Item 1. Business
Item 1. Business.
Overview
We are developing quantum computers and networks designed to solve some of the world’s most complex problems, and transform business, society and the planet for the better. We believe that our proprietary technology, our architecture, and the technology exclusively available to us through license agreements will offer us advantages both in terms of research and development, as well as the commercial value of our intended product offerings.
Today, we sell specialized quantum computing and networking hardware together with related maintenance and support. We also sell access to several quantum computers of various qubit capacities and are in the process of researching and developing technologies for quantum computers with increasing computational capabilities. We currently make access to our quantum computers available via three major cloud platforms, Amazon Web Services’ (“AWS”) Amazon Braket, Microsoft’s Azure Quantum and Google’s Cloud Marketplace, and also to select customers via our own cloud service. This cloud-based approach enables the broad availability of quantum-computing-as-a-service (“QCaaS”).
We supplement our offerings with professional services focused on assisting our customers in applying quantum computing and networking to their businesses. We also expect to sell full quantum computing systems to customers, either over the cloud or for local access. We also offer quantum networking products which offer customers secure communication networks and enable networked quantum computing.
We are still in the early stages of commercial growth. Since our inception, we have incurred significant operating losses. Our ability to generate revenue sufficient to achieve profitability will depend heavily on the successful development and further commercialization of our quantum computing systems. Our net losses were $331.6 million, $157.8 million and $48.5 million, for the years ended December 31, 2024, 2023 and 2022, respectively. As of December 31, 2024, we had an accumulated deficit of $683.7 million. We expect to continue to incur significant losses for the foreseeable future as we prioritize reaching the technical milestones necessary to achieve an increasingly higher number of algorithmic qubits and higher levels of fidelity than presently exists—prerequisites for quantum computing to reach broad quantum advantage.
The Quantum Opportunity
Throughout human history, technological breakthroughs have dramatically transformed society and altered the trajectory of economic productivity. In the 19th century, it was the industrial revolution, powered by the scientific advances that brought us steam-powered machines, electricity, and advanced medicine. These technologies drastically improved human productivity and lengthened life expectancy.
In the 20th century, computing—arguably the greatest of all human inventions—leveraged human intelligence to run complex calculations, paving the way for profound advances in virtually every realm of human experience, including information processing, communication, energy, transportation, biotechnology, pharmaceuticals, agriculture and industry.
Since classical computing emerged in the mid-twentieth century, there has been exponential progress in computer design, with processing power roughly doubling every few years (Moore’s law). The true economic and social impact of computing is difficult to measure because it has so thoroughly permeated every aspect of life, altering the trajectory of society.
However, as transformative as computing has been, many classes of problems strain the ability of classical computers, and some will never be solvable with classical computing. In this traditional binary approach to computing, information is stored in bits that are represented logically by either a 0 (off) or a 1 (on). Quantum computing uses information in a fundamentally different way than classical computing. Quantum computers are based on quantum bits (qubits), a fundamental unit that can exist in both states 0 and 1 simultaneously (superposition). As a result, we believe that quantum computers can address a set of problems classical computing may never solve. The types of problems that currently defeat classical computing include: the simulation of quantum systems (e.g., in materials science or pharmaceuticals); number factoring for decryption; and complex optimization problems. Many of these problems are fundamental, involving society’s most pressing needs, such as how to live sustainably on our planet, how to cure diseases, and how to efficiently move people and goods. Classical computers cannot solve these problems because the calculations would take far
4
too long (i.e., millions to trillions of years) or because the problems involve quantum systems that are far too complex to be represented on a classical computer, even if their remarkable pace of development were to continue indefinitely. While these problems are not solvable by today’s quantum computers, we believe that a quantum computer currently offers the best possibility for computational power that could be used to solve them.
The future success of quantum computing will be based on the development of a computer with a substantially higher number of qubits than our current computers. We believe that we will find solutions to these challenges and that our proprietary technology and architecture and the technology exclusively available to us through exclusive license agreements will offer advantages both in terms of research and development as well as the ultimate product we wish to offer customers.
There are certainly thousands, if not millions, of important and fundamental unanswered questions about how the universe works and opportunities associated with the answers to those questions. We envision a future powered by quantum computing and believe the 21st century is poised to be the dawn of this era.
Our Strategy
Our mission is to be the leading quantum computing company enabling the new era of quantum computing. We intend to fulfill our mission by:
• Leveraging Our Technology . We believe that our technology offers substantial technological advantages compared to other competing quantum computing and networking systems. We intend to build upon our technological lead by leveraging our world-class team of leaders and engineers who are pioneers in quantum computing and networking, with proven track records in innovation and technical leadership. To date, we have developed and assembled ten generations of quantum computer prototypes and systems, have constructed quantum operating systems and software tools, and have worked with leading cloud vendors, quantum programming languages and quantum software development kits (“SDKs”).
• Selling Direct Access to Quantum Computers . We sell specialized quantum computing and networking hardware to select customers, complemented by access to quantum experts and algorithm development capabilities. We sell direct access to the quantum computers we manufacture, with units offered on a whole system or usage basis. We believe that by offering direct access to quantum computing, supplemented by our professional services, we can assist select customers in deepening their application of quantum solutions.
• Offering QCaaS . We provide QCaaS, complemented by access to quantum experts and algorithm development capabilities. We manufacture, own and operate quantum computers. Our quantum computing solution is currently delivered via AWS’s Amazon Braket, Microsoft’s Azure Quantum and Google’s Cloud Marketplace as well as on our own cloud platform. We believe that by offering QCaaS, we can accelerate the adoption of our quantum computing solutions, while efficiently promoting quantum computing across our partner ecosystems.
• Continuing to Enhance Our Proprietary Position . We have exclusively licensed our core technology from the University of Maryland and Duke University (together, the “Universities”), and our complex technology is protected by an extensive patent portfolio. We intend to continue to drive innovation in quantum computing and networking and seek intellectual property protection where appropriate to enhance our proprietary technology position.
• Further Developing Our Quantum Computing and Networking Partner Ecosystem . We believe our relationships with leading technology enterprises, government agencies, and university research institutes will accelerate innovation, distribution and monetization of our quantum capabilities.
Market Opportunity: A Future Driven by Quantum Computing
The potential uses for quantum applications are widespread and address a number of problems that would be impossible to solve using classical computing technology. Below are a few of the use cases in which we believe quantum computers, if they are successfully developed, will become an important tool for businesses to remain competitive in the market over the coming years.
Quantum Simulations in Chemistry
We believe that there are thousands of problems that could benefit from these quantum algorithms across the pharmaceutical, chemical, energy and materials industries. An example of such a simulation problem is modeling the core molecule in the nitrogen fixation process to make fertilizer. Nature is able to fixate nitrogen (i.e., turn atmospheric nitrogen into more useful ammonia) at room temperature. Scientists, however, have only been able to achieve fixation using a resource-intensive, high-temperature, high-pressure process, called the Haber-Bosch process. A cornerstone of the global agriculture industry, the Haber-Bosch process consumes about one percent of the world’s energy and produces about one percent of the world’s carbon dioxide. Agronomists have attempted to model the core molecule in nature’s nitrogen fixation process, but the molecule is too large for today’s classical supercomputers to
5
simulate. Understanding the quantum process used in nature to fixate nitrogen could lead directly to more efficient ways for scientists to do the same.
Quantum chemistry simulation is expected to impact multiple markets and become an essential tool in chemical industries. For example, computer-aided drug discovery in the pharmaceutical industry is limited by the computing time and resources required to simulate a large enough chemical system with sufficient accuracy to be useful. If future generations of more powerful quantum computers are successfully developed, we believe that we could improve the speed and accuracy of virtual high-throughput screening and improve the molecular docking predictions used in structure-based drug discovery, dramatically reducing the development cost of new drugs and reducing the time to market. Similarly, we believe that developing a detailed understanding of chemical reactions critical to various industries, such as catalytic reaction in battery chemistry for electric vehicles, can lead to higher performing solutions with extended energy storage capacity.
Quantum Algorithms for Monte Carlo Simulations
Monte Carlo simulations are probability simulations used to calculate the expected distribution of possible outcomes in hard-to-predict processes involving random variables. Such simulations are used pervasively in finance, banking, logistics, economics, engineering and applied sciences. A key parameter of Monte Carlo simulations is the degree of accuracy desired to attain with the result. To obtain 99.9% accuracy, a classical computer requires around one million simulations. Quantum algorithms, however, can achieve the same accuracy using only one thousand simulations, thereby significantly reducing the time it takes to perform Monte Carlo simulations. This is especially important when running these simulations is expensive.
One application of the quantum Monte Carlo algorithm is to price options for the financial industry. Simple options models are used ubiquitously in finance, the most famous of these being the Black-Scholes model. However, these models fail to capture the complexities of real markets, and financiers use more sophisticated simulations to obtain better model predictions. Currently, many of these models are limited by the number of simulations required to reach the desired accuracy within a fixed time budget. Quantum algorithms for Monte Carlo simulations could give some financial firms a competitive advantage by enabling them to price options more quickly.
Quantum Algorithms for Optimization
Optimization problems have enormous economic significance in many industries, and they often cannot be solved with classical computers due to their daunting complexity. Quantum algorithms are naturally suited for problems in which an exponential number of possibilities must be considered before an optimized output can be identified. It is widely believed that quantum computers will be able to arrive at a better approximate optimization solution than classical computers can, and with reduced computational cost and time. One method of quantum optimization is a hybrid method called the Quantum Approximate Optimization Algorithm, in which layers of quantum computations are executed within circuit parameters optimized using classical high-performance computers. Because optimization issues bedevil so many complicated processes in industries ranging from logistics to pharmaceutical drug design to climate modeling, the application of quantum algorithms to optimization problems could have far-reaching impacts on society.
Quantum Machine Learning
Quantum computers can generate probability distributions that cannot be efficiently simulated on a classical computer. Similarly, there are probability distributions that can only be efficiently distinguished from each other using a quantum computer. In these examples, models utilizing quantum circuits can be used to capture complex internal structures in the data set much more effectively than classical models. In other words, quantum computers can “learn” things that are beyond the capabilities of classical computers. Quantum computing is likely to offer new machine-learning modalities, greatly improving existing classical machine learning when used in tandem with it. Examples of areas where quantum machine learning could have an impact are risk analysis in finance, natural language processing, and classification of multivariate data such as images and chemical structures. Machine learning is used broadly in industry today, and we believe quantum machine learning could have a similarly broad impact.
As with any completely new technology, the use cases imagined by us today are only a subset of the opportunities that will emerge if future generations of more powerful quantum computers are successfully developed, as users understand the power of quantum algorithms.
Remaining Challenges in Quantum Computing Evolution
One can compare any particular quantum algorithm’s performance to the best classical algorithm for the same problem. The point at which a quantum computer is able to perform a particular computation that exceeds its classical counterpart in speed or reduces its cost to solution is known as the point of “quantum advantage.”
6
Given the substantial research and development required to build a modern quantum computer that is both functional and practical, industry experts describe the remaining challenges in quantum computing to achieve quantum advantage as being solved in three phases. Although none of these challenges have yet been fully solved, we believe that we are well positioned to do so. A 2019 publicly available report by a leading third-party consulting firm describes these phases—and the associated technical barriers—as paraphrased below:
• Noisy and intermediate-scale quantum (NISQ) computers : The earliest stage of development will see component demonstrations and intermediate-scale system development with limited commercial application. The main technical barrier involves the mitigation of errors through improved fabrication and engineering of underlying qubit devices and advanced control techniques for the qubits. These devices are used for developing and validating fundamentally new quantum approaches to tackling difficult problems, but are not expected to generate substantial commercial revenues.
• Broad quantum advantage : In this stage, quantum computers are expected to provide an advantage over classical computers with a meaningful commercial impact. The main technical barrier is the deployment of quantum error-correcting codes that allow bigger applications to be executed. If this barrier can be overcome, we believe that quantum computing will offer practical solutions to meaningful problems superior to those provided by classical computers.
• Fault-tolerant quantum computing (FTQC) : This last stage will see large modular quantum computers with enough power to tackle a wide array of commercial applications relevant to many sectors of the economy. At this stage, classical computers are expected to no longer compete with quantum computers in many fields. The technical barrier will be the adoption of a modular quantum computer architecture that allows the scalable manufacturing of large quantum computer systems.
In a 2024 update of the previously referenced publicly available report, the third-party consulting firm detailed $50 billion in total government and private investment in quantum technology and estimated up to $2 trillion in economic value from quantum computing in the next ten years.
Building a Quantum Computer or a Quantum Network
Requirements for Building Useful Quantum Computers and Networks
Quantum computers and networks are difficult to build and operate because the physical system of qubits must be nearly perfectly isolated from its environment to faithfully store quantum information. Yet the system must also be precisely controlled through the application of quantum gate operations, and it must ultimately be measured with high accuracy. A practical quantum computer or network requires well-isolated, near-perfect qubits that are cheap, replicable and scalable, along with the ability to initialize, control and measure their states. Breakthroughs in physics, engineering, and classical computing were prerequisites for building a quantum computer or network, which is why for many decades the task was beyond the limits of available technology.
To execute computational tasks, a quantum computer must be able to (i) initialize and store quantum information in qubits, (ii) operate quantum gates to modify information stored in qubits and (iii) output measurable results. Each of these steps must be accomplished with sufficiently low error rates to produce reliable results. Moreover, to be practical, a quantum computer or network must be economical in cost and scalable in compute power (i.e., the number of qubits and the number of gate operations) to handle real world problems.
The development of large-scale quantum computing and networking systems is still in early stages, and several potential engineering architectures for how to build a quantum computer or network have emerged. We are developing quantum computers based on individual atoms as the core qubit technology, which we believe has key advantages in scaling. The ability to produce cheap error-corrected qubits at scale in a modular architecture is one of the key differentiators of our approach. We have achieved many engineering firsts in this field and we believe that, with our focus on achieving additional technical milestones over the next few years, we are well positioned to bring quantum computing advantage to the commercial market.
Scientific Approaches to Quantum Computing
There are a variety of different approaches to (or architectures for) building a quantum computer, each of which involves tradeoffs in meeting the three functional and practical requirements outlined above. Roughly, approaches to performing a quantum computation fall into one of three categories: natural quantum bits, solid state or classical computer simulation.
7
Natural quantum bits : In natural qubit-based quantum computers, a system is built around naturally occurring substrates exhibiting quantum properties.
• Atoms : In atomic-based quantum computers, the qubits are represented by internal states of individual atoms trapped and isolated in a vacuum. There are two categories within this approach: the use of ionized (charged) atoms and the use of neutral atoms.
• Photons : In this approach, the state of a photon, a particle of light, is used as the qubit. Various aspects of a photon, such as presence/absence, polarization, frequency (color) or its temporal location can be used to represent a qubit.
Solid state : In solid-state-based quantum computers, the qubits are engineered into the system.
• Spins in semiconductors : This approach uses the spins of individual electrons or atomic nuclei in a semiconductor matrix. There are two categories within this approach: (1) the use of electrons trapped in quantum dot structures fabricated by lithographic techniques and (2) the use of atomic defects (or dopants) that capture single electrons. The nuclear spin of the dopant atoms, or the nearby atoms to defects, are often used to store qubits.
• Superconducting circuits : This approach uses circuits fabricated using superconducting material that features quantum phenomena at cryogenic temperatures. Two states of the circuit, either charge states or states of circulating current, are used as the qubit.
Classical computer simulation : Classical computers in a data center can be used to simulate quantum computers. Although useful for small-scale quantum experiments, quantum simulation on classical computers is still bound by the same limitations of classical computing and would require an impractical number of data centers to tackle meaningful quantum problems.
Our Technology Approach
Our Approach to Quantum Computing: Trapped Ions
We have adopted the atom-based approach described above and use trapped atomic ions as the foundational qubits to construct practical quantum computers. We are pursuing a modular computing architecture to scale our quantum computers, meaning that, if successful, individual quantum processing units will be connected to form increasingly powerful systems. We believe that the ion trap approach offers the following advantages over other approaches:
• Atomic qubits are nature’s qubits : Using atoms as qubits means that every qubit is exactly identical and perfectly quantum. This is why atomic qubits are used in the atomic clocks that do the precise timekeeping for mankind. Many other quantum systems rely upon fabricated qubits, which bring about imprecisions such that no single qubit is exactly the same as any other qubit in the system. For example, every superconducting qubit comes with a different frequency (or must be tuned to a frequency) due to manufacturing imprecision. Overall, we believe that systems relying upon fabrication of their qubits are more susceptible to error.
• Trapped ion qubits are well-isolated from environmental influences : When a quantum system interacts with its environment, the quantum state loses coherence and is no longer useful for computing. For example, in a superconducting qubit, the qubit tends to lose its coherence within approximately 10 to 50 microseconds. Even neutral atoms are perturbed to some extent when they are trapped in space. In contrast, trapped ion qubits are confined via electric fields in an ultra-high vacuum environment, and their internal qubits are hence perfectly isolated. As a result, the coherence of trapped ions can be preserved for about an hour, and may be able to be preserved for longer if isolation technology improves. Longer coherence times mean more computations can be performed before noise overwhelms the quantum calculation and are key to minimizing the overhead of error correction needed for large-scale quantum computers.
• Lower overhead for quantum error-correction . Quantum error-correction will likely be necessary to reduce the operational errors in any large-scale quantum computations relevant to commercial problems. Quantum error-correction uses multiple physical qubits to create an error-corrected qubit with lower levels of operational errors. For solid-state architectures, we estimate that it may take at least 1,000 physical qubits to form a single error-corrected qubit, while for near-term applications with ion traps the ratio is closer to 16:1.
• Trapped ion quantum computers can run at room temperature : Solid-state qubits currently require temperatures close to absolute zero (i.e., -273.15°C, or -459.67°F) to minimize external interference and noise levels. Maintaining the correct temperature requires the use of large and expensive dilution refrigerators, which can hamper a system’s long-term scalability because the cooling space, and hence the system space, is limited. Trapped ion systems, on the other hand, can operate at room temperature. This is because the qubits themselves are not in thermal contact with the environment, as they are electromagnetically confined in free space inside a vacuum chamber. Although modest cryogenics (< 10 degrees above absolute zero) can be used to dramatically improve the vacuum environment, the inherent properties of the qubits
8
themselves do not degrade at room temperature. The laser-cooling of the qubits themselves is extremely efficient because the atomic ions have very little mass and this requires just a single low-power laser beam (microwatts). This allows us to minimize the system size as technology progresses, while scaling the compute power and simultaneously reducing costs.
• All-to-all connectivity : In superconducting and other solid-state architectures, individual qubits are connected via physical wires, hence a particular qubit can only communicate with a further-removed qubit by going through the qubits that lie in-between. In the trapped ion approach, however, qubits are connected by electrostatic repulsion rather than through physical wires. As a result, qubits in our existing systems can directly interact with any other qubit in the system. Our modular architecture benefits from this flexible connectivity, significantly reducing the complexity of implementing a given quantum circuit.
• Ion traps require no novel manufacturing capabilities : Ion trap chips consist of electrodes and their electrical connections, which are built using existing technologies. The trap chips themselves are not quantum materials. They simply provide the conditions for the ion qubits to be trapped in space, and in their current state, they can be fabricated with existing conventional and standard silicon or other micro-fabrication technologies. By contrast, solid-state qubits, such as superconducting qubits or solid- state silicon spins, require exotic materials and fabrication processes that demand atomic perfection in the structures of the qubits and their surroundings; fabrication with this level of precision is an unsolved challenge.
Technological Complexity Creates Significant Barriers to Entry
Alongside the benefits of the trapped ion approach, there are several challenges inherent in it that serve as barriers-to-entry, strengthening the advantages of our systems. These key challenges include:
• Complex laser systems : One of the challenges of trapped ion quantum computing is the set of lasers required and the degree to which they must be stable to operate the system. Traditionally, these laser systems were assembled on an optical table on a component-by-component basis, which led to serious stability and reliability issues. We believe that we have resolved this issue from an engineering standpoint and that our future roadmap will further improve manufacturability.
• Ultra-high vacuum (UHV) technology : The conventional method to achieve UHV conditions for ion trapping experiments involves using vacuum chamber designs with carefully chosen materials, assembly procedures with cumbersome electrical connections, and a conditioning procedure to prepare and bake the chamber at elevated temperatures for extended periods of time. We have developed new approaches, such as environmental conditioning, that we believe will substantially reduce the time and cost to prepare the UHV environment to operate the quantum computer.
• Executing high fidelity gates with all-to-all connectivity : While trapped ion qubits feature the highest fidelity entangling gates, it is nevertheless a major technical challenge to design a control scheme that enables all qubits in a system to form gates with each other under full software control. Through innovation in gate-implementation protocols, we believe that we have developed laser delivery and control systems that will allow us to implement fully programmable, fully connected gate schemes in our system.
• Slow gate speeds : Compared to their solid-state counterparts, trapped ions are widely believed to have slow gate speeds. While slow gate speeds are the case for many systems in operation today, both theoretical analyses and experimental demonstrations suggest this may not be a fundamental limit of trapped ion qubits (although this has not yet been demonstrated in commercial applications). In fact, high-fidelity gates with speeds comparable to those of solid-state qubits have been realized in several research laboratories. We expect that our future quantum computers based on barium ions will be faster, more powerful, more easily interconnected, and will feature more uptime for customers. Moreover, we believe that as systems with other qubit technologies scale up, their restricted connectivity and high error-correction overhead will significantly slow down their overall computation time, which we believe will make the trapped ion approach more competitive in terms of operational speed.
Our Trapped Ion Implementation
The specific implementation of our trapped ion systems leverages the inherent advantages of the substrate and creates what we believe is a path for building stable, replicable and scalable quantum computers.
Trapped Ion Infrastructure
Our systems are built on individual atomic ions that serve as the computer’s qubits. Maintaining identical, replicable, and cost-effective qubits is critical to our potential competitive advantage, and we have developed a process to produce, confine and manipulate atomic ion qubits.
9
To create trapped atomic ion qubits using our approach, a solid source containing the element of interest is either evaporated or laser-ablated to create a vapor of atoms. Laser light is then used to strip one electron selectively from each of only those atoms of a particular isotope, creating an electrically charged ion. Ions are then confined in a specific configuration of electromagnetic fields created by the trapping structure ( i.e. , the ion trap), to which their motion is confined due to their charge. The trapping is done in an UHV chamber to keep the ions well-isolated from the environment. Isolating and loading a specific isotope of a specific atomic species ensures each qubit in the system is identical. Two internal electronic states of the atom are selected to serve as the qubit for each ion. The two atomic states have enough frequency separation that the qubit is easy to measure through fluorescence detection when an appropriate laser beam is applied.
To build quantum computers, many atomic ions are held in a single trap, and the repulsion from their charges naturally forces them into a stable linear crystal (or chain) of qubits. The qubits are highly isolated in the UHV chamber, only perturbed by occasional collisions with residual molecules in the chamber, which provides near-perfect quantum memory that lasts much longer than most currently envisioned quantum computing tasks require. The qubits are initialized and measured through a system of external gated laser beams. An additional set of gated laser beams applies a force to selected ions and modulates the electrical repulsion between the ions. This process allows the creation of quantum logic gates between any pair of qubits, regardless of their distance within the crystal, which can be arbitrarily reconfigured in software.
System Modularity and Scalability
Today, all qubits in our systems are stored on a single chip, referred to as a quantum processing unit (“QPU”). QPUs can have several cores, or zones for trapping chains of ions, comparable to multicore central processing unit (“CPU”) chips in classical computing. Each core can contain up to about 100 qubits in a linear crystal, and dozens of cores can potentially be co-located in a single QPU. Within a QPU, some qubits can be physically moved between cores to accommodate quantum communication between the cores. This process of moving ions within a QPU is called “shuttling” and is achieved by modifying the electromagnetic fields that form the trap.
