When I last wrote about D-Wave (NYSE:QBTS) in March, its shares were priced at $1.87, and I expressed my concerns that commercial adoption was not sufficient to drive the company towards profitability. I said I needed to see more evidence of commercial traction before I invest. It was clear at that point that the D-Wave annealing machine could solve very hard optimization problems, and that some companies were beginning to explore its potential.
At the recent QUBITS conference in Boston (June 17th – 18th) it became clear that D-Wave is gaining some real traction among a growing customer base and that the newly released D-Wave Advantage2 architecture has delivered true Quantum Supremacy. (Some of the talks from the QUBITS conference can now be viewed on the D-Wave YouTube Channel, if you are interested, I would advise watching the CEO and his keynote address as it summarizes the whole conference.)
In my opinion D-Wave, Microsoft (MSFT) and QuEra appear to be the three companies pulling ahead in the Quantum race, I already own MSFT I cannot buy into QuEra as they are still private. After the conference, I wanted to add D-Wave to my portfolio, so I conducted another deep dive, deciding to continue watching and waiting.
Quantum Supremacy
When I wrote about IONQ (IONQ) in April, they took issue with my use of the terms Quantum Supremacy and Quantum Advantage, saying in an email to me (covered as a pinned comment to that article) that they prefer the term Commercial Advantage due to the many fallacies used with other terms.
The original definition of Quantum Supremacy was proposed by John Preskill in 2012, he said:
Quantum Supremacy or Quantum Advantage is demonstrating that a programmable quantum computer can solve a problem that no classical computer can solve in a feasible amount of time.
That definition is easily understood, but it sets a very high benchmark, a benchmark so high that it has proved impossible to clear. As a result, companies have tried to lower the benchmark to one that can be achieved.
Alphabet (GOOGL) published a claim that it had achieved supremacy, which was quickly disproved, and International Business Machines (IBM) later disproved the same claim from Google. I covered these claims in My Microsoft article Preparing for a quantum leap.
IBM lowered its benchmark to Quantum Utility in Nov 2023: defining Quantum Utility as “quantum computation that provides reliable, accurate solutions to problems that are beyond the reach of brute force classical computing methods,”
It is, as IONQ described, a fallacy. “Brute Force” would never be used to solve a difficult problem, brute force is just trying many answers to see if one works. Classical computers would attempt to use a heuristic or neural network, not simply a brute force attack.
IONQ and its commercial advantage, “the point at which quantum computers begin to solve problems classical computers cannot” is another such fallacy, begin to solve means nothing.
The original benchmark is high; however, D-Wave has stayed with it and passed it. They have delivered Quantum Supremacy on a useful problem. D-Wave published a paper, Computational Supremacy in quantum simulation, in March this year. I have read it, but its 55 pages go beyond my knowledge in many areas. In summary, the paper describes how a specific problem, the simulation of the dynamics of a magnetic spin system, a meaningful and valuable requirement to model quantum phase transition, was solved using D-Waves annealing computer.
The paper also explains D-Wave’s attempts to solve the problem on classical computers using various techniques, concluding that no known method can achieve the same results in a reasonable time frame.
The D-Wave CEO confirmed in his QUBITS keynote speech that no challenges to the paper had been received. The previous claims of quantum supremacy from Google and IBM received their challenges, proving they were wrong in days, and it has now been over three months since D-Wave published. The D-Wave CEO said in his speech that the D-Wave annealer took 20 minutes to solve this problem and that the world’s largest supercomputer, Frontier, would take more than one million years to do it.
For those of a more technical mind, I should point out that the paper also shows that D-Wave has violated Bells inequality, a necessary condition of supremacy that no other system has so far achieved.
Further evidence of Quantum Supremacy came from YapiKredi Technology (a Turkish software company with 1,700 staff focused on developing financial and banking products). At the QUBITS conference, they described a financial crash forecasting and estimation package they designed for their banking partner. The package includes 4,500 firms and reaches its solution in seven seconds. The representative from YapiKredi said that their previous research and attempts to solve this problem had suggested that a package with only 30 firms would have taken thirteen billion years to solve on their classical computers and the biggest problem they had was explaining the speed-up to their banking partner.
