Earlier this week, Honeywell announced plans to merge its quantum computing business, Honeywell Quantum Solutions (HQS), which focuses on trapped ion hardware, with the U.K.-based Cambridge Quantum Computing (CQC), which focuses on quantum software, to form a new, separate company. This follows Honeywell’s entry into commercial quantum computing development space three years ago and its announcement of its first “commercial” system (H-1) in 2020. The deal is expected to close in Q3 2021.
Ostensibly, the merger is intended to create a hardware-plus-full-stack quantum computing company, attracting more investors. In the official release, Honeywell chairman and CEO Darius Adamczyk, said: “Since we first announced Honeywell’s quantum business in 2018, we have heard from many investors who have been eager to invest directly in our leading technologies at the forefront of this exciting and dynamic industry – now, they will be able to do so. The new company will provide the best avenue for us to onboard new, diverse sources of capital at scale that will help drive rapid growth.”
When the deal closes, Honeywell will own 54 percent of the new company and Cambridge shareholders will own 46 percent. However, as Tony Uttley, currently president of the Honeywell quantum business and tapped for the same job at the new company, confirmed in a briefing with HPCwire, by this time next year Honeywell is unlikely to still be the majority owner.
Depending on your perspective, the merger plan is evidence of a potential quantum computing gold rush – at least in terms of a rush to raise capital and, to a lesser extent, to create vertically integrated (hardware and software) entities – or evidence of industry consolidation. It comes at a time when potential investors are sitting on huge hoards of cash (The Wall Street Journal reported yesterday that companies have been stockpiling cash throughout the pandemic and some banks are no longer taking deposits – “Some banks, awash in deposits, are encouraging corporate clients to spend the cash on their businesses or move it elsewhere”)[i] .
The quantum computing development community has been taking notice and seeking to capitalize.
IonQ, another ion trap-based quantum computing company announced plans to go public via the SPAC (Special Purpose Acquisition Company) route in March. Uttley wouldn’t rule out such a move for the new company. Given the promise of quantum computing, the rise of well-funded government quantum initiatives around the world, and cash-rich investors seeking opportunities, a boomlet in funding for quantum companies wouldn’t be surprising. That would be a change from just a few years ago when quantum computing’s technology hurdles and predicted years-long wait for payoff made investors wary. Recently, quantum technology advances and growing public sector attention are changing the quantum expectations landscape.
Veteran quantum watcher Paul Smith-Goodson, senior quantum & AI analyst, Moor Insights & Strategy, said the planned Honeywell-Cambridge merger is a positive step for both companies.
“The deal makes a lot of sense for both HQS and CQC. HQS was limited to only funding provided to it as a business unit of Honeywell International. To compete with the likes of IonQ, who will get $600 million in cash when it goes public, this path is necessary. Additionally, these quantum startups need long term funding to stay in the game for the next five to ten years. It will take that long to move from a few hundred qubits to a few million qubits needed for a fault-tolerant quantum machine suitable for complex computations in a production environment. And, the truth is we really don’t know how yet to do that. We have an idea of how long it will take, but we really don’t know with certainty,” Smith-Goodson told HPCwire.
“The new company will get somewhere around $300 million from Honeywell corporate, and that will help. More importantly, the merger sets up the new company in a position where it can decide if or when it goes public. I suspect that is a large part of the motivation for the merger. Both companies are doing well now, but as a new company it will do even better. After the merger, I expect HQS to release its model H2 quantum computer or increase the number of qubits and more zones in its current H1 model. Also at the same time, I expect CQC to announce a new commercial quantum product shortly after the merger is done. No NDA (confidential knowledge) here, I’m just connecting the dots,” said Smith-Goodson.
While acknowledging that attracting investment is a key objective, Uttley emphasized the end goal is to create and succeed as a quantum computing company. Honeywell and Cambridge have been collaborating for three years and Cambridge has been the single largest user of Honeywell’s quantum systems, said Uttley. Where Honeywell has focused on hardware, Cambridge expertise spans quantum algorithm and application development along with middleware and programming tools.
Uttley said: “We came to a collective realization that if you want to have the best company out there in quantum, the way to do it is to fully integrate, to have a full solution – applications, middleware, operating system, and hardware – that can be collectively designed to go after some of these really critical challenges that are going to be the hallmark of quantum computing.”
He said the merger/spinout approach is new for Honeywell. “This is unique in the history of Honeywell to go and do something like this, where we’ve taken a breakthrough growth initiative (HQS) and said, the right thing to do right at this moment is to actually put it into a position where we can combine it [with a complementary partner], turn it into its own company, set it up with a capital structure immediately by giving it $270 to $300 million of cash, and then allow it to onboard investments.”
Honeywell has set up a structure, said Uttley, to ensure it has continued, significant interaction with the new company. “For example, we have a long-term supply agreement to be able to develop the ion traps using Honeywell’s foundry [which is] part of Honeywell’s aerospace group [for] trap development. Honeywell is also one of our customers right now; we have projects [with Honeywell’s] performance materials, technologies group and Cambridge Quantum Computing using [our] quantum computer to do work in chemistry and novel compounds. We expect that Honeywell will continue to be the best customer of this new company.”
Currently, many qubit technologies (superconducting, ion trap, optical, topological, cold atom) are being developed. Honeywell has bet big on trapped ion tech, which has several advantages, including longer coherence times. Honeywell also sought to leverage its control systems expertise for use with quantum systems. The company embraced quantum volume (QV) – an IBM based quantum performance metric – and has touted its early systems QV ratings. Just this March, Honeywell reported its System Model H1 achieved a QV of 512, a fourfold jump over the QV 128 that H1 had demonstrated at its launch in September 2020, and the highest QV so far of an industry system yet.
Qubit count on the Honeywell systems remain relatively small (10 qubits on System H1) but other significant advances, such as mid-circuit measurement and re-use of qubits are notable. Using this capability, Honeywell developed what it calls the holoQUADS algorithm to run more complex simulations with fewer qubits.
“What holoQUADs was a demonstration of how we can take our trapped ion quantum computer – right now we have 10 qubits – and in this case, we use nine qubits to act like 32 qubits. The way we were able to do that is, using a circuit that would otherwise have required 32 qubits, there were times along the entire length of the computation where you stopped needing a particular qubit. There was nothing else you were going to go do with it. So you can measure it. But now since we’ve measured it, we can actually take that qubit and put it back into a circuit, and we have enough coherence time that we can keep on recycling after we measure them. That’s what this mid-circuit measurement and qubit reuse is all about. You measure the qubit and then can put it back into the circuit somewhere else.”
Cambridge has a wider focus in that its software strategy is to be hardware-agnostic and work with any qubit technology. Its centerpiece product is t|ket⟩, which CQC describes as an “architecture agnostic quantum software stack and ‘best in class’ compiler. t|ket⟩ translates machine independent algorithms into executable circuits, optimizing for physical qubit layout whilst reducing the number of required operations. t|ket⟩’s state of the art qubit scheduling and routing protocol ensures optimal results even in the Noisy Intermediate-Scale Quantum (NISQ) era.”
CQC founder lyas Khan will be CEO of the new company, which is expected to be named when the deal formally completes. CQC also works on algorithm development and recently published a paper on an algorithm to accelerate Monte Carlo calculations on quantum computers.
Uttley contends there are two powerful pieces to the merger story.
“The first is Cambridge, quantum computing is and will continue to be hardware agnostic. So as they do software development in chemistry and machine learning, cybersecurity, it is going to be across multiple platforms and multiple technologies. That will continue. The trapped ion hardware has very particular use cases for which it excels, such at the cybersecurity and chemistry. There are other use cases where other technologies are just adequate or better. One of the things Cambridge Quantum does is think through the logic of what a customer is trying to accomplish, and point where should the job be done. That’s actually part of the power of the platform to be able to say this job should be done on let’s say, IBM’s superconducting system, and this one needs to be done Honeywell,” he said.
“The second point is that if you want to build something that’s going to add value, quickly, it’s not about universal quantum computers. So universality is great [but still distant]. But if you can design architectures that are fit for the application, you can make them extremely efficient. We have the ability now to go from the trap all the way up to the operating system to the combination layer of the compiler and the application. And that doesn’t have to be ubiquitous across every application area; it absolutely won’t be, it’ll be designed to accelerate how fast quantum computing can have an impact in [particular applications] by years.”
To date, Honeywell and Cambridge have said little about who their early users are and what they are working on beyond work with JPMorgan on security issues. Both are already part Microsoft’s Azure Quantum portal. Uttley said they’ve had “dozens” of substantial engagements and cites demonstration projects for “BMW, DHL, and Samsung.”
Stay tuned for the company naming.