Dozens of companies, startups, and government-backed labs around the world are building quantum computers right now, each betting on different underlying technologies. The field is split between tech giants like IBM, Google, and Microsoft, a wave of well-funded startups, and major national programs in the United States, China, and Europe. Here’s who’s doing what and how far along they are.
IBM: The Largest Superconducting Machines
IBM currently holds the record for the largest superconducting quantum processor. Its Condor chip, announced in late 2023, was the first to break the 1,000-qubit barrier with 1,121 qubits. But IBM’s strategy isn’t just about raw qubit count. Its public roadmap lays out a step-by-step plan: demonstrate error correction by 2025, achieve the first example of scientific quantum advantage by 2026, and deliver the first fault-tolerant quantum computer by 2029. That 2029 machine is projected to run circuits of 100 million operations on 200 qubits, a milestone that would make quantum computing practically useful for certain problems.
By 2033 and beyond, IBM aims to scale fault-tolerant systems to 1 billion operations on up to 2,000 qubits. Its newer processors, like the 156-qubit Fez chip, already show strong performance in benchmarks, with some of the lowest error rates among superconducting machines. IBM also makes its hardware available through the cloud, giving researchers and businesses access without needing to own a machine.
Google and Its Quantum Ambitions
Google made headlines in 2019 when its Sycamore processor completed a specific calculation in roughly 200 seconds that would have taken a classical supercomputer thousands of years. Google continues to develop superconducting quantum hardware and has invested heavily in error correction, which is widely seen as the biggest hurdle to practical quantum computing. Its research teams at Google Quantum AI are focused on building a large-scale, fault-tolerant system, though the company has been less public with year-by-year roadmap details than IBM.
Microsoft: A Completely Different Approach
Microsoft is pursuing something no other major company has attempted at scale: topological quantum computing. Instead of using the standard types of qubits, Microsoft’s approach relies on exotic quantum states called Majorana zero modes, hosted in specially engineered materials known as topological superconductors. In 2025, a Microsoft team led by physicists at UC Santa Barbara unveiled Majorana 1, an eight-qubit topological quantum processor, the first of its kind. The results, published in Nature, confirmed the creation of a new state of matter.
Eight qubits is tiny compared to IBM’s machines, but topological qubits are theoretically far more resistant to errors, which could make them easier to scale. Microsoft has published a roadmap for growing this technology into a fully functional computer. In the meantime, Microsoft also operates Azure Quantum, a cloud platform that gives customers access to hardware from IonQ, Quantinuum, Rigetti, and Pasqal.
Quantum Startups and Their Technologies
A growing number of startups are building quantum computers using approaches that differ from the superconducting circuits favored by IBM and Google. The main alternatives are trapped ions, neutral atoms, and photonics.
Trapped Ions: IonQ and Quantinuum
IonQ and Quantinuum both build quantum computers by trapping individual charged atoms and manipulating them with lasers. These systems have a key advantage: every qubit can interact directly with every other qubit, eliminating the complex routing that superconducting chips require. In a benchmark comparison, a 20-qubit problem needed only 190 core operations on a trapped-ion machine, while the same problem required 570 operations on an IBM chip due to physical routing constraints.
Quantinuum’s H2-1 processor has 56 qubits with the lowest error rates in the same benchmark study, making it one of the most accurate quantum machines available. IonQ’s Forte system has 36 qubits. Trapped-ion machines are slower per operation (microseconds versus nanoseconds for superconducting chips), but their precision often compensates for the speed difference. Both companies offer access through major cloud platforms including Microsoft Azure.
Neutral Atoms: Atom Computing, QuEra, and Pasqal
Neutral atom quantum computers use arrays of individual atoms (not electrically charged) held in place by laser beams. This technology has recently produced some of the highest qubit counts in the industry. Atom Computing, a California-based startup, built a system with a 1,225-site atomic array populated with 1,180 qubits, actually surpassing IBM’s Condor in raw qubit number.
QuEra, based in Boston, published a roadmap targeting 10,000 physical qubits and 100 logical qubits. Paris-based Pasqal, which uses rubidium atoms, announced a similar goal: 10,000 physical qubits by 2026 and full fault tolerance with 128 or more logical qubits by 2028. Pasqal’s current machines contain around 100 qubits and are roughly the size of a small car. Neutral atom systems are attractive because atoms are naturally identical, which reduces manufacturing inconsistencies, and qubit counts are scaling quickly.
Photonics: PsiQuantum
PsiQuantum, based in Palo Alto, is betting on photonic quantum computing, which uses particles of light instead of matter as qubits. The company’s pitch is that photonic chips can be manufactured in existing semiconductor fabrication plants, potentially making them easier to mass-produce than other approaches. PsiQuantum has secured billions in funding and partnerships with chipmakers, though it has released fewer public benchmarks than its competitors.
Government Programs: The U.S. and China
Quantum computing development isn’t just a private-sector race. Governments see it as a strategic technology and are pouring resources into national programs.
In the United States, the federal government funds quantum research through the Department of Energy, the National Science Foundation, the National Institute of Standards and Technology, and the Department of Defense. The DOE operates five National Quantum Information Science Research Centers and supports quantum testbeds at Sandia National Laboratories (trapped ions), Berkeley Lab (superconducting systems), and Oak Ridge National Lab (quantum networking). Argonne National Lab and SLAC run quantum foundries that grow and fabricate qubits. NIST created the Quantum Economic Development Consortium to connect government, academia, and companies. The NSF funds Quantum Leap Challenge Institutes and an international research network linking U.S. universities with partners in Europe and Japan.
China takes a state-led approach, coordinating research through five-year plans that link the Chinese Academy of Sciences, state-owned enterprises, and top universities. The results have been dramatic. Researchers at the University of Science and Technology of China unveiled the Jiuzhang photonic quantum computer in 2020, which performed a specific sampling task in 200 seconds that would take a classical supercomputer an estimated half-billion years. The same university developed the Zuchongzhi superconducting system, starting at 62 qubits and later expanding to 176. China has also built major quantum communication infrastructure, including the Micius quantum satellite and a quantum communication backbone linking Beijing and Shanghai.
The European Union has its own Quantum Technologies Flagship program, and countries like the UK, France, Germany, and Japan all fund domestic quantum research initiatives.
How the Technologies Compare
No single approach has proven itself the clear winner. Each technology trades off different strengths:
- Superconducting circuits (IBM, Google, Rigetti, IQM) offer the fastest individual operations, measured in nanoseconds, but have higher error rates and require complex wiring between qubits.
- Trapped ions (IonQ, Quantinuum) deliver the highest accuracy and full connectivity between qubits, but each operation takes microseconds, roughly a thousand times slower.
- Neutral atoms (QuEra, Pasqal, Atom Computing) are scaling qubit counts rapidly and offer flexible connectivity, though the technology is less mature in commercial deployments.
- Photonics (PsiQuantum, China’s Jiuzhang) could leverage existing chip manufacturing, but faces challenges in making photons interact reliably.
- Topological qubits (Microsoft) promise built-in error protection, but the first processor has only eight qubits and the physics is still being validated.
Cloud Access Is Already Available
You don’t need to own a quantum computer to use one. Microsoft’s Azure Quantum platform connects users to hardware from IonQ, Quantinuum, Rigetti, and Pasqal. Amazon’s Braket service provides similar access to multiple vendors. IBM offers cloud access to its own processors. This means researchers, developers, and companies can experiment with real quantum hardware today, testing algorithms and building expertise while the machines continue to improve. The practical, fault-tolerant quantum computer that can solve problems no classical machine can touch is still years away, but the ecosystem of builders, technologies, and access points is already remarkably broad.