Technological progress
D-Wave appears to be the only company designing a quantum annealer. This year, they announced three important upgrades to their offering.
Quantum Annealing: A recap
I covered this tech in detail in my first article on D-Wave Quantum Computing Might Be With Us, but to recap; a quantum annealer uses quantum properties to solve optimization problems. The machine consists of superconducting qubits (a small loop of superconducting wire connected by a Josephson junction). The qubits will naturally exist in their lowest energy state. A magnetic field is applied to the qubit, controlled by the machine, and the qubit will move to its lowest energy state under the influence of the magnetic field. The qubits are linked together by coupling devices that utilize the quantum property of entanglement; linked qubits will move to the lowest energy state for the entire linked system rather than the lowest state for individual qubits.
To solve a problem, the computer couples qubits together and applies different magnetic fields to each of them, representing the constraints on the problem being solved. (in its simplest case, the magnetic fields might represent the distance between two drop-off points for a delivery driver, the longer the distance between the drop-offs, the higher the magnetic field). When the system has applied the magnetic fields, the computer runs “the anneal,” and the qubits will fall to the lowest energy state for the entire system. The value of each qubit can then be measured and would give the optimal solution to the problem (in the delivery driver case, the lowest number of miles to complete all the drops).
Hardware Progress
The new Advantage2 processor is a significant technological advancement. It has 1,200 qubits with a new architecture and fabrication process. At 1,200 qubits Advantage2 is more powerful than the previous 5,000 qubit Advantage processor. A 4,800 qubit Advantage 2 processor has already been fabricated and should be on the D-Wave Leap cloud offering later this year. Advantage2 has double the coherence time of the previous model, increased connectivity and a significantly higher energy scale.
The energy scale is important as it is the basis of how the system reaches its solutions. A higher energy scale means higher quality solutions. Two Advantage2 computers are already running in the Leap Cloud, and the Advantage2 was the processor used to deliver the supremacy result I discussed. The previous Advantage could not deliver the answer, despite D-Waves efforts.
Software Upgrades
A new annealing protocol called Fast Anneal was released this year, and the longer coherence times of Advantage2 means the anneal now takes place within the coherence time of the processor. (Anneal is the term for finding the answer)
A new hybrid solver, again released this year, allows nonlinear relationships between variables, leading to previously intractable problems being solved.
With the new hardware and software upgrades, the gap between D-Wave and the rest of the quantum industry has grown. They have a large-scale quantum computer that is reliable and accurate enough to solve complex problems today. It excels at solving optimization problems that have real value for many large businesses, and it appears that these businesses are taking note.
Commercial Traction
From my last article to the QUBITS conference, the number of commercial customers has grown from 60 to 80 and the number of problems solved on D-Wave computers jumped from 50 million to over 100 million. Several companies discussed their experience using D-Wave at the conference.
Pattison Food Group
Pattison Food Group, part of the Jim Pattison conglomerate, one of Canada’s largest, has 11 retail brands and operates 300 food and drug retail locations. They have four wholesale business locations, five food and drug production facilities, a large set of own brand labels and a delivery service.
The executive from Pattison explained that they had spent more than a decade working with computing and software suppliers and failed to find a software solution capable of scheduling their complex business.
In 2020, Pattison started working with D-Wave looking to optimize its auto-scheduling and in store personnel scheduling. Labor is the biggest cost for Pattison and the number of constraints they have to meet for scheduling, including union membership rules (of which they have many) age restrictions for some products, safety training, seniority, individual preference and the requirement for no split shifts was beyond any classical computer.
The initial plan was to design a scheduling program with the help of D-Wave, which they call a tool, and conduct a proof of concept in a single non-unionized store. Covid led to this being shelved.
During Covid, the team assembled to design the scheduling tools, began working on the Pattison e-commerce delivery scheduling problem, which grew enormously during the pandemic.
Working manually, it took 4 people in the head office 80 hours to complete the schedule each week, plus ongoing maintenance to the schedule with last-minute changes.
This is the process they went through to design the scheduling software. A key element is the mathematical optimization equations, without these you cannot write the code needed to solve the optimization problem on a D-Wave machine, it is a critical part of the process and Pattison had to hire ten people to perform this task.
Looking at the mathematical structure of the problem gives an idea of how mathematical this has become and how new this will be to many organizations.
When the system went live, all criteria were met and 80% of time spent on scheduling was saved by the automated scheduling program.
Development started in April 2022, the production launch was in October 2022 the tool has been running without issue since, a seamless transition to quantum.
The Pattison leadership team was impressed by the results and asked the team to revisit the original store scheduling problem.
Using the knowledge gained from the delivery schedule, the team coded the problem without further help from D-Wave, and it worked the first time. The store scheduling tool saves 12 working hours per store per week and delivers more consistent schedules that appear fair to everyone. “It just works” said the Pattison executive, senior store managers are now on the shop floor helping run the store more efficiently rather than sitting in the office writing schedules.
The Retail scheduling tool went to proof of concept in one store in April 2023 with two further pilots in different stores launched in August and October, the three stores’ estimated savings are 1,248 hours a year.
The tool has since been rolled out to 58 stores, saving 50,000 hours. Senior management has decided to roll out the system to Pattisons’ other banners in 2024, all 300 stores are going quantum. They intend to continue building the system into supply chain management, possibly the biggest savings of all are still to come.
Hermes Germany
The success of Pattison Food Group is leading to other companies exploring D-Wave. This month Hermes Germany GmbH, responsible for over 30% of German business to consumer parcels with 18,000 employees and shipments of 817 million per year, announced that it has partnered with D-Wave and Quantum Basel to look at designing a new route optimization package to improve its efficiency and environmental impact. This is the definition of commercial traction, one company tries it and reports success and other companies follow.
Ford Transit Production
Ford Otosan (Turkish Company 41% owned by Ford Motor Company (F)) explained in the conference how it was using D-Wave to deal with unexpected problems in the body shop. The specific example given was of a single part being delayed. The company has developed three optimization algorithms that can reschedule production so transit vans not requiring the missing part can be brought forward and production continue. Ford Otosan is looking to extend the software upstream and downstream from the body shop.
It is not too difficult to see how this program, if successful, could quickly gain traction in Ford production facilities worldwide and then into the entire automotive industry.
Traction in Other Areas
The QUBITS conference provided evidence of traction in many other areas. Davidson Technologies, the Alabama-based defense software contractor, announced they will host a new D-Wave annealing computer at their Global HQ in Huntsville. The machine will be initially available as part of the leap cloud. Still, it will be taken offline and become a secure facility for missile control and scenario development for the US military.
Triumf, the Canadian particle accelerator, discussed using the D-Wave annealer to develop synthetic data to allow them to understand and model their experiments.
The number of companies presenting at the conference was substantial and varied from Mastercard and Deloitte in finance, Pattison and Ford in operations, defense contractors and scientists.
It was clear that most of these companies are just beginning the quantum journey and are still some way from fully deploying solutions. Pattison foods was the clear exception to this, it looked like the Pattison journey took around 20 months from the decision being made to commercial deployment, but probably took more than 24 months for that decision to be made.
This implies that the quantum journey may be four years from the initial concept to full-scale deployment.
Forecasting D-Wave
I have built a three-statement mathematical model for D-Wave, I do not believe it is ready to provide a discounted cash flow valuation, but it will help monitor the performance of D-Wave going forward and hopefully highlight a good time to invest.
The model has three scenarios, Optimistic, Base and Worse. The key line items are highlighted below and present the average of the three projected scenarios.
The model gives a 2024 EBITDA of ($95) million, but management are providing guidance of ($54) million (Q1 2024 earnings call CFO prepared remarks). A significant difference that is accounted for in the cost section of the model.
The optimistic case assumes that D-Wave will be able to continue reducing its operating costs, they declined 17% year-over-year in Q1 2024 giving a total “optimistic” operating cost of $66 million. D-Wave guidance matches the optimistic case.
The base case assumes operating expenses will be the figure for Q1 2024 multiplied by four, and the worse case assumes they will continue to increase in line with the increase from 2022-2023.
Revenue Forecasts
2024 revenue forecasts were as follows
Worse case = Q1 2024 revenue x four; historically, Q1 has shown poor seasonality relative to the other quarters. I set the base case 20% higher than the worst case and the Optimistic case 50% higher.
The average for 2024 came in at $12 million, which seems to fit with previous results. From 2025 onwards, I am forecasting significant growth as the products gain traction in the commercial field and companies begin to increase the use of the D-Wave cloud offering.
In future years, I assume near-exponential growth will occur. These figures are highly speculative and will need significant adjustment as actual figures arrive.
Management is not currently providing guidance in regard to revenue, however Q1 2024 revenue increased 54% over Q1 2023 and bookings increased 54% to $4.5 million. (10Q Q1 2024)
The model suggests a break-even date sometime in 2029, but that would come forward to 2027 if management hits their forecast for reducing costs and revenue grows according to the base case scenario.
The Balance Sheet
The Balance sheet is far from perfect. D-Wave has had negative equity since early 2022. The Q1 2024 report showed total debt at $63 million, with cash and short-term investments of $27 million. In the May earnings call, the CFO reported a cash balance of $33 million after raising additional funds. The CFO said that total liquidity available to the company came from a $175 million shelf registration that became effective on April 12th, and an equity line of credit with Lincoln Park that has $82 million available.
The cash position is better than 12 months ago, in Q1 2023 they reported a cash balance of $9 million.
In Q1 2024, D-Wave reported a net decrease in Cash and equivalents of $14 million, using $12 million of cash to fund operating activities. Only the optimistic case on my model suggests that D-Wave will have enough available liquidity to get to positive cash generation and that would depend on the share price holding above $1, which is a term of the Lincoln Park eloc agreement.
Conclusion
D-Wave is leading the quantum computing industry towards commercial operations, it has a fault-tolerant quantum annealing machine with thousands of qubits. The new Advantage2 architecture, the improved annealing protocol and the addition of additional solving software has provided a machine that meets the strict definition of Quantum Supremacy.
The optimization benefits of the D-Wave annealer have been proven in a real-world business setting and a growing number of companies are exploring its potential and some are beginning to deploy applications.
The financial situation is far from ideal, with negative equity and a large shelf in place.
My mathematical model raises questions about when D-Wave will reach profitability, highlighting the near exponential revenue growth and continued cost-cutting needed to achieve it within the forecast period.
I really want to add D-Wave to my portfolio, but it does not yet meet my Strong Buy criteria. As a result, I will continue to watch carefully and update my model after each quarter’s results and considering any news releases. My Strong-Buy portfolio (companies I publish with a strong buy recommendation) has a hit rate of over 80% for 2024 and 65% over the long term. I am now only investing when companies meet this higher criterion.
For D-Wave to become a strong buy, I need clarity on the path toward profitability. Specifically, evidence of sales volume growing in line with the base or optimistic case in my model, costs reducing as indicated by management, and an improved balance sheet. At the moment, D-Wave has a market cap of $190 million and a shelf in place for $175 million, suggesting the possibility of a near 100% dilution. For me, the perfect time to buy would be when the shelf has been used, positive equity has been restored, and sales volume is growing with a clear line of sight to profitable operations.
Editor’s Note: This article covers one or more microcap stocks. Please be aware of the risks associated with these stocks.
Read the full article here