What changed in this filing

/stocks — the workspaceLOADING

Item 1. Business

2026-03-04 compared with 2025-03-07 · 143 added, 146 removed, 413 unchanged (41% of the section changed)

Our mission is to build the world’s most powerful computers to help solve some of humanity’s most important and pressing problems.

Our mission is to build the world’s most powerful computers to help solve humanity’s most important and pressing problems.

Our strategy is to be at the forefront of superconducting quantum computing.

5 unchanged sentences

When scaled, we believe these systems are poised to solve problems of staggering computational complexity at unprecedented speed.

The availability of scalable quantum computers is expected to enable scientists and engineers to address problems in areas like climate change, fusion energy, quantitative finance, drug development and discovery, materials science, and artificial intelligence.

To unlock this opportunity, we have developed the world’s first multi-chip quantum processor for scalable quantum computing systems.

We expect this patented and patent pending, modular chip architecture to be the building block for new generations of quantum processors that we expect to achieve a clear advantage over classical computers.

Our quantum computers are based on superconducting qubits, which we believe is the leading quantum computing modality based on their fast gate speeds and defined pathway to scaling.

Our quantum computers currently achieve gate speeds of 50-70 nanoseconds, which is about 1,000 times faster than other modalities such as trapped ions or pure atoms based on publicly available information.

We have developed the world’s first multi-chip quantum processor for scalable quantum computing systems.

We expect this patented and patent pending modular chip architecture to be the building block for new generations of quantum processors that we expect to achieve an advantage over classical computers.

We have already demonstrated the potential of our modular chip architecture with the launch of Cepheus-1-36Q, which is based on four 9-qubit “chiplets” tiled together.

We are a vertically integrated company.

4 unchanged sentences

We have been deploying our quantum computers to end users over the cloud since 2017.

We offer our full-stack quantum computing platform as a cloud service to a wide range of end-users, directly through our Rigetti QCS platform, and also through cloud service providers.

We offer our full-stack quantum computing platform as a cloud service to a wide range of end-users, directly through our Rigetti Quantum Cloud Services (QCS®) platform, and also through public cloud service providers.

We began selling quantum computers to end users in 2023.

In December 2023, we launched the Novera™ QPU, our first commercially available QPU, which includes a 9-qubit chip that features tunable couplers for fast 2-qubit operations and a 5-qubit chip for testing single-qubit operations.

In December 2023, we launched the Novera™ QPU, our first commercially available quantum processing unit (“QPU”), which includes a 9-qubit chip that features tunable couplers for two-qubit operations and a 5-qubit chip for testing single-qubit operations.

The Novera QPU is based on our fourth generation Ankaa™-class architecture.

In the fourth quarter of 2024, we announced the public launch of our 84-qubit Ankaa-3 system, our newest flagship quantum computer featuring an extensive hardware redesign.

We also achieved key two-qubit gate fidelity milestones with Ankaa-3:

successfully halving error rates in 2024 to achieve a 99.0% median two-qubit iSWAP gate fidelity, as well as demonstrating a 99.5% median two-qubit fidelity with fSim gates based on our internal testing.

For information on iSWAP gate fidelity and fSIM gates, see “—Our Technology—Our Superconducting Quantum Processors—Fidelity”

Ankaa-3 is available to our partners via the Rigetti Quantum Cloud Services platform (QCS ® ) and is expected to be available on Amazon Braket and Microsoft Azure in the first quarter of 2025.

Ankaa-3 is intended to enable users to operate our universal iSWAP gates for a wide range of algorithmic research, with a median gate time of 72 nanoseconds.

The more specialized fSim gates provide a median gate time of 56 nanoseconds and are useful for specific algorithms such as random circuit sampling.

For more information, see “—Our Technology—Our Superconducting Quantum Processors—Fidelity

The Ankaa-3 system features scalable chip architecture with 3D signal delivery while incorporating major enhancements to key technologies.

Leveraging our full-stack platform and in-house quantum foundry capabilities, we believe that Ankaa-3 demonstrates our ability to deliver increasingly higher performance quantum computers.

In December 2024, we sold a Novera QPU to Montana State University, which was our first QPU delivered to an academic institution.

In 2025, we received purchase orders for two Novera systems totaling approximately $5.7 million.

Both systems are upgradeable, allowing the customers to increase the system qubit count for more complex computations and research.

In the fourth quarter of 2024, we announced the public launch of our 84-qubit Ankaa-3 system, which featured an extensive hardware redesign.

We achieved a key two-qubit gate fidelity milestone with Ankaa-3:

successfully halving error rates in 2024 to achieve a 99.0% median two-qubit gate fidelity based on our internal testing.

For information on gate fidelity, see “—Our Technology—Our Superconducting Quantum Processors—Fidelity.”

In the second quarter of 2025, we announced the public launch of our 36-qubit Cepheus-1-36Q system, our newest flagship quantum computer that utilizes our modular chip architecture and demonstrates our path to scaling to higher qubit count and higher performing systems.

Made of four 9-qubit “chiplets,” we believe that Cepheus-1-36Q is the industry’s largest multi-chip quantum computer.

As of January 2026, we achieved a 99.6% median two-qubit gate fidelity (based on internal testing) with Cepheus-1-36Q, successfully halving our error rate from our previous, single-chip 84-qubit Ankaa-3 system.

Ankaa-3 and Cepheus-1-36Q are available to our partners via the Rigetti QCS platform.

Cepheus-1-36Q is intended to enable users to operate our universal CZ gates for a wide range of algorithmic research, with a median gate time of 76 nanoseconds.

Our CZ gates are designed to be optimized for fast gate times while reducing coherent errors, which improves fidelity and is key for executing quantum error correction techniques.

Cepheus-1-36Q features scalable chip architecture with 3D signal delivery while incorporating enhancements to key technologies, such as enhanced intermodule coupler design to enable higher performance.

Leveraging our full-stack platform and in-house quantum foundry capabilities, we believe that Cepheus-1-36Q demonstrates our ability to deliver increasingly higher performance quantum computers with larger qubit counts using our proprietary chiplet-based architecture.

We have developed strong customer relationships and collaborative partnerships for the purpose of accelerating the development of key technologies for high-value use cases to potentially unlock strategic market opportunities.

Our partners and customers include commercial enterprises such as Amazon Web Services (“AWS”) Standard Chartered Bank and Moody’s, along with U.S.

government organizations such as Defense Advanced Research Projects Agency (“DARPA”), Department of Energy (“DOE"), and Air Force Research Laboratory (“AFRL”) and international government entities.

In February 2024, Rigetti UK Limited, a wholly owned subsidiary of our Company, announced that it was awarded a Small Business Research Initiative grant funded by Innovate UK to develop and deliver a 24-qubit quantum computer to the National Quantum Computing Centre.

The Company is enabled by a deep technical team that includes global experts in quantum chip design and manufacturing, quantum computing systems architecture, quantum software, and quantum algorithms and applications.

government organizations such as Defense Advanced Research Projects Agency (“DARPA”), Department of Energy (“DOE”), and Air Force Research Laboratory (“AFRL”), and international government entities such as India’s Centre for Development of Advanced Computing (“C-DAC”), India’s premier R&D organization of the Ministry of Electronics and Information Technology.

In April 2025, Rigetti UK Limited, our wholly owned subsidiary, announced that it was selected as one of the winners of Innovate UK’s Quantum Missions Pilot Competition to advance quantum error correction capabilities on superconducting quantum computers.

As part of the project, we are upgrading our existing quantum computer hosted at the UK’s National Quantum Computing Centre to a larger 36-qubit quantum processing unit and integrating it with Riverlane Ltd.’s quantum error correction stack.

In January 2026, Rigetti Computing India P L, a wholly owned subsidiary of Rigetti Computing, Inc., announced that it received an $8.4 million purchase order to deliver a 108-qubit quantum computer to C-DAC.

The system will be installed on-premises at C-DAC’s Bengaluru center and is expected to be deployed in the second half of 2026.

We are focused on continuing to improve our system performance.

We recently achieved a two-qubit gate fidelity as high as 99.9% at 28 nanosecond gate speed on a prototype platform by using a new proprietary adiabatic CZ scheme.

We continue to be at 99.9% one-qubit gate fidelity.

In January 2026, we announced achievement of a median two-qubit gate fidelity (based on internal testing) of 99.7% on our 9-qubit system, 99.6% on our 36-qubit system and 99.0% on our 108-qubit system (Cepheus-1-108Q).

Cepheus-1-108Q is based on twelve 9-qubit chiplets and leverages our proprietary modular chip architecture.

We are enabled by a deep technical team that includes global experts in quantum chip design and manufacturing, quantum computing systems architecture, quantum software, and quantum algorithms and applications.

Powered by the production of our scalable multi-chip quantum processors in Fab-1 and our full-stack product development approach, we are working to develop quantum computing systems that demonstrate clear performance advantages over classical computing alternatives for multiple high-impact application areas.

Quanta Collaboration Agreement

In February 2025, our wholly-owned subsidiary, Rigetti & Co, LLC (“Rigetti Sub”), entered into a Collaboration Agreement (the “Collaboration Agreement”) with Quanta Computer Inc., a Taiwan corporation (“Quanta”), whereby the parties may enter into written statements of work from time to time pursuant to which Quanta will develop Covered Components listed in such statement of work that meet the specifications and requirements provided by Rigetti Sub.

“Covered Components” may include control systems, dilution refrigerators, flexible cables, and select other non-quantum processing unit (“QPU”) components suitable for Rigetti Sub’s quantum computing products.

No statements of work were entered into by the parties in connection with the entry into the Collaboration Agreement.

In addition, the parties have each agreed to invest at least $250 million over the next five years in the field of quantum computing (and Quanta’s investment will be towards personnel and capital expenditures for developing products and services and manufacturing capability in furtherance of the Rigetti Sub product roadmap).

Further, in connection with the Collaboration Agreement, on February 27, 2025, we entered into a securities purchase agreement (the “Securities Purchase Agreement”) with Quanta, pursuant to which we agreed to sell and issue to Quanta in a private placement transaction 3,020,412 shares of our Common Stock at a price per share of $11.58782, for an aggregate value of approximately $35.0 million.

The closing of the private placement transaction is subject to regulatory clearance.

In February 2025, we entered into a Collaboration Agreement (the “Collaboration Agreement”) with Quanta, whereby the parties may enter into written statements of work from time to time pursuant to which Quanta will develop Covered Components listed in such statement of work that meet the specifications and requirements provided by us.

“Covered Components” may include control systems, dilution refrigerators, flexible cables, and select other non-QPU components suitable for our quantum computing products.

Further, in connection with the Collaboration Agreement, in April 2025 Quanta purchased approximately $35 million of shares of Common Stock at approximately $11.59 per share, pursuant to a securities purchase agreement.

Potential Market Opportunity

1 unchanged sentence

Many of these types of problems are approached through the use of High-Performance Computing (“HPC”), which relies primarily on large classical computers located either in the cloud or on-premises.

We believe our quantum computers will be able to solve many computational problems with greater speed and at a lower cost than today’s high performance computers, thereby unlocking considerable value for the users of current HPC systems.

We believe that quantum computers will be able to solve many computational problems with greater speed and at a lower cost than today’s high-performance computers, thereby unlocking considerable value for the users of current HPC systems.

Furthermore, we believe that quantum computing will be applicable to many use cases that today lie within the realm of the much larger cloud computing market.

10 unchanged sentences

Qubits do not need the latest semiconductor lithography node and, in fact, can be made using 1990’s era lithography.

Emerging Quantum Advantage (“eQA”) Phase

This phase is characterized by the availability of practical, fully functional and operational quantum computers, whose capabilities do not yet enable them to demonstrate clear performance advantages relative to traditional computers.

Currently, our quantum computers are of sufficient scale and capability to be useful in applied research for quantum algorithm development, the exploration of potential applications of quantum computing, and for understanding the skill gaps an organization must resolve in order to be prepared to take advantage of quantum computing capabilities.

We consider the eQA phase to have begun five years ago, and during this time we have worked with business and government researchers, commercial software developers and academic institutions who access our quantum computers via cloud-based services.

We have also sold our QPUs to U.S.

national labs and others who wish to have their own quantum computer on-site and have launched Novera, our first commercially available QPU, which features a 9-qubit chip, tunable couplers for fast 2-qubit operations and a 5-qubit chip for testing single-qubit operations.

We anticipate that this phase will end when there are repeated demonstrations solving practical problems, of substantial commercial or customer value, with a level of performance that is competitive with the best available classical computing performance.

Narrow Quantum Advantage (“nQA”) Phase

If and when our quantum computing processing capabilities have scaled to the point where they can be used to solve practical, operationally relevant problems with improved accuracy, speed or cost over classical computers, we believe we will have reached the phase of nQA.

In the nQA phase, we expect that large enterprises and government organizations will increase their investment in quantum computing as the superior computational capabilities of the technology will have progressed from projected to verifiably advantaged for certain applications.

In addition to quantum-based research and development, quantum machine learning (“QML”) is likely to emerge as a strong avenue for growth as it can be leveraged in a wide range of business and scientific applications.

Research into quantum simulation and quantum optimization opportunities is predicted to increase in the nQA phase.

Broad Quantum Advantage (“bQA”)

We will consider the phase of bQA to have begun if and when our quantum computing processing abilities have scaled to the point where they can be used to solve practical problems that would be physically impossible to solve on any classical computer.

At such point, with both scaled qubit counts and strong error correction capabilities, we believe our quantum computers would be suitable for many applications of quantum machine learning and begin to be used for a growing number of quantum simulation and quantum optimization problems.

In the event we demonstrate bQA, we expect many new potential clients to emerge as the range and value of the problems that are addressable by our quantum computing systems significantly increases.

Quantum Advantage (“QA”)

We define QA as the point at which quantum computers can solve a practical problem that would be physically impossible to solve on a classical computer.

We believe that a quantum computer with over 1,000 qubits with a two-qubit gate fidelity of above 99.9% and gate speed of less than 50 nanoseconds is needed to achieve QA.

Upon achievement of QA, we believe a quantum computer would be suitable for many applications, including quantum machine learning, quantum simulation, and quantum optimization problems.

If QA were to be demonstrated, we would expect a meaningful number of new potential clients to emerge, as the range and value of the problems that would be addressable by quantum computing systems would significantly increase.

Large-Scale Fault Tolerant Quantum Computing (“LFTQC”)

We will consider the phase of lFTQC to begin if and when systems are available with hundreds of logical qubits, which can be universally controlled and measured with substantially error-free operation through the full course of a quantum computation.

It is currently believed in the quantum computing industry that this likely requires systems with 10,000 to 1,000,000 physical qubits.

We believe our scalable multi-chip architecture paves the way to scale up to these large systems.

We will consider the phase of LFTQC to begin if and when quantum computing systems are available with hundreds of logical qubits, which can be universally controlled and measured with substantially error-free operation through the full course of a quantum computation.

It is currently believed in the quantum computing industry that this is likely to require systems with 10,000 to 1,000,000 physical qubits.

We believe our multi-chip architecture provides a pathway to scale up to these large systems.

We anticipate the beginning of the large-scale fault tolerant phase to be roughly a decade away.

As quantum computing further matures through this phase, systems will likely continue to grow in scale and performance, culminating in full-scale fault tolerance that operates using potentially thousands of effectively perfect logical qubits.

As quantum computing matures through this phase, systems would likely continue to grow in scale and performance, culminating in full-scale fault tolerance that operates using potentially thousands of effectively logical qubits.

This ultimate goal of full-scale fault tolerance represents the largest commercial opportunity.

Business Strategy

Our approach to developing and sustaining what we believe is a strong competitive advantage relies on a four-pronged strategy:

Our approach to developing and sustaining what we believe is a strong competitive advantage relies on a six-pronged strategy:

● Develop superconducting gate-based quantum computers based on advantages in gate speeds and scalability.

We believe the superconducting modality of quantum computing offers advantages in scalability and gate speeds as opposed to other modalities.

Based on publicly available information, superconducting quantum computers have gate speeds that are about 1,000 times faster than other modalities such as trapped ions and pure atoms.

We believe fast gate speeds are important because they enable quantum computers to operate efficiently in a hybrid computing environment.

We also believe that a superconducting modality utilizing a multi-chip architecture provides a defined pathway to scaling up to the large systems needed for LFTQC.

● Create high-performance quantum computing systems through full-stack product development.

2 unchanged sentences

This was underscored by our announcement of the industry’s first multi-chip quantum processor for scalable quantum computers, a capability realized through many innovations from Fab-1.

● Utilize an open modular architecture that allows for integration of innovative solutions.

We have adopted an open modular architecture for our quantum computers that allows for integration of innovative solutions developed by third parties into our technology stack.

We believe that customers value the flexibility provided by an open modular architecture and that our approach will provide us with an advantage over competitors who utilize a closed architecture.

● Provide broad access to our quantum computers.

We sold our first QPU in 2023 and in December 2023 launched Novera ™, our first commercially available QPU, which features a 9-qubit chip with tunable couplers for fast 2-qubit operations and a 5-qubit chip for testing single-qubit operations.

We sold our first QPU in 2023 and in December 2023 launched Novera ™, our first commercially available QPU, which features a 9-qubit chip with tunable couplers for fast two-qubit operations and a 5-qubit chip for testing single-qubit operations.

In 2021, we began selling full-scale quantum computing systems, supporting national laboratories and quantum computing centers.

We have been providing cloud access to our quantum computers since 2017 and have since expanded the availability of our machines through distribution agreements with other solution providers, including Amazon Bracket among others.

10 unchanged sentences

We intend to continue deeply investing in finding and fostering the talent required to remain at the forefront of quantum computing innovation, while protecting our growing base of intellectual property.

Key achievements in 2024 include the launch of the 84 qubit Ankaa™-3 system to customers via Rigetti Quantum Cloud Services (QCS).

Ankaa-3 is our newest flagship quantum computer featuring an extensive hardware redesign that is intended to enable superior performance.

We also achieved key two-qubit gate fidelity milestones with Ankaa-3:

successfully halving error rates in 2024 from our error rates in 2023 to achieve a 99.0% median two-qubit iSWAP gate fidelity, as well as demonstrating a 99.5% median two-qubit fidelity with fSim gates based on our internal testing.

For information on iSWAP gate fidelity and fSIM gates, see “—Our Technology—Our Superconducting Quantum Processors—Fidelity”.

Ankaa-3 is available to our partners via the Rigetti Quantum Cloud Services platform (QCS ® ) and is expected to be available on Amazon Braket and Microsoft Azure in the first quarter of 2025.

Ankaa-3 is intended to enable users to operate iSWAP gates for a wide range of algorithmic research, with a median gate time of 72 nanoseconds.

The more specialized fSim gates provide a median gate time of 56 nanoseconds and are useful for specific algorithms such as random circuit sampling.

For information, see “—Our Technology—Our Superconducting Quantum Processors—Fidelity”.

The Ankaa-3 system continues to feature our scalable chip architecture with 3D signal delivery while incorporating major enhancements to key technologies.

Leveraging our full-stack platform and in-house quantum foundry capabilities, we believe Ankaa-3 demonstrates our ability to deliver increasingly higher performance quantum computers.

In 2025, we plan to introduce the next generation of our modular system architecture, while aiming to continue to increase fidelities.

By mid-year 2025, we expect to release a 36-qubit system based on four 9-qubit chips tiled together, with a target 2x reduction in error rates from our error rates achieved at the end of 2024.

By the end of 2025, we expect to release a system with over 100 qubits with a targeted 2x reduction in error rates from our error rates achieved at the end of 2024.

We will continue to pursue sales of Novera™ , our first commercially available QPU, which features a 9-qubit chip, tunable couplers for fast 2-qubit operations and a 5-qubit chip for testing single-qubit operations.

We believe that we will be able to achieve our plans for 2025 described above and elsewhere in this Annual Report on Form 10-K;

We believe that we will be able to achieve our plans described above and elsewhere in this Annual Report on Form 10-K;

however, we face various risks and uncertainties relating to our business that could cause actual results to differ materially from our expectations stated herein.

7 unchanged sentences

Our QPUs are designed and fabricated at Fab-1, leveraging novel manufacturing methods to create state-of-the art superconducting qubits.

Novera™ , our first commercially available QPU, includes a 9-qubit chip that features tunable couplers for fast 2-qubit operations and a 5-qubit chip for testing single-qubit operations.

Novera™, our first commercially available QPU, includes a 9-qubit chip that features tunable couplers for fast two-qubit operations and a 5-qubit chip for testing single-qubit operations.

The Novera™ QPU is based on our fourth generation Ankaa-class architecture.

We announced our most technically advanced QPU yet, the 84-qubit Ankaa-3, featuring an extensive hardware redesign.

We announced our most technically advanced QPU yet based on our proprietary chiplet-based architecture, the 36-qubit Cepheus-1-36Q system, which we believe is the industry’s largest multi-chip quantum computer.

We have also released our 84-qubit Ankaa-3 quantum computer, which is available on Amazon Braket.

We intend to design and fabricate more advanced QPUs in the future with improved fidelities, faster gate speeds and higher qubit counts.

We believe these anticipated improvements and advances in technology will hopefully lead to nQA, bQA and LFTOC in the coming years.

We believe these anticipated improvements and advances in technology will hopefully lead to QA and LFTQC in the future.

Quantum Computing as a Service (QCaaS)

6 unchanged sentences

The product is designed to meet the needs of a diverse set of customers that all benefit from the high-performance nature of its core computational capabilities.

Central to QCS are two very powerful sets of technologies developed by us, our quantum processing units (“QPUs”), described above, and our quantum operating system, as described below:

Central to QCS are two very powerful sets of technologies developed by us, our QPUs, described above, and our quantum operating system, as described below:

Quantum Operating System Software

4 unchanged sentences

● Facilitating the implementation of high performance public and private clouds with ultra-low latency connectivity between classical hardware and our QPUs.

Customers who purchase an on-premises quantum system from us also have access to QCS Outpost, our distributed software environment for operating, administering, and monitoring the overall system.

In addition, QCS Outpost includes utilities for QPU characterization and calibration, user access management, quantum program compilation, scheduling, and execution.

QCS Outpost also serves as the foundation for integrating with other systems, in particular high-performance computing systems, and can be accessed through a variety of tools and web-based services as well as through software development kits (SDKs) that support Python, C, and Rust.

Direct QCaaS Distribution

10 unchanged sentences

For example, one cloud provider or HPC operator might need deep and high-performance integration with a specific machine learning service they provide, while another might desire a fast and easy way for small customers to be introduced to quantum computing.

We have signed a distribution agreement with Amazon’s Braket service and Microsoft’s Azure Quantum Service, providing access to our quantum computing systems to AWS and Azure customers.

We have distribution agreements with Amazon’s Braket service and Microsoft’s Azure Quantum Service, providing access to our quantum computing systems to AWS and Azure customers.

Key Technology Development Partnerships

6 unchanged sentences

DOE’s Superconducting Quantum Materials and Systems Center (“SQMS”), to advance the development of scalable and high performance quantum processors;

● AFRL to harness our fabrication capabilities for quantum networking hardware research and development.

● AFRL to harness our fabrication capabilities for quantum networking hardware research and development, and to advance superconducting quantum computing networking;

● DARPA and National Aeronautics and Space Administration (“NASA”) to create quantum computing systems, software and algorithms for optimization applications;

● Innovate UK, as part of the British government’s effort to accelerate commercialization of quantum computing in the United Kingdom and to pursue practical applications in machine learning, molecular simulation and financial optimization.

● Quanta for control systems, dilution refrigerators, flexible cables, and select other non-QPU components .

We expect to add new development partnerships as the capabilities of our quantum computer systems grow and the market’s readiness and interest in quantum computing continues to mature.

11 unchanged sentences

optimization, machine learning, simulation and quantum mechanical system simulation.

The computational properties of a quantum computer inherently support the problem-solving requirements of extremely complex optimization calculations because quantum computers possess the ability to simultaneously evaluate very large numbers of variables, and each additional qubit in a quantum computer exponentially scales our information processing capacity.

We believe that quantum computers could allow highly accurate optimization models to be continuously refreshed to reflect the impact of changing conditions on available solutions, ultimately leading to better and more responsive plans and decision-making.

Many of the world’s largest and most valuable industries could potentially benefit from enhanced optimization enabled by quantum computing.

In financial services, optimization could be applied to portfolio management, algorithmic trading and risk assessment.

In telecommunications, optimization could be applied to call routing and network capacity planning.

In manufacturing, optimization could help with workforce, warehouses and supply chain planning.

In transport, there are logistics applications like fleet routing, driver scheduling, and package loading and delivery that could benefit from further optimization.

These types of problems can quickly overwhelm classical computers due to the large numbers of variables that need to be evaluated, which exponentially scales the computational power required with each additional possibility to be considered.

For example, in a vehicle routing problem involving roundtrips to just 10 destinations there can be more than 300,000 permutations to be considered;

with 15 destinations, the number of possible routes exceeds 87 billion.

If you factor in other real-world considerations such as delivery cost, fleet size, driver availability, or service level agreements, you can see the intractable nature of these kinds of problems in full display.

One of the most active fields of quantum algorithm research is the area of constrained combinatorial optimization.

These mathematical equations can arrive at approximate results with a close-to-optimal solution across many possible outcomes-a result that would create high value in many different industries, particularly when the exact solution is unknowable utilizing a classical computer.

In September 2023, we were awarded a DARPA project as part of the Imagining Practical Applications for a Quantum Tomorrow (IMPAQT) program to advance the state-of-the-art in quantum algorithms for solving combinatorial optimization problems.

Our project, “Scheduling Problems with Efficient Encoding of Qubits” (SPEEQ), seeks to develop a novel and efficient encoding of optimization problems onto qubits, with the goal of enabling larger problems to be mapped to currently available NISQ-era quantum computers.

The project will specifically address scheduling problems, which are among the best-known and most pervasive types of combinatorial optimization problems across numerous industries, as well as some of the most challenging to solve.

In November 2023, we were awarded Phase 2 of the DARPA Quantum Benchmarking Program to develop benchmarks for quantum application performance on large-scale quantum computers.

The goal of the DARPA Benchmarking Program was to create key quantum computing metrics for fault tolerant quantum computing, make those metrics testable, and estimate the required quantum and classical resources needed to reach critical performance thresholds.

Rigetti was awarded Phase 1 in March 2022.

The key output of Phase 1 of this program was the development of a resource estimation framework to provide insight into the requirements of a superconducting quantum computing system necessary for solving large-scale, complex problems.

Phase 2 is expected to entail refining and optimizing our estimates for selected utility-scale problems, delivering new upper bounds on these requirements.

Another anticipated benefit of this resource estimation framework is to enable a cost benefit analysis into whether the resources needed to run a quantum application will be met by the value of solving the particular problem.

A challenge in developing quantum algorithms is understanding how a problem will scale, and at what point a dataset is large or complex enough to benefit from the unique properties of quantum computing.

Estimating the amount of time, the number of qubits, and the energy required could accelerate the work towards designing an optimized algorithm.

Phase 2 is expected to be heavily focused on researching fault-tolerant quantum applications.

Of particular interest are dynamical chemistry simulations and modeling the dynamics of quantum systems.

Discrete optimization, also known as combinatorial optimization, focuses on problems where variables are restricted to specific, discrete values (e.g., 0 or 1).

Unlike continuous variables, which can assume any value, discrete variables are constrained.

The primary goal of this field is to determine the optimal assignment of values to these discrete variables to solve a given problem.

This is typically achieved by formulating the problem around an objective function—a mathematical construct that one seeks to either minimize or maximize through the selection of the variables' values.

Such optimization problems are ubiquitous, spanning areas like logistics, routing, manufacturing, scheduling, telecommunication, energy, chemistry, biology, physics, finance, and basic science.

For example, in financial services, optimization can be applied to portfolio management, algorithmic trading, and risk assessment.

In telecommunications, optimization can be applied to call routing and network capacity planning.

In manufacturing, optimization can help with workforce, warehouses, and supply chain planning.

In transport, there are logistics applications like fleet routing, driver scheduling, and package loading and delivery that can benefit from further optimization.

In energy, optimization can be applied to effectively deliver power distribution over the grid.

In biology, optimization can be applied to accelerate or improve drug discovery processes.

Optimization problems can be computationally intensive, and the run-time required for classical solvers to identify an optimal solution increases exponentially with the number of variables.

For problems of meaningful scale, obtaining a provably optimal solution is effectively infeasible, necessitating the use of estimated or approximate solutions.

As a result, optimization in practice focuses on identifying the best attainable solution within practical run-time limitations, which are shaped by industry-specific factors.

Because even marginal improvements in solution quality can translate into meaningful gains in areas such as cost, timing, or resource allocation, there is demand for the development of more efficient and accurate optimization methods.

Quantum computers introduce a fundamentally different computational model that extends beyond classical computer systems.

It leverages distinctive mechanisms—such as entanglement, superposition, and interference—to enable new algorithmic strategies for addressing complex optimization problems.

These mechanisms underpin a range of quantum algorithmic frameworks, including quantum adiabaticity, variational quantum circuits, and quantum interferometry, among others, which may be applied in optimization contexts.

While certain quantum optimization algorithms are designed to offer mathematical performance guarantees, others are developed as heuristics.

Whether quantum optimization algorithms can outperform current state-of-the-art classical methods in terms of accuracy or run-time remains an open question.

We have conducted research in the field of quantum optimization for several years, pioneering original work and collaborating with experts to develop, understand, benchmark, and apply quantum optimization algorithms.

We have examined the role of entanglement in quantum optimization, including its tradeoff with hardware noise, and have identified instances in which quantum optimization exhibits capabilities that exceed those of classical approaches.

We have invented several quantum algorithms, and in 2025 we unified these techniques under the umbrella of quantum preconditioning.

Quantum preconditioning is a quantum-boosted heuristic algorithm, which uses the quantum computer to modify a problem in a manner that makes it more readily solvable on a classical solver.

We investigated the potential for QA on standard benchmark problems as well as an energy-grid optimization problem, targeting the QA window described above.

We have also investigated other approaches to optimization, such as a qubit-efficient solver, a quantum-based multilevel approach, and quantum algorithms with co-designed circuits or logical gates.

In addition to numerous peer-reviewed publications in leading journals, our works have been presented at a number of conferences nationally and internationally.

Our expertise in quantum optimization has been recognized through our participation in U.S.

government programs such as DOD’s DARPA ONISQ (2020-2024), DOD’s DARPA IMPAQT (2023-2024), and DOE’s National Quantum Initiative with the superconducting quantum materials and systems center.

Machine Learning

Machine learning is a well-established field, with broad application, that today is already having a transformative impact on a myriad of markets.

At the core of any machine learning application is a series of computations, typically expressed in linear algebra, applied to vast amounts of data in order to do things such as reliably classifying objects and making data-driven predictions.

Today, cloud computing and HPC have been the predominant sources of the computational capabilities required to create effective machine learning algorithms, models and data analysis applications.

But the efficiency of HPC-powered machine learning algorithms is limited when faced with richer and larger data sets.

For that reason, computer scientists have looked toward the computational promise of quantum computers, and the development of quantum-based algorithms, as a means of both accelerating current machine learning algorithms and creating new approaches that are currently impossible on classical computers.

Given these factors, the emerging field of QML is the focus of much of the current research and development occurring on quantum computers today.

Machine learning is a well-established computer science field that is already having a transformative impact on a myriad of markets today.

At the core of any machine learning algorithm is a series of computations, typically linear algebra, applied to vast amounts of data that can reliably classify objects in pictures and make data-driven forecasts, for example.

Today, cloud computing and HPC have been the predominant sources of the computational capabilities required to create and deploy effective machine learning algorithms, models and data analysis applications.

When faced with increasing amounts of data and while trying to grasp more complex patterns, the energy consumption of HPC-powered machine learning systems may become exorbitant due to computational and cooling demands.

For that reason, computer scientists have looked toward the promise of higher computational efficiency of quantum computers, and the development of quantum machine learning (“QML”) algorithms, as a means of both accelerating current machine learning applications and creating new approaches that are currently impossible with classical computers and may lead to more efficient and accurate models.

Given these factors, the emerging field of QML is the focus of much of the current quantum computing research.

We already see emerging machine learning algorithms that take advantage of the unique capabilities of quantum computing to tackle the complex linear algebra problems at the heart of many machine learning tasks.

9 unchanged sentences

In November 2023, we were awarded an Innovate UK grant as part of the Feasibility Studies in Quantum Computing Applications competition.

Joining us in this work are Amazon Web Services (AWS), Imperial College London, and Standard Chartered.

The consortium aims to use quantum computing to improve current classical machine learning techniques used by financial institutions to analyze complex data streams.

Joining us in this work were Amazon Web Services (AWS), Imperial College London, and Standard Chartered.

The consortium aimed to use quantum computing to improve current classical machine learning techniques used by financial institutions to analyze complex data streams.

Financial institutions need to continuously interpret complex data streams to extract information necessary for providing accurate credit risk evaluation, managing market-making services, and predicting emissions in the context of green finance, among other things.

1 unchanged sentence

Combining quantum computing with classical machine learning methodology could offer more powerful resources for processing these data streams, given the potential for quantum computers to process some types of information more efficiently than with classical resources alone.

The aim of the consortium is to address the following research objectives:

(1) further develop quantum signature kernels and quantum-enhanced feature maps, (2) benchmark the results against classical machine learning methods for streamed data, and (3) build and study quantum algorithms for computing signatures and signature kernels for long and high-dimensional data streams efficiently.

In October 2023, we were awarded a separate Innovate UK grant as part of the Feasibility Studies in Quantum Computing Applications competition.

Joining us in this work are HSBC, the Quantum Software Lab (QSL) based at the University of Edinburgh, and the National Quantum Computing Centre (NQCC).

Together, the consortium aims to enhance existing anti-money laundering techniques by using quantum machine learning techniques with the goal of improving the performance of current-state-of-the-art machine learning algorithms.

Money laundering poses a significant threat to financial institutions and society.

Machine learning technology has the power to detect and prevent financial crime by flagging suspicious transactions and adapting to ever-changing criminal behavior.

Quantum computing has the potential to enhance existing classical computing workflows, and in turn could offer improved machine learning methods.

In this work, the consortium will aim to extend current anomaly detection quantum machine learning models to detect anomalous behavior indicating money laundering.

Joining us in this work were HSBC, the Quantum Software Lab (QSL) based at the University of Edinburgh, and the National Quantum Computing Centre (NQCC).

Together, the consortium aimed to enhance existing anti-money laundering techniques by using quantum machine learning techniques with the goal of improving the performance of current-state-of-the-art machine learning algorithms.

In this work, the consortium aimed to extend current anomaly detection quantum machine learning models to detect anomalous behavior indicating money laundering.

Classical computers have been used for decades in critical applications that model real-world processes or systems in order to study their behaviors over time.

43 unchanged sentences

Achieving quantum advantage imposes requirements on the quantum computer itself, the most important of which relate to the above performance factors of scale , fidelity , speed , co-processing , and re-programmability .

In order for quantum computers to solve problems out of reach for classical computers, such as modeling molecules with many electrons in order to enhance drug discovery, they require a significant number of high-performing qubits, likely starting at between a few hundred to 1,000 qubits.

A gate fidelity estimates the reliability of an operation.

1 unchanged sentence

Errors can be caused by imperfect control, natural manufacturing variations, finite qubit lifetimes (coherence) or other sources.

Overall, high fidelities of over 99% are likely necessary to enable performance benefits on practical workloads.

Overall, high fidelities of close to 99.9% are likely necessary to enable performance benefits on practical workloads.

An error per operation is defined as (1-fidelity).

4 unchanged sentences

Hybrid architectures that leverage quantum computers as co-processors, pioneered by us since the company’s inception, have now become widely adopted in the quantum computing industry.

Quantum co-processing delivered over the cloud, such as Rigetti Quantum Cloud Services platform, is the predominant framework for building and using quantum computers today.

Quantum co-processing delivered over the cloud, such as our QCS platform, is the predominant framework for building and using quantum computers today.

In this paradigm, quantum processors are tightly integrated with classical computing systems and infrastructure to ensure the rate of data flowing in and out of the quantum processor can meet the needs of commercial applications.

16 unchanged sentences

As an example, in June 2011, the largest algorithms demonstrated on programmable, gate model quantum computers across these modalities were in the range of a few qubits.

In the ensuing twelve-year period from 2012 to 2024, superconducting systems have successfully scaled up to the range of 100 or more qubits, including demonstrations of quantum supremacy.

In the ensuing period from 2012 to 2025, superconducting systems have successfully scaled up to over 100 qubits, including demonstrations of quantum supremacy.

This rate of scaling has easily outpaced other approaches.

49 unchanged sentences

There are a variety of metrics that are used to measure these errors;

we currently report performance and indicate a measure of error through a fidelity metric applied to 2-qubit gate error or fidelity, usually expressed as a percentage.

we currently report performance and indicate a measure of error through a fidelity metric applied to two-qubit gate error or fidelity, usually expressed as a percentage.

Gate fidelity represents the reliability of an operation.

For example, a 2-qubit gate with a fidelity of 99% means that 99 out of 100 times the measurement of the gate will produce the correct result.

For example, a two-qubit gate with a fidelity of 99% means that 99 out of 100 times the measurement of the gate will produce the correct result.

Fidelities are related to errors in the following way:

2 unchanged sentences

There are a number of standard benchmarks that are used to measure qubit errors and are explained further below.

We measure the performance of iSWAP gates with an industry standard technique called Randomized Benchmarking, a commonly used method to measure fidelity.

We measure the performance of iSWAP and CZ gates with an industry standard technique called Randomized Benchmarking, a commonly used method to measure fidelity.

This protocol requires creating random sequences of quantum gates of different sequence length, executing each sequence, and then measuring the outcome of the execution against the mathematically expected results.

We also implement a family of 2-qubit gates referred to as fSim.

On iSWAP-enabled devices, such as Ankaa-3, we also implement a family of two-qubit gates referred to as fSim.

Generally, any specific fSim gate may not be part of a universal gate set.

2 unchanged sentences

This protocol requires creating random circuits from the provided gate set measuring the results and comparing the outcomes to an expected probability distribution of outcomes.

In the past we have implemented gate sets based on 2-qubit gates other than iSWAP and fSIM, and may, in the future, choose different gate sets.

In the past we have implemented gate sets based on two-qubit gates other than CZ, iSWAP and fSIM, and may, in the future, choose different gate sets.

At the moment there is no standard set of gates agreed on in the industry, and there may never be.

1 unchanged sentence

Accordingly, undue reliance should not be placed on the fidelity measures that we present.

See also “Risk Factors— If our computers fail to achieve quantum advantage, our business, financial condition and future prospects may be harmed.

Moreover, the standards by which we measure our progress may be based on assumptions and expectations that are not accurate or that may change as quantum computing evolves .”

Ankaa-3 is Rigetti’s newest flagship quantum computer featuring an extensive hardware redesign that is intended to enable superior performance.

With Ankaa-3 we successfully halved our error rates in 2024 from our error rates in 2023, achieving a 99.0% median two-qubit fidelity with iSWAP gates and a 99.5% median two-qubit fidelity with fSim gates based on internal testing.

Ankaa-3 is designed to enable users to operate the iSWAP gates for a wide range of algorithmic research, with a median gate time of 72 nanoseconds.

The more specialized fSim gates provide a median gate time of 56 nanoseconds and are useful for specific algorithms such as random circuit sampling.

Improving our median 2-qubit fidelities is a crucial part of our mission to build the world’s most powerful computers.

See also “Risk Factors— We face significant technical and engineering challenges in completing the development of our quantum computers, producing our quantum computers at scale, achieving our targeted performance milestones, and realizing quantum advantage or LFTQC, any of which if not accomplished would adversely impact our business, financial condition, and results of operations.”

Cepheus-1-36Q is our newest flagship quantum computer featuring our proprietary modular chip architecture, optimized two-qubit gates and advances in intermodule coupler design that is intended to enable superior performance.

With Cepheus-1-36Q, we successfully halved our error rates from our previous Ankaa-3 system, achieving a median two-qubit fidelity of 99.6% (based on internal testing) as of January 2026.

We believe Cepheus-1-36Q is the first multi-chip quantum computer in the industry to achieve this level of performance based on publicly available information.

Improving our median two-qubit fidelities is a crucial part of our mission to build the world’s most powerful computers.

Useful quantum computers will need not only a large number of qubits, but also high-quality qubits.

Reaching 99.0% median two-qubit fidelity with iSWAP gates and a 99.5% median two-qubit fidelity with fSim gates on the Ankaa-3 system in 2024 based on our internal testing is the result of years of innovation and commitment from our teams across the technology stack.

We have already designed and deployed a modular architecture, tiling multiple chips together demonstrating what we believe is the way forward towards building larger systems.

Reaching 99.6% median two-qubit fidelity on the Cepheus-1-36Q system is the result of years of innovation and commitment from our teams across the technology stack.

We believe that Cepheus-1-36Q validates our modular architecture approach to scaling.

Tiling multiple chips together demonstrates what we believe is the way forward towards building larger systems.

We believe a densely connected square lattice with tunable couplers that allows us to control qubit interactions is the foundation for driving qubit performance.

2 unchanged sentences

The speed of gate operations in superconducting qubits are determined by the intentional design of circuit elements on-chip and their optimized parameters, rather than relying on atomic properties.

Our recently introduced Ankaa-3 system achieves a median gate time of 72 nanoseconds with universal iSWAP gates.

A median gate time of 56 nanoseconds was achieved with the more specialized fSim gates.

Our recently introduced Cepheus-1-36 system achieves a median gate time of 76 nanoseconds with universal CZ gates.

Median gate time is measured by internal testing.

21 unchanged sentences

For example, re-setting registers of qubits conditional on the outcomes of previous measurements, can increase overall quantum circuit throughput by 5x relative to non-dynamic implementations of the same workload.

The QPU control system includes hardware for networking, classical microprocessors, FPGAs for control and readout pulse sequencing, and analog signal processing.

The QPU control system includes hardware for networking, classical microprocessors, field-programmable gate arrays (FPGAs) for control and readout pulse sequencing, and analog signal processing.

The integrated system is designed and built to meet the requirements for co-processing and reprogrammability over the cloud.

15 unchanged sentences

Planar codes are expected to show a high error threshold of approximately 1% error probability per operation.

This means that if error rates are below the required threshold (e.g.

1%), then increasing the redundancy ( i.e.

This means that if error rates are below the required threshold (e.g., 1%), then increasing the redundancy ( i.e.

, the number of physical qubits making up a single logical qubit) results in an exponential reduction in logical error.

14 unchanged sentences

As of December 31, 2025, we have 121 patents issued and 160 patents pending that are designed to protect our full-stack technology across hardware, software, and services.

These patents cover a broad range of key technology areas of the business including (i) quantum computing systems, software and access;

These patents cover a broad range of key technology areas of the business including:

(i) quantum computing systems, software, and access;

(ii) quantum processor hardware;

4 unchanged sentences

Sales & Marketing

During this period of eQA, our go-to-market strategy is focused on being a leader in the key market segments driving the early application of quantum computing.

During this period of emerging quantum advantage, our go-to-market strategy is focused on being a leader in the key market segments driving the early application of quantum computing.

Our sales and marketing efforts are focused on technology development and distribution partnerships with the leading organizations in these markets.

government, for example, the Departments of Defense and Energy have each been making significant investments in quantum computing, and we have technology development partnerships with leading agencies and national laboratories.

government, for example, the Departments of Defense and Energy have each been making significant investments in quantum computing.

We have technology development partnerships with leading government agencies and national laboratories.

We are pursuing similar arrangements with customers in other important vertical market segments, like finance, where we are developing specific expertise in several application areas and are collaborating with Moody’s, HSBC and Standard Chartered Bank.

We also have distribution relationships with customers like Amazon Web Services, Microsoft Azure and Strangeworks.

In connection with our reorganization announced in February 2023, we reduced our investment and expenses in sales and marketing to focus our resources on technology development.

As we work to develop new generations of our hardware with the goal of continuing to scale and achieve nQA and then BQA, we anticipate increasing our investment and expenses in both sales and marketing in the future to expand the number of enterprise companies buying our QPUs and directly licensing our QCS platform.

We source our components from multiple industries including:

from the electronics and semi-conductor industries with low-noise microwave components, CPUs, GPUs, FPGAs;

As we work to develop new generations of our hardware with the goal of continuing to scale and achieve QA, we anticipate increasing our investment and expenses in both sales and marketing in the future to expand the number of enterprise companies buying our QPUs and directly licensing our QCS platform.

We source our components from multiple industries including from the electronics and semi-conductor industries with low-noise microwave components, CPUs, GPUs, FPGAs;

from the cryogenic industry with dilution refrigerators and associated helium gas products;

10 unchanged sentences

● We also enter into multi-year technology development partnerships with organizations that possess specialized technical expertise and strong interests in advancing the development of quantum computing (as referenced in Business - Key Technology Development Partnerships ).

These organizations include DARPA, SQMS, and Innovate UK.

These organizations include DARPA, SQMS, Innovate UK and Quanta.

The quantum computing market is evolving and highly competitive.

3 unchanged sentences

In addition, because of the importance of quantum computing, most large public cloud providers and traditional chip makers are researching and investing in quantum computing initiatives, in some cases seeking to build quantum computers.

For example, Amazon and Intel are engaged in the research and development of quantum computers.

For example, Amazon is engaged in the research and development of quantum computers.

A number of development-stage companies are also seeking to build quantum computers, quantum software and applications, and quantum cloud computing services.

14 unchanged sentences

We believe that we are favorably positioned to compete on the basis of these factors.

However, we face various risks relating to competition as described in “ Risk Factors-Risks Related to Rigetti’s Business and Industry-The quantum computing industry is competitive on a global scale and Rigetti may not be successful in competing in this industry or establishing and maintaining confidence in our long-term business prospects among current and future partners and customers .”

However, we face various risks relating to competition as described in “ Risk Factors-Risks Related to Our Business and Industry-The quantum computing industry is in its early stages and volatile and is competitive on a global scale and we may not be successful in competing in this industry or establishing and maintaining confidence in our long-term business prospects among current and future partners and customers .”

government contracts, grants, and agreements are subject to regulations and procurement laws.

43 unchanged sentences

Compared sentence by sentence after normalising whitespace, quotation marks, case and digits, so re-formatting and restated figures do not read as changed language. Wording changes appear as one removal and one addition. The current filing and the prior one are authoritative.

What changed 10-K

Item 1. Business