Quantum computing leaders see early uses before large-scale commercial value

Mark Bergen asks what a future quantum computer could do that current supercomputers cannot. Carmen Palacios-Berraquero offers breaking public-key encryption as an example of a potential exponential advantage, though it is not the application that excites her.

She stresses that her comparison is still being calculated: matching a proposed five-hour quantum attack using conventional data centers might require an area equivalent to twice Earth’s land surface. She places a machine capable of such work around 2033–35, while urging listeners to treat the estimate cautiously.

Jan Goetz distinguishes problems solved once, such as predicting the properties of a new material, from recurring optimization tasks. He cites work with a German rail operator on rescheduling trains after disruptions: even a limited number of stations and trains creates an enormous number of combinations.

Goetz expects a hybrid system rather than quantum computers replacing supercomputers. His company puts quantum hardware in computing centers alongside CPUs and GPUs, connecting the machines so different parts of a problem can be allocated to each.

Ebba Carbonnier says useful experiments require logical qubits built from physical qubits, and points to 50–60 logical qubits as a promising stage. She wants to simulate molecules and electrons for drug discovery, including photodynamic cancer therapy, as well as chemistry and materials for more efficient fuel cells.

Carbonnier distinguishes the intermediate stage from the fault-tolerant machines the industry ultimately seeks. Asked whether a drug will be discovered within five years, she says there is no timeline; QuNorth expects a 50-logical-qubit system the following year and is working with industries on smaller pieces of larger problems.

The approach begins with the user’s problem, she says. Teams then break it into subproblems and learn which algorithms are suitable or still need to be developed, because the complete applications require more powerful hardware.

Bergen asks why National Grid Partners invested in Palacios-Berraquero’s company. She says grid simulation is already difficult and will become more complex as power networks connect many sources, including solar panels, wind farms, batteries and users.

For grid applications, Palacios-Berraquero points to fault-tolerant systems around 2033 or later. She describes a nearer goal of early quantum advantage in one or two years: a problem that classical machines cannot solve and that has some real use, even a small academic one.

Commercial utility is a higher threshold: solving an industry problem with a traceable economic benefit. She says US and UK government programs have used roughly 2033 as a horizon for that goal, though neither date is a guarantee.

Goetz says building a quantum industry requires both supply chains and demand. He sees the business as global but wants Europe to strengthen its own capabilities while deciding which international suppliers are trusted partners.

He credits European public procurement through EuroHPC with placing small quantum machines in supercomputing centers, giving European suppliers revenue and helping attract private investment. Longer-term European programs now target later commercial challenges.

Asked whether US partners remain trustworthy, Goetz says his company works with American suppliers and hopes transatlantic collaboration continues. Carbonnier says the field’s long history of research and commercial cooperation makes the prospect of a single country withholding the technology less of a personal concern.

Bergen asks Carbonnier why QuNorth works with Microsoft after controversy over some of its quantum research. She distinguishes Microsoft’s own topological hardware research from QuNorth’s machine: its neutral-atom hardware comes from Atom Computing, while the software stack is developed with Atom and Microsoft.

Carbonnier says quantum hardware still has particular error and noise problems, so hardware, error correction, compilation, algorithms and applications must be developed together. She regards the collaboration as a way to work across that whole stack.

On “Q-day,” the point at which a quantum machine could break vulnerable public-key cryptography, Palacios-Berraquero says an annual expert survey has long centered around 2033. Unexpected advances in hardware, software or algorithms could bring it forward, she adds.

Goetz says some cryptocurrencies are vulnerable to quantum attacks to varying degrees, while other encryption methods are quantum-safe. He argues that the immediate task is to migrate underlying cryptographic schemes, noting that government procurement is already pushing organizations toward transition plans.

The panel distinguishes engineering roadmaps from algorithmic breakthroughs that can reduce the required hardware. Goetz expects AI to contribute to further algorithm improvements, potentially allowing useful work on smaller machines.

Carbonnier says quantum development kits can use AI assistance to help select relevant algorithms for tasks such as molecular simulation. She compares this to AI-supported development of conventional software, while noting that the library of quantum algorithms remains small.

Asked about AI developing more powerful quantum algorithms, Goetz welcomes the possibility because greater efficiency could make existing hardware useful earlier. Palacios-Berraquero adds that designing an algorithm does not provide the hardware needed to run it; AI can also help calibrate fragile qubits.

An audience member asks who might become the Nvidia of quantum computing. Goetz describes collaboration in which Nvidia works on GPUs while his company develops quantum hardware and their systems work together, including for calibration and error correction.

Goetz expects a mix of players, potentially including quantum foundries and design houses, but says roles are still fluid. Asked about China, the panelists describe its quantum computing progress as very advanced and fast-moving.

On protecting intellectual property, the speakers say their companies take security seriously, train staff and use available government support. Bergen then asks about public markets after IQM’s listing.

Palacios-Berraquero says Nu Quantum may eventually turn to public markets, but for now private capital can fund large rounds and support its position in quantum networking. Bergen closes by noting that the discussion has only begun to cover the field.