Most quantum computing companies are pursuing gate-based digital quantum systems. These machines manipulate qubits through long sequences of quantum logic gates, similar in concept to how conventional processors execute instructions. But every interaction with a qubit risks introducing noise and decoherence, forcing digital systems to rely heavily on error correction and large numbers of redundant physical qubits.

A small group of quantum computing companies is pursuing an alternative architecture known as analog quantum computing, arguing that it may reach useful applications sooner than many gate-based quantum systems. Among them is Qilimanjaro Quantum Tech, a Barcelona-based startup.

“We believe in this vision of multimodality,” Marta Estarellas, CEO of Qilimanjaro, told EE Times. “If you have a problem that is intrinsically continuous, like chemistry or electrodynamics in materials, analog is more suitable. If you have a more structured mathematical problem, such as in cryptography, use a digital approach.”

Qilimanjaro is one of the few firms pursuing analog quantum computing. According to Olivier Ezratty, a quantum computing academic and co-founder of the Quantum Energy Initiative, other companies in this category include D-Wave, Pasqal, and QuEra.

Reducing the gate problem

Qilimanjaro’s central argument is that analog quantum systems may be able to reduce the accumulation of errors by avoiding many of the gate operations required in digital quantum computing.

“There’s indeed no such thing as gate operations in quantum annealing and analog quantum simulation,” Ezratty told EE Times.

In digital quantum computers, each gate operation requires interacting with fragile qubits using fast microwave pulses or other control mechanisms. Those interactions can disturb the system and create errors that compound throughout a computation.

Qilimanjaro instead uses analog quantum evolution. Rather than repeatedly manipulating qubits through discrete gate sequences, the system prepares a quantum state and allows it to evolve more naturally toward a minimum-energy solution.

“You prepare your quantum chip, and you let it go,” Victor Canivell, chairman and co-founder of Qilimanjaro, said. “Because you don’t interact with it all the time, you create far fewer errors.”

The company uses superconducting qubits—similar to those used by IBM and Google—but of a slightly different type, the fluxonium qubit, which allows for analog control on top of digital.

Qilimanjaro believes this could make analog systems particularly useful for problems involving continuous physical processes, including chemistry, materials science, optimization, and, potentially, AI training.

Industrial companies are already exploring some of those use cases. Repsol, a Spanish multinational energy company headquartered in Madrid, sees analog quantum systems as potentially useful for large-scale optimization and energy system design. The company is now working with Qilimanjaro to find out more.

“At Repsol, we are increasingly facing industrial challenges where complexity grows faster than compute, particularly in optimization problems,” Emilia Martínez, CTO of Repsol, told EE Times. “Quantum computing offers a fundamentally different way to explore these complex solution spaces.”

AI becomes a new target

Although quantum computing has long been associated with chemistry simulation and optimization, Qilimanjaro increasingly sees AI as one of the most promising future applications for analog quantum systems.

The transition comes as hyperscalers and AI companies confront growing power consumption challenges associated with large-scale AI training and inference.

“We are seeing now a huge interest in quantum computing that we never saw before,” Estarellas said. “The digital world is realizing that we need a more sustainable way of doing computation.”

Qilimanjaro believes analog quantum systems could eventually become specialized accelerators inside future AI infrastructure.

The company is particularly focused on “quantum reservoir” approaches to AI training. These systems use the complex dynamics of quantum systems themselves as computational resources for machine-learning tasks, particularly for the prediction of time series.

According to Estarellas, recent theoretical papers from outside research groups have strengthened interest in the concept. “This can become the biggest revolution since Nvidia exploded with GPUs,” she said.

The company argues that quantum systems may eventually support larger and more complex neural networks than classical systems can efficiently simulate. “As the quantum systems grow bigger and bigger, the complexity of the neural network that we can encode grows exponentially with the number of qubits,” Estarellas said.

Still, Qilimanjaro acknowledges that practical quantum advantage in AI remains unproven. “We still need to run those algorithms on real quantum hardware, and that is why we are building our next-gen chips,” Estarellas said.

Ezratty also cautioned that proving meaningful quantum advantage remains difficult across the analog quantum sector.

Building a hybrid quantum infrastructure

Qilimanjaro positions its systems not as standalone replacements for classical computing but as components within future hybrid infrastructures that combine HPC, AI accelerators, and multiple types of quantum hardware.

The company has deployed systems at the Barcelona Supercomputing Center (BSC), where researchers are experimenting with hybrid HPC-quantum workflows.

According to BSC senior researcher Alba Cervera-Lierta, the center is integrating both analog and digital quantum systems into a dedicated partition of the MareNostrum 5 supercomputer called “MareNostrum ONA.” Researchers are exploring approaches that distribute workloads across classical HPC systems and quantum computers.

Cervera-Lierta also noted that quantum simulation is already emerging as a promising application area for hybrid HPC-quantum systems because some physics simulations stretch beyond the capabilities of classical supercomputers.

Qilimanjaro sees such deployments as early models for future data centers. “The future of the data center will be traditional computing, AI computing, and quantum computing together,” Canivell said.

Barcelona Supercomputing Center’s Alba Cervera-Lierta standing under the first quantum computer in Spain, developed with 100% European technology. The quantum computer is integrated with the MareNostrum 5 supercomputer at BSC. (Source: Mario Ejarque/Barcelona Supercomputing Center)

Scaling challenges remain

Despite the optimism surrounding analog quantum systems, major technical challenges remain.

Qilimanjaro currently operates prototypes with 15 analog qubits and is developing systems with roughly 50 analog qubits. The company believes that useful “quantum utility” or business advantage in specific applications could emerge within two to five years.

“I think this will happen within a range of two to five years, probably in the lower range,” Estarellas said.

Quantum Energy Initiative’s Olivier Ezratty (Source: Franck Disegni)

But commercial-scale systems remain far more difficult.

Superconducting quantum systems require extreme cryogenic cooling, sophisticated control electronics, and dense cabling infrastructure. Scaling those systems introduces fabrication, packaging, and noise challenges.

“The fabrication process is still very artisanal,” Estarellas said. “You make human mistakes. The yield of those devices is not always very good, and we must start automating those processes.”

Cervera-Lierta also pointed to broader industry hurdles surrounding fault tolerance and scaling. She said large-scale practical quantum computing may eventually require networking multiple quantum systems together because future machines may not fit within a single dilution refrigerator.

Ezratty added that analog quantum systems face their own scaling limits, particularly around coherence time and the physical behavior of large quantum systems.

Even so, Qilimanjaro argues that analog quantum systems may provide a nearer-term route to practical applications than fully fault-tolerant, gate-based quantum computers.

The company said it deliberately positions itself against some of the hype surrounding quantum computing.

“We decided to follow the anti-hype approach,” Estarellas said. “Every claim, every presentation we made, was based on scientific grounds.”

That strategy may help the company stand out as the quantum industry increasingly shifts attention from long-term theoretical potential toward practical utility, infrastructure integration, and AI-era computing demands.

Read also: