Insider Brief
- Quantum computing hardware remains divided across multiple competing chip modalities in 2026, with no single approach establishing clear dominance.
- Major companies are advancing superconducting, trapped-ion, photonic, neutral-atom, silicon spin qubit, and annealing architectures, each with distinct technical advantages and manufacturing challenges.
- The industry continues to invest across several hardware approaches as researchers and companies work toward scalable, fault-tolerant quantum computing
Quantum computing hardware in 2026 looks like several distinct technologies, each with different physics, different manufacturing constraints, and different ideas about what fault-tolerant quantum computing should look like when it arrives.
Superconducting qubits, trapped ions, photonics, neutral atoms, and silicon spin qubits are all in active commercial development. No modality has established the kind of dominance that x86 holds in classical computing. That reflects genuine technical uncertainty – the field has not yet reached the point where the performance gaps between approaches are large enough to foreclose alternatives. What it means in practice is that the hardware landscape remains unusually competitive, and the companies in it are making different bets.
This article covers the companies building quantum chips today – what they have shipped, what their hardware does, and what distinguishes their approach.
Companies Building Quantum Computing Chips
The following is a non-exhaustive selection of companies. The landscape is broad and evolving rapidly, and the inclusion or omission of any entry should not be interpreted as a ranking or endorsement.
IBM
IBM’s quantum hardware roadmap is anchored by the Heron processor family. The current generation – Heron r1 (133 qubits) and Heron r2/r3 (156 qubits) – uses fixed-frequency transmon qubits with tunable couplers, an architecture IBM developed specifically to reduce crosstalk errors that affected earlier designs. The r3 variant, released in beta in July 2025 with the ibm_pittsburgh system, achieved IBM’s best coherence and readout fidelity to date across the Heron line.
The Nighthawk processor, which features 120 qubits on a square lattice with four-degree connectivity, represents the next architectural direction – a topology designed to increase workload complexity beyond what heavy-hex lattices support. Condor, at 1,121 qubits, serves as a scaling milestone rather than a production system, demonstrating that the physical qubit counts required for error-corrected computing are achievable.
Access is available via IBM Quantum, which gives researchers and enterprise customers cloud access to production systems.
Google’s Willow processor is a 105-qubit superconducting system that produced two significant results in a short period. In December 2024, Willow demonstrated below-threshold quantum error correction – the first time exponential error suppression was observed as qubit counts increased rather than degraded, a milestone researchers had been working toward for years.
In October 2025, Google announced what it described as the first verifiable quantum advantage using Willow and a new algorithm called Quantum Echoes – technically an out-of-time-order correlator (OTOC). The result showed Willow running the algorithm 13,000 times faster than the best classical algorithm on one of the world’s fastest supercomputers. Google framed the result as the first quantum advantage on a verifiable, reproducible algorithm with real-world scientific applications – distinguishing it from the 2019 random circuit sampling result, which used a problem specifically constructed to be difficult classically. Willow is accessible through a proposal-based research access program in partnership with the UK government.
Intel
Intel’s Tunnel Falls processor is a 12-qubit silicon spin qubit chip, distinct from the superconducting approach used by IBM, Google, and others. Silicon spin qubits encode quantum information in the spin states of electrons confined in silicon structures, leveraging the same manufacturing infrastructure Intel uses for classical chips. The manufacturing compatibility is the strategic rationale – Intel’s thesis is that volume production at advanced nodes will eventually provide a scaling pathway that purpose-built quantum fabs cannot match.
Intel’s modular architecture approach facilitates hybrid configurations combining multiple quantum cores, addressing scaling through parallelization. The Tunnel Falls roadmap targets thousands of qubits within several years, though the current qubit count remains early-stage relative to competing modalities.