In addition to increasing the number of qubits per QPU, we believe we have identified, and we are currently developing, the technology needed to connect qubits between trapped ion QPUs, which may be commercially viable in the future. This technology, known as a photonic interconnect, uses light particles to communicate between qubits while keeping information stored stably on either end of the interconnect. The basic protocol for this photonic interconnect between ion traps in two different vacuum chambers was first realized in 2007. We believe this protocol can be combined with all-optical switching technology to enable multi-QPU quantum computers at large scale. We at IonQ have assembled a team with deep expertise in photonics and are designing Photonic Interconnects that will enable our systems to compute with entangled qubits spanning multiple QPUs. We believe this can open up the possibility of scaling quantum computers indefinitely, similar to how high-performance computers and data centers have been scaled.
Our quantum architecture is modular, meaning that if development of this architecture is successful, the number of qubits in a QPU, or the number of QPUs in a system, could be scaled. Also, by allowing for each qubit in a system to entangle with any other qubit in that system, we believe that a system’s number of quantum gates could increase rapidly with each additional qubit added. This all-to-all connectivity is one of the key reasons we believe our systems will be computationally powerful. Notably, our architectural approach to scaling quantum computers across several QPUs has also contributed to IonQ's work in quantum networking.
Gate Configuration
Our qubits are manipulated (for initialization, detection, and forming quantum logic gates) by shining specific laser beams onto the trapped ions. Our systems employ a set of lasers and a sophisticated optical system to deliver beams precisely tailored to achieve this manipulation. The laser beams are tailored by programming radio frequency (“RF”) signals using state-of-the-art digital chipsets, which are custom-configured to generate the signals for qubit manipulation. An operating system manages the quantum computer, maintaining the system in operation. It includes software toolsets for converting quantum programs from users into a set of instructions the computer hardware can execute to yield the desired computational results. To support system access from the cloud, we offer cloud management tools and application programming interfaces (“APIs”) that permit programming jobs to run remotely.
Our quantum gates are fully programmable in software; there is no “hard-wiring” of qubit connections in the quantum computing hardware. The structure of a quantum circuit or algorithm can therefore be optimized in software, and the appropriate laser beams can then be generated, switched, or modulated to execute any pattern of gate interactions. Our programmable gate configurations make our systems adaptable. Unlike quantum computer systems that are limited to a single class of problems due to their architecture, we believe that any computational problem with arbitrary internal algorithmic structure could be optimized to run on our system, although this has not been demonstrated at scale.
10
Quantum Error Correction
A key milestone in building larger quantum computers is achieving fault-tolerant quantum error-correction. In quantum error-correction, individual physical qubits prone to errors are combined to form an error-corrected qubit (sometimes referred to as a logical qubit) with a much lower error rate. Determining how many physical qubits are needed to form a more reliable logical qubit (the resource “overhead”) depends on both the error rate of the physical qubits and the specific error-correcting codes used. In 2020, a team of researchers at the University of Maryland, including some current IonQ employees, demonstrated the first fault-tolerant error-corrected qubit using 13 trapped ion qubits. In 2024, we announced a partial error correction scheme which proposes a 3:1 qubit overhead to reduce noise and error in certain quantum gates that we believe we may implement in future generations of IonQ quantum computers. With our unique architecture, we believe quantum error-correction can be coded in hardware and software, allowing varying levels and depths of quantum error-correction to be deployed as needed. Because the ion qubits feature very low idle and native error rates and are highly connected, to achieve the first useful quantum applications we expect the error-correction overhead to be significantly lower than other approaches, for which we estimate the overhead to be on the order of at least 1,000:1.
We believe our architectural decisions will make our systems uniquely capable of achieving scale. We have published a roadmap for scaling to larger quantum computing systems, with concrete technological innovations designed to significantly improve the performance of the systems. For example, in 2022, we announced that through our partnership with the U.S. Department of Energy’s Pacific Northwest National Laboratory (“PNNL”), we were able to shrink the barium source material down to a microscopic scale. We believe this is significant because it is evidence of our continued progress to reduce the size of core system components, an important step in the creation of quantum computers small enough to be networked together. However, meeting future milestones included in our roadmap is not guaranteed and is dependent on various technological advancements, which could take longer than expected to realize or turn out to be impossible to achieve. We believe that, with engineering advancements and firsts yet to be achieved, our quantum computers will become increasingly compact and transportable, opening up future applications of quantum computing at the edge.
Our Forward-Looking Roadmap
In December 2020, we publicly released a forward-looking technical roadmap for the next eight years. Our technical roadmap was designed to provide transparent guidance to our quantum computer users regarding when we expect certain quantum computing capabilities to become available. As part of this roadmap, we introduced the notion of “algorithmic qubits” as a metric to measure progress, and a detailed description of how to define and measure the number of algorithmic qubits (#AQ) in early 2022. Roughly speaking, #AQ represents the total number of qubits that can be used to perform a quantum computational task that involves an order of ~(#AQ)^2 entangling gate operations in a list of quantum algorithms that reflect representative real-world use cases of a quantum computer. This metric provides a simple and effective measure to estimate the computational power of each generation of quantum computers. At low #AQ, the size of the problem the quantum computer can tackle is limited by the error rate of the entangling gate operations, rather than by the number of physical qubits available in the computer. The aggressive push for improving the power of quantum computers, including the early introduction of quantum error-correction, is intended to significantly compress the time required for reaching the point when we expect quantum computers may become commercially impactful at scale. We believe that many of the technological components needed to accomplish the performance goals of the roadmap, such as high-fidelity gate operations, photonic interconnects and quantum error-correction, have been realized in proof-of-concept demonstrations in trapped ion systems. Given our track record of engineering and technology development, we believe that, over time, we will be able to successfully translate these technology components into products, which may enable successful deployment of our quantum computers and deliver material commercial value to customers.
We are targeting a Modular Architecture, Designed to Scale, resulting in Smaller Systems and Cheaper Compute Power for Each Generation
The scaling of classical computer technology, which unlocked continuously growing markets over many decades, was driven by exponential growth in computational power coupled with exponential reduction in the cost of computational power for each generation (Moore’s law). The key economic driver permitting the expansion of digital computer applications to new segments of the market was this very phenomenon of capability doubling in each generation with costs rising only modestly. We believe the scaling of quantum computing may follow a similar trajectory: as the #AQ available in each generation scales, the per-AQ cost is also reduced and enables true scaling of quantum computers. Our systems have benefited from years of architectural focus on scalability that addresses both #AQ and per-AQ cost and, as such, we believe that if we are able to successfully solve remaining scalability challenges, these systems may become increasingly powerful and accessible in tandem.
At the heart of our approach is the modular architecture that may enable such growth. We expect our future systems to be modular networks of many QPUs working together as a large quantum computer, similar to how classical data centers are designed, constructed and operated today. Our engineering effort is focused on reducing the size, weight, cost, and power consumption of the QPUs that will be the center of each generation of the modular quantum computer, while increasing the number of QPUs
11
manufactured each year. We intend to focus on achieving these engineering efforts over the next several years. If successful, we expect that we may be able to achieve compact, lightweight and reliable quantum computers, which can be deployed at the edge, similarly to how personal computers have enabled new applications for both government and commercial use.
Our Business Model
Quantum Computing and the Compute Access Model
As quantum hardware matures, we expect the quantum computing industry to increasingly focus on practical applications for real-world problems, known as quantum algorithms. Today, we believe that there are a large number of quantum algorithms widely thought to offer advantages over classical algorithms in that each of these algorithms can solve a problem more efficiently, or in a different manner, than a classical algorithm. Our business model is premised on the belief that businesses with access to quantum computers and networks will likely have a competitive advantage in the future.
We provide quantum computing and networking services, complemented by access to quantum experts and algorithm development capabilities, designed to solve some of the most challenging issues facing corporations, governments and other large-scale entities today. We manufacture, own and operate quantum computers and networks, with compute units being offered to potential customers through system hardware sales and on a QCaaS basis and network units being offered through hardware sales. We manufacture specialized quantum computers for specific use cases for customers including government agencies. We also provide quantum networking solutions to augment existing networks with quantum capabilities.
We expect our target markets to experience two stages of quantum algorithm deployment: the development stage and the application stage. We expect our involvement in these two stages, to the extent they will take place, to be as follows:
• During the development stage, our experts will assist customers in developing an algorithm to solve their business challenges. Customers may be expected to pay for quantum compute usage, in addition to an incremental amount for the consulting and development services provided in the creation of algorithms. We may choose to sell this computing time to customers in a variety of ways. In this stage, we expect revenue to be unevenly distributed, with individual customers potentially contributing to peaks in bookings.
• During the application stage, once an algorithm is fully developed for a market, we anticipate that customers would be charged to run the algorithm on our hardware. Given the mission critical nature of the use cases we anticipate quantum computing will attract, we believe a usage-based revenue model will result in a steady stream of revenue while providing the incremental ability to grow with customers as their algorithm complexity and inputs scale.
Our Customer Journey
In each new market that stands to benefit from quantum computing, we intend to guide our customers and partners through two stages: the development phase and the application phase.
Development Phase: This first stage focuses on quantum algorithm development and we expect it to involve deep partnerships between us and our customers to lay the groundwork for applying quantum solutions to the customer’s industry. We also anticipate uneven revenue for this period given that the quantum computing market is still nascent. We expect the development phase for each market to be characterized by the following go-to-market channels:
• Co-development of quantum applications with strategic partners. We intend to form long-term partnerships with select industry-leading companies (aligned with our technology roadmap) to co-develop end-to-end solutions for the partner and to provide an early-adopter advantage to the partner in their industry. IonQ has announced co-development agreements with Ansys for computer aided design (CAD) and engineering and with the US Defense Advanced Research Projects Agency (DARPA) to help establish the next generation of benchmarking for quantum computers.
• Preferred compute agreements with clients . We expect our preferred offerings to give the customer’s application engineers direct access to our cutting-edge quantum systems, as well as technical support to pursue their solution development.
• Dedicated hardware . We sell certain specialized quantum computing and networking hardware to select customers. We also expect to manufacture and sell complete quantum systems and networks for dedicated use by a single customer, to be hosted on premises by the customer or remotely by us.
• Cloud access to quantum computing . Our current and future cloud partnerships with AWS’s Amazon Braket, Microsoft’s Azure Quantum, Google’s Cloud Marketplace and other cloud providers are designed and will continue to be designed to make access to quantum computing hardware available to a broader community of quantum programmers.
12
Application Phase : This second phase is expected to commence if we are successful in demonstrating the commercial viability of quantum advantage in the industry and can therefore commence with developing commercial applications and applying that advantage broadly throughout the market with new customers.
• Delivery of a full-scale quantum compute platform . For customers who have worked alongside us in the development phase to curate deep in-house technical expertise in quantum computing capabilities at the time quantum advantage is achieved for the customer’s application, our preferred compute agreements, cloud offerings, and dedicated hardware sales are expected to offer sufficient quantum computational capacity.
• Packaged solution offerings . When appropriate, we may develop full-stack quantum solutions that can be provided directly to customers, regardless of their in-house quantum expertise.
• Accelerated high-impact applications development . We intend to provide opportunities for accelerated applications development to customers seeking compressed development timelines to solve some of their biggest problems and drive efficiencies.
We expect the technical complexity of the solutions required for quantum algorithms to address how each application area will impact the timing of that market’s inflection point and transition from the development phase to the application phase. We expect computational chemistry and optimization to be among the first solutions to transition into broadly available applications. Additional markets taking advantage of quantum material science research and optimization speed-ups may come online next if broad-scale quantum advantage becomes accessible. If our quantum computers achieve full-scale fault tolerance, a diverse array of industries, ranging from quantum machine learning to deeper optimization, may be able to be transitioned to the application phase.
Customers and Prospects
Quantum Computing and Networking Systems and Hardware
We sell certain specialized quantum computing and networking hardware to select customers. We are also engaged with certain prospects who are interested in purchasing partial or entire quantum computing and networking systems, either over the cloud or for local access.
Direct Access Customers
By directly integrating with us, customers can reserve dedicated execution windows, receive concierge-level application development support, gain early access to next-generation hardware, or host their own quantum computer. Such access is currently limited to a select group of end-users.
We expect our standard offerings will include additional bundled value-add services in exchange for an annual commitment, such as reserved system time, consultations with solution scientists, and other application and integration support.
QCaaS
We provide access to our quantum computing solutions via AWS’s Amazon Braket, Microsoft’s Azure Quantum, and Google’s Cloud Marketplace, and sell access directly to select customers via our own cloud service. Making systems available through the cloud in both cases enables wide distribution. Through our cloud service providers, potential customers across the world in industry, academia and government can access our quantum hardware with just a few clicks. These platforms serve an important purpose in the quantum ecosystem, allowing virtually anyone to try our systems without an upfront commitment or needing to integrate with our platform.
Government Agencies
Our customers, potential customers and partners include government agencies such as the United States Air Force Research Lab. Government agencies and large organizations often undertake a significant evaluation process. Our contracts with government agencies are typically structured in phases, with each phase subject to satisfaction of certain conditions.
Commercial Agreements with the University of Maryland
Lease with the University of Maryland
In March 2020, we entered into an amended and restated office lease with UMD for the lease of our corporate headquarters and our research and development and manufacturing facility. This lease expires on November 30, 2030.
13
Contracts with the University of Maryland
In September 2021, we entered into a contract with UMD to provide certain quantum computing services and facility access, and in July 2022, we entered into a contract with UMD to provide customized quantum computing hardware.
Competition
There are many other approaches to quantum computing that use qubit technology besides the trapped ion approach we are taking. In some cases, conflicting marketing messages from these competitors can lead to confusion among our potential customer base. Large technology companies such as Google and IBM, and startup companies such as Rigetti Computing, are adopting a superconducting circuit technology approach, in which small amounts of electrical current circulate in a loop of superconducting material (usually metal where the electrical resistance vanishes at low temperatures). The directionality of the current flow, in such an example, can represent the two quantum states of a qubit. An advantage of superconducting qubits is that the microfabrication technology developed for silicon devices can be leveraged to make the qubits on a chip; however, a disadvantage of superconducting qubits is that they need to be operated in a cryogenic environment at near absolute-zero temperatures, and it is difficult to scale the cryogenic technology. Compared to the trapped ion approach, the qubits generated via superconducting suffer from short coherence times, high error rates, limited connectivity, and higher estimated error-correction overhead, ranging from 1,000:1 to 100,000:1 to realize the error-corrected qubits from physical qubits.
There are companies pursuing photonic qubits, such as PsiQuantum and Xanadu, among others. PsiQuantum uses photons (i.e., individual particles of light) as qubits, whereas Xanadu uses a combination of photons and a collective state of many photons, known as continuous variable entangled states, as the qubits. Each company’s approach leverages silicon photonics technology to fabricate highly integrated on-chip photonic devices to achieve scaling. The advantages to this approach are that photons are cheap to generate, they can remain coherent depending on the property of the photons used as the qubit, and they integrate well with recently-developed silicon photonics technology; however, the disadvantages of photonic qubit approaches include the lack of high-quality storage devices for the qubits (photons move at the speed of light) and weak gate interactions (photons do not interact with one another easily). Both of these problems lead to photon loss during computation. Additionally, this approach requires quantum error correcting protocols with high overhead (10,000:1 or more).
Several other companies use a trapped ion quantum computing approach similar to ours, including Quantinuum Ltd. and Alpine Quantum Technologies GmbH. These companies share the fundamental advantages of the atomic qubit enjoyed by our approach. The differences between our technology and that of these companies lies in our processor architecture, system design and implementation and our strategies to scale. Based on publicly available information, Quantinuum processors operate with the application circuits broken down to two qubits at a time, with a bus width of two, and the ion qubits are shuffled between each gate operation. Our processor core involves a wide-bus architecture, where the interaction among a few dozens of atomic ion qubits can be controlled using programmable laser pulses. This typically allows quantum logic gates between all possible pairs of qubits in the processor core without extraneous operations, which will enable us to operate some quantum gates that are not possible on other quantum architectures. We have also demonstrated the ability to shuttle multiple processor cores on the same chip, increasing the potential qubit capacity of a system. At scale, we believe these architectural features will confer benefits in the speed and efficiency of running algorithms. At a higher level, our scaling architecture will exploit optical interconnects among multiple QPUs in a way that allows full connectivity between any pair of qubits across the entire system. The modular scaling of multiple QPUs with photonic interconnects is unique in our architecture.
Lastly, there are alternative approaches to quantum computing being pursued by other private companies as well as the research departments at major universities or educational institutions. For example, D-Wave computing produces quantum annealers, a separate form of computing technology that hopes to tackle a class of problems with some overlap to those solved by quantum computing. Another example is QuEra, which hopes to use neutral rubidium atom arrays to build quantum computers.
Intellectual Property
We rely on a combination of the intellectual property protections afforded by patent, copyright, trademark and trade secret laws in the United States and other jurisdictions, as well as license agreements and other contractual protections, to establish, maintain and enforce rights in our proprietary technologies. Unpatented research, development, know-how and engineering skills make an important contribution to our business. We pursue patent protection only when it is consistent with our overall strategy for safeguarding intellectual property.
In addition, we seek to protect our intellectual property rights through non-disclosure and invention assignment agreements with our employees and consultants and through non-disclosure agreements with business partners and other third parties. We have accumulated a broad patent portfolio, both owned and exclusively licensed, across a range of technological fronts that relate to our systems and will continue to protect our inventions in the United States and other countries. Our patent portfolio is deepest in the area
14
of devices, methods and algorithms for controlling and manipulating trapped ions for quantum computing. Our trade secrets primarily cover the design, configuration, operation and testing of our trapped-ion quantum computers.
As of February 1, 2025, we own or license, on an exclusive basis, 132 issued U.S. patents and 204 pending or allowed U.S. patent applications, 140 issued foreign patents and 134 pending or allowed foreign patent applications, 13 registered U.S. trademarks and 11 pending U.S. trademark applications, and 23 registered international trademarks and 2 pending international trademark applications. Our issued patents expire between 2029 and 2043.
Exclusive License Agreement with the University of Maryland and Duke University
In July 2016, we entered into a license agreement with the University of Maryland (“UMD”) and Duke University (“Duke”), or the Universities, which was subsequently amended in September 2017, October 2017, October 2018, February 2021, April 2021, September 2021, January 2023, and February 2024 (as amended, the “License Agreement”), under which we obtained a worldwide, royalty-free, sublicensable license under certain patents, know-how and other intellectual property to develop, manufacture and commercialize products for use in certain licensed fields, the scope of which includes the application of the licensed intellectual property in ion trap quantum computing. The License Agreement provides an exclusive license under the Universities’ interest in all patents (and non-exclusive for other types of intellectual property), subject to certain governmental rights and retained rights by the Universities and other non-profit institutions to use and practice the licensed patents and technology for internal research and other non-profit purposes. We also entered into an exclusive option agreement (“Option Agreement”) with each of the Universities in 2016 whereby we have the right to exclusively license additional intellectual property developed by the Universities by exercising an annual option and issuing a certain number of common shares to each of Duke and UMD.
We are obligated to use commercially reasonable efforts to commercialize the inventions covered by the licensed patent rights and achieve certain milestones, including the hiring of a Chief Executive Officer, obtaining equity financing by specified times and such other milestones that we may specify in a development plan provided by us to the Universities. We have met all existing milestones as provided for in the License Agreement, have not included any additional milestones in any development plan provided to the Universities, and no longer have any obligation to submit any future development plans to the Universities. We are also responsible for the prosecution and maintenance of the licensed patents, at our expense and using commercially reasonable efforts. We have the sole right to enforce the licensed patents, at our expense.
We may terminate the License Agreement at any time for any reason with at least 90 days’ written notice to UMD. UMD and Duke may terminate the License Agreement if we enter into an insolvency-related event or in the event of our material breach of the agreement or other specified obligations therein, in each case, that remains uncured for 90 days after the date that it is provided with written notice of such breach by either university.
In consideration for the rights granted to us under the License Agreement, we issued UMD and Duke shares of our common stock.
Option Agreement with Duke University
In July 2016, we entered into an option agreement with Duke, which was subsequently amended in December 2020 and March 2021 (as amended, the “Duke Option Agreement”), under which we obtained the right to add Duke's interests in certain patents or other intellectual property to the License Agreement, including if they were developed by Jungsang Kim, Christopher Monroe, a professor at Duke and our former Chief Scientist, or Kenneth Brown, a professor at Duke, or by individuals under their respective supervision and such patents or intellectual property relates to the field of quantum information processing devices. We have added patents and other intellectual property to the License Agreement through the Duke Option Agreement. Pursuant to the terms of the Duke Option Agreement, we issued Duke shares of common stock, including shares of common stock issued pursuant to the amendment of the Duke Option Agreement. The Duke Option Agreement terminates in July 2026.
Human Capital Management
Our employees are critical to our success. As of December 31, 2024, we had a 407 person-strong team of quantum hardware and software developers, engineers, and general and administrative staff. Approximately 31% of our full-time employees are based in the greater Washington, D.C. metropolitan area and approximately 35% of our full-time employees are based in the greater Seattle, WA metropolitan area. We also engage a small number of consultants and contractors to supplement our permanent workforce. A majority of our employees are engaged in research and development and related functions, and a significant portion of our research and development employees hold advanced engineering and scientific degrees, including many from the world’s top universities.
To date, we have not experienced any work stoppages and maintain good working relationships with our employees. None of our employees are subject to a collective bargaining agreement or are represented by a labor union at this time.
15
Corporate Information
IonQ, formerly known as dMY Technology Group, Inc. III (“dMY”) was incorporated in the state of Delaware in September 2020, and formed as a special purpose acquisition company. Our wholly owned subsidiary, IonQ Quantum, Inc. (formerly known as IonQ, Inc., and referred to as “Legacy IonQ” herein), was incorporated in the state of Delaware in September 2015.
On March 7, 2021, Legacy IonQ entered into an Agreement and Plan of Merger (the “Merger Agreement”), with dMY and Ion Trap Acquisition Inc., a direct, wholly owned subsidiary of dMY (the “Merger Sub”). Pursuant to the Merger Agreement, on September 30, 2021, the Merger Sub was merged with and into Legacy IonQ with Legacy IonQ continuing as the surviving corporation following the merger, becoming a wholly owned subsidiary of dMY and the separate corporate existence of the Merger Sub ceased (the “Business Combination”). Commensurate with the closing of the Business Combination, dMY changed its name to IonQ, Inc. and Legacy IonQ changed its name to IonQ Quantum, Inc.
Our principal executive offices are located at 4505 Campus Drive, College Park, MD 20740, and our telephone number is (301) 298-7997. Our corporate website address is www.ionq.com . Information contained on or accessible through our website is not a part of this Annual Report, and the inclusion of our website address in this Annual Report is an inactive textual reference only.
Available Information
Our website address is www.ionq.com . We make available on our website, free of charge, our Annual Reports, our Quarterly Reports on Form 10-Q and our Current Reports on Form 8-K and any amendments to those reports filed or furnished pursuant to Section 13(a) or 15(d) of the Exchange Act, as soon as reasonably practicable after we electronically file such material with, or furnish it to, the Securities and Exchange Commission (the “SEC”). The SEC maintains a website that contains reports, proxy and information statements and other information regarding our filings at www.sec.gov . The information found on our website is not incorporated by reference into this Annual Report or any other report we file with or furnish to the SEC.
16