFQMC that already improve on classical trial states.

Enhanced solar — singlet fission. The Shockley-Queisser limit caps single-junction panels near 29%. Singlet-fission materials could theoretically push toward 45%, but they're quantum-mechanically nasty to simulate. Seven to ten years out; if it lands, the LCOE implications are structural, not incremental.

Nuclear R&D. Not "quantum replaces HPC." Nuclear modelling is already HPC-bound — a single XGC1 fusion production run can consume ~1.5 million MPP-hours. Quantum is a candidate accelerator for specific kernels: optimization of fuel reloads, uncertainty quantification on Monte Carlo transport, and electronic-structure calculations for actinides that DFT handles badly. The value is faster iteration, not a different science.

Finance × energy. Quantum Amplitude Estimation can, under strong assumptions, reduce Monte Carlo sample complexity quadratically. That translates directly into derivatives pricing, VaR / CVaR, and portfolio optimization — all central to how energy markets and carbon markets clear. HSBC × IBM (2025) reported a 34% uplift on European corporate-bond trade prediction; JPMorgan × Quantinuum published a 100× improvement on a QAOA benchmark. Useful caveat: quantum here is a generic Monte Carlo speedup, not an energy-specific algorithm.

Quantum sensing. Different category. Not computing — hardware. NV-centre magnetometers, atom interferometers, gravity gradiometers. In O&G this reads on subsurface surveys and pipeline leak detection; in CCUS it reads on reservoir monitoring. Present-day deployments exist; the sector just hasn't priced them yet.

The four overarching findings

We earned these four the hard way, in the April 10 walkthrough with Ben and Trent. Every application reduces to them.

1. Accuracy drives timing. Error correction, not qubit count, is the bottleneck.
2. Economic benefit is a separate axis from technical fit. A problem can be a perfect quantum fit and still be economically boring — and vice versa. Don't collapse the two.
3. Quantum augments, not replaces, classical computing. Every credible near-term deployment is a hybrid workflow with classical HPC doing the heavy lifting and quantum accelerating a specific kernel.
4. Impact will arrive as a step change, not a drip. A particular threshold — logical-qubit count, gate fidelity, algorithmic improvement — unlocks an application, and until that threshold is crossed the curve is flat. Plan capital deployment accordingly.

Policy recommendations, in two lines

Don't invest as if quantum is already here. Invest so that, when it arrives, you haven't sunk capital into infrastructure the capability will devalue. Migrate to post-quantum cryptography now — the shelf-life of today's encryption is finite and adversaries are already harvesting ciphertext to decrypt later.

Method notes

• Six weeks of background reading (December–January), covering the McKinsey and BCG primers, the Qiskit Finance tutorials, and the IEEE / arXiv literature on QUBO formulations for grid and fusion problems.
• A dozen industry interviews, including Pete Shadbolt (PsiQuantum), Zach Rainey (Fujitsu), Max Wang, Kaizhao Wang, Russ Fein, Tyler Christeson, and Dr. Reichman — split two team members per interview, notes shared within 24 hours.
• An Excel scoring harness. Three dimensions, roughly four questions per dimension, five possible responses each — about a dozen inputs per application producing a composite score. The Excel is the single source of truth for the matrix.
• Two presentations at Wood Mackenzie's Hudson Yards office: midterm on March 13, final on April 24.

The deck

The final deck is included below as a B&W contact sheet. It runs forty-three slides: cover → glossary → project overview → QC 101 → methodology → the six applications with matrix placements → policy → conclusion → appendix.

The complete source deck, in full colour, is here: view original — Woodmac_SIPA_Deck_vF.pdf.

Disclaimer

This brief was prepared for Wood Mackenzie as a SIPA Capstone project and is shared here for academic and portfolio purposes only. It is not an investment recommendation, a solicitation to buy or sell any security, or professional financial or technical advice. Redistribution, republication, or further sharing of the deck or its contents — in whole or in part — is not permitted. Figures, scenarios, and scoring judgments reflect the team's reading of public literature and interviews as of April 2026 and may be revised without notice. Any errors are the team's; nothing on this page reflects the views of Wood Mackenzie or Columbia University.