Microsoft
Microsoft’s quantum approach differs fundamentally from competitors. Rather than superconducting or trapped-ion qubits, the company is developing topological qubits based on Majorana zero modes – exotic quantum states that in theory offer inherent error protection. The approach has attracted sustained scrutiny from physicists, and Microsoft’s 2025 Majorana 1 announcement prompted significant debate within the community over the evidence supporting those states.
In June 2026, Microsoft reported advances in its Majorana 2 processor, replacing aluminum with lead in its superconducting material stack. The change more than doubled the topological gap protecting quantum states and improved parity lifetimes from milliseconds to over 20 seconds – a more than 1,000-fold improvement. Microsoft cited these results in accelerating its roadmap to a scalable quantum computer by 2029, though the current device remains a prototype and significant scaling work lies ahead.
Rigetti Computing
Rigetti’s Ankaa processor uses superconducting qubits with emphasis on hybrid classical-quantum algorithm execution. Rigetti’s approach integrates tightly with its quantum-classical hybrid programming model, designed for parametrized quantum circuits and variational algorithms. Ankaa is accessible via Amazon Braket, and Rigetti’s focus on near-term hybrid approaches positions it toward practical applications ahead of fully error-corrected systems.
D-Wave
D-Wave’s Advantage2 processor is a quantum annealer with 4,400+ qubits, built for a different purpose than gate-based systems. Quantum annealing is designed for optimization and sampling problems – finding low-energy configurations in complex systems – rather than universal quantum computation. It does not require quantum error correction for near-term applications, which gives D-Wave a different commercial profile from companies building fault-tolerant universal systems. In 2025, D-Wave demonstrated results on quantum simulation that it described as surpassing classical methods for specific problem classes.
IQM
IQM develops superconducting quantum processors for cloud and on-premises deployment, with a strong focus on European national quantum initiatives. The company’s current production system is the IQM Garnet QPU, a 20-qubit processor with a median two-qubit gate fidelity of 99.5%, designed to scale to 150 qubits. In 2024, IQM demonstrated 99.9% two-qubit gate fidelity on a planar transmon qubit – a milestone the company cited as evidence its fabrication technology is ready to support the next generation of high-performance processors.
In November 2025, IQM launched the Halocene product line, an open and modular on-premises system targeting quantum error correction research. The first Halocene system will feature 150 qubits with support for NISQ algorithms and up to five logical qubits, targeting commercial availability by end of 2026, with future releases planned beyond 1,000 qubits. IQM has also introduced the Star topology – a superconducting qubit-resonator architecture that implements all-to-all connectivity through a central resonator, departing from the conventional lattice layouts used by most competing superconducting processors.
Quantinuum
Quantinuum’s H-Series trapped-ion processors deliver gate fidelities that superconducting systems have not matched at comparable qubit counts – 99.9%+ on single-qubit gates and 99.5%+ on two-qubit gates. The H2 system, featuring 56 all-to-all connected ytterbium qubits, achieved a quantum volume of 2²⁵, or 33,554,432 in september 2025, a milestone reflecting both gate fidelity and qubit connectivity.
The next-generation Helios system introduces 98 barium-ion qubits using quantum charge-coupled device (QCCD) architecture – a design that improves ion controllability by physically transporting ions within the trap structure rather than relying on optical addressing alone. Quantinuum’s software stack, including the TKET compiler, is available through Microsoft Azure Quantum.
Atom Computing
Atom Computing’s neutral atom platform scales to 1,200+ physical qubits in its AC1000 system, using Rydberg-blockade interactions to implement quantum gates. The neutral atom approach achieves all-to-all qubit connectivity and uniform qubit properties through optical tweezer arrays, with mid-circuit measurement and real-time conditional branching available in the current generation.
In January 2025, Atom Computing and Microsoft announced plans to deliver Magne, an error-corrected system targeting approximately 50 logical qubits. In July 2025, Denmark’s QuNorth initiative announced it would acquire a Magne system, backed by €80 million from EIFO and the Novo Nordisk Foundation, with the system expected to become operational by the end of 2026. The next-generation system projects 10,000+ physical qubits, which Atom estimates would yield 100+ logical qubits – placing neutral atoms among the more credible near-term paths to fault-tolerant operation.
Pasqal
Pasqal develops neutral-atom quantum processors using Rydberg-based technology, with systems deployed at HPC centers across Europe and the Middle East. Its current Orion QPU controls 200 programmable qubits, with a system operational at Aramco’s data center in Dhahran since November 2025 – the first quantum computer deployed in Saudi Arabia and the region’s first commercial QCaaS platform.
In 2026, Pasqal published results showing quantum advantage in materials simulation – a one-to-one quantum simulation of the frustrated magnet TmMgGaO4 using 256 qubits, producing results that matched experimental measurements and accessing non-equilibrium dynamics at timescales where entanglement growth is considered to place the problem beyond classical reach. In May 2026, Pasqal demonstrated for the first time that logical qubits outperform physical qubits on solving differential equations on real hardware. The company’s published roadmap targets 200+ logical qubits by 2029, and Pasqal has announced plans to go public via a SPAC merger with Bleichroeder Acquisition Corp. II.
Infleqtion
Infleqtion develops neutral-atom quantum computing and quantum sensing systems under the Sqale platform, spun out of ColdQuanta. In December 2025, the company delivered the UK’s only operational 100-physical-qubit quantum computing system at the National Quantum Computing Centre (NQCC) – a milestone the NQCC identified as a critical objective for the UK’s quantum strategy.
Infleqtion has achieved 12 logical qubits with error detection and loss correction ahead of schedule, and its 99.73% two-qubit gate fidelity on the Sqale system is among the higher benchmarks in neutral-atom hardware. The company’s roadmap targets 30 logical qubits in 2026 and 1,000 logical qubits by 2030. Infleqtion completed its SPAC merger with Churchill Capital Corp X in early 2026, raising approximately $550 million and listing on the NYSE under the ticker INFQ – becoming the first publicly traded neutral-atom quantum technology company.
QuEra Computing
QuEra’s Aquila is a 256-qubit neutral atom quantum processor built on arrays of rubidium atoms controlled and trapped using lasers, available via Amazon Braket – the first commercially accessible neutral-atom system on a public cloud. Aquila uses QuEra’s FPQA (field-programmable qubit array) technology, which allows flexible reconfiguration of qubit positioning for each computation, and supports analog quantum processing for optimization, simulation, and machine learning applications. In 2025 and early 2026, QuEra’s academic partners at Harvard and MIT published multiple Nature papers demonstrating fault-tolerant architectures with up to 96 logical qubits and the first logical-level magic state distillation on neutral atom hardware.
QuEra’s published roadmap targets 30 logical qubits from 3,000+ physical qubits in 2025, scaling to 100 logical qubits from 10,000+ physical qubits in 2026. The company has been selected by DARPA for its quantum computing programs and is accessible through the NERSC quantum computing access program at Lawrence Berkeley National Lab.
Xanadu
Xanadu’s Borealis is a 216-qubit photonic processor that demonstrated quantum advantage on Gaussian boson sampling in 2022.
In January 2025, Xanadu unveiled Aurora, a 12-qubit universal photonic quantum computer composed of four modular interconnected server racks – the first photonic system to demonstrate real-time error correction decoding. Aurora is accessible via cloud through Xanadu’s PennyLane software framework. Xanadu announced in November 2025 that it will go public via SPAC merger at a pro forma market capitalization of approximately $3.6 billion, which would make it the first publicly traded pure-play photonic quantum company. Room-temperature operation positions photonic systems for quantum networking and distributed quantum computing architectures.
IonQ
IonQ operates multiple trapped-ion platforms across different ion species and control technologies. The commercial Forte Enterprise system uses ytterbium ions with acousto-optic deflector (AOD) laser control, reaching 36 algorithmic qubits in 2024. The Tempo system, built on barium ions, reached #AQ 64 three months ahead of schedule in 2025 – delivering what IonQ describes as a computational space 36 quadrillion times larger than leading commercial superconducting systems.
In October 2025, IonQ announced 99.99% two-qubit gate fidelity using Electronic Qubit Control (EQC) technology, acquired through the September 2025 completion of the Oxford Ionics acquisition. EQC replaces laser-based qubit control with precision electronics manufactured on standard semiconductor chips, enabling qubit control mechanisms to be integrated directly onto microfabricated chips. IonQ became the first quantum company to cross the four-nines benchmark – achieved on R&D prototypes that will form the basis for 256-qubit systems planned for 2026. IonQ systems are accessible via Amazon Braket, Microsoft Azure, and Google Cloud.
In January 2026, IonQ also announced an agreement to acquire SkyWater Technology for approximately $1.8 billion in cash and stock. SkyWater is the largest exclusively US-based pure-play semiconductor foundry and previously fabricated superconducting qubits for D-Wave’s Advantage2 system. The acquisition would create a vertically integrated quantum platform combining IonQ’s trapped-ion hardware with domestic fabrication capabilities. Stockholders approved the merger in May 2026, with the transaction pending regulatory approval.
PsiQuantum
PsiQuantum’s Omega photonic chipset, announced in February 2025, is designed as a manufacturable platform toward million-qubit-scale quantum computers. The chipset is fabricated on GlobalFoundries silicon photonics wafers in New York, integrating high-performance single-photon sources, superconducting single-photon detectors, and barium titanate optical switches. Reported performance metrics include 99.98% single-qubit state preparation and measurement fidelity, 99.5% two-photon quantum interference visibility, and 99.72% chip-to-chip quantum interconnect fidelity.
PsiQuantum’s photonic architecture eliminates the need for dilution refrigerators – replacing them with datacenter-style rack cooling – which addresses one of the primary infrastructure constraints in scaling superconducting systems. The company raised $1 billion in a Series E round in September 2025 and broke ground on a facility in Illinois targeting commercial fault-tolerant systems. Total funding now exceeds $2.3 billion.
How Quantum Chips Are Manufactured
Fabrication approaches vary considerably across modalities, and those differences have direct implications for how quickly production can scale.
Superconducting qubits are fabricated in semiconductor foundries using modified CMOS-compatible processes – depositing aluminum films, patterning resonators and control lines through photolithography, and etching precise geometries. IBM, Google, and Intel work with partners including GlobalFoundries for chip production. The constraint is testing: each chip must be validated in dilution refrigerators before deployment, which creates bottlenecks that volume production cannot easily resolve.
Trapped-ion systems use custom-engineered electrode arrays fabricated through micromachining, with dimensional tolerances and yields significantly lower than semiconductor fabs. The advantage is operational – trapped-ion systems run in room-temperature vacuum chambers, eliminating dilution refrigeration costs. IonQ’s acquisition of Oxford Ionics in September 2025 reflected strategic interest in advancing trap fabrication – Oxford Ionics had developed Electronic Qubit Control (EQC), a technology that uses standard semiconductor chips to control ions via electronic signals rather than lasers, offering a pathway to manufacture ion traps at semiconductor scale.
Photonic quantum chips are the most manufacturing-compatible approach, using mature silicon photonics fabs. PsiQuantum’s fabrication partnership with GlobalFoundries and Xanadu’s use of standard photonic integrated circuit manufacturing both demonstrate this compatibility. The challenges are different from superconducting systems – single-photon sources and detectors require specialized components, and quantum memory remains an active research problem.
Neutral atom systems do not involve traditional chip fabrication. Atom Computing and QuEra build precision laser systems and optical tweezer arrays that trap and manipulate atoms. Manufacturing expertise centers on precision optics and laser engineering, allowing qubit counts to scale without chip-level miniaturization.
Silicon spin qubits, Intel’s approach with Tunnel Falls, use the same silicon fabrication infrastructure as classical chips. Quantum information is encoded in electron spin states in silicon quantum dots. The manufacturing pathway to scale is potentially the most accessible, though the technology is earlier-stage than competing modalities.
For a deeper look at how these modalities work, TQI’s guide to understanding the quantum computing hardware landscape covers each approach in depth.
What No Clear Winner Means for the Field
The absence of a dominant modality in 2026 reflects genuine uncertainty rather than delay. Superconducting qubits scale to higher qubit counts but require expensive cryogenic infrastructure. Trapped ions offer superior gate fidelities with slower scaling. Neutral atoms achieve large qubit counts with flexible connectivity. Photonics operate at room temperature but face photon loss and memory challenges. Silicon spin qubits offer manufacturing compatibility at an earlier technical stage.
A central question is whether one model will eventually dominate. The presence of several viable hardware strategies introduces complexity for investors and enterprise adopters. Classical computing went through a similar period of experimentation before consolidating around a small number of architectures – and that consolidation is what made it possible to build a global industry around the technology.
According to Techtarget, Carl Dukatz, global lead of the quantum program at Accenture, suggested history may point in that direction. “If we look at history, it tells us that there will generally be one that’s selected as the way, the preferred device, simply for the economies of scale,” Dukatz said.
Not everyone agrees on when – or whether – that consolidation arrives. Michael Biercuk, founder and CEO of Q-CTRL, whose firm develops quantum control infrastructure that works across platforms, offers a measured view: “Each modality has its own strengths and weaknesses. We don’t have a favorite.”
Government programs through the US National Quantum Initiative, EU Quantum Flagship, and national quantum programs elsewhere continue funding hardware development across multiple modalities – supporting diversity in the field rather than pushing convergence around a single approach.
The more likely near-term outcome is a heterogeneous landscape, meaning, superconducting systems in dedicated quantum data centers, trapped ions for error-corrected applications requiring high fidelity, photonics for distributed quantum networks, neutral atoms for simulation and optimization. Whether one of these eventually becomes the x86 of quantum computing, nobody in the field is willing to say with confidence. Which is either a sign that quantum computing is still early – or that it is more interesting than classical computing ever was.
Frequently Asked Questions
Why do different companies use different qubit types?
Different qubit modalities represent fundamental tradeoffs in quantum computing. Superconducting qubits offer maturity, shorter gate times, and thousands of devices worldwide; trapped ions deliver exceptional gate fidelities and long coherence times but require complex equipment; neutral atoms scale to thousands of qubits easily but face coherence challenges; photonic qubits operate at room temperature but photon loss limits scaling. No perfect solution exists – each modality excels in different metrics. Companies choose based on team expertise, manufacturing capabilities, target applications, and investment theses. This diversity drives competition and innovation.
When will quantum computers become practical?
Practicality depends on application. D-Wave’s annealing systems solve real optimization problems today. IBM, Google, and other gate-based systems show near-term utility in specific domains including quantum simulation, certain optimization problems, and materials science. However, quantum advantage – where quantum computers outperform classical computers on real-world problems – requires larger qubit counts and lower error rates. Based on published hardware roadmaps and industry planning assumptions, meaningful quantum advantage for specialized applications is generally projected within 3-7 years, with mainstream utility estimated at 7-15 years. The timeline depends on qubit scaling, error rates, and algorithm development progressing simultaneously.
Can quantum chips be manufactured in ordinary semiconductor fabs?
Partially. Superconducting qubits can be manufactured using modified semiconductor processes, though dilution refrigeration infrastructure is specialized. Trapped-ion systems require custom-engineered electrode arrays fabricated through micromachining, with dimensional tolerances and yields significantly lower than semiconductor fabs – IonQ’s acquisitions of Oxford Ionics and SkyWater Technology reflect the manufacturing challenge involved. Neutral atoms use optical tweezers requiring specialized photonics and precision laser systems rather than traditional chip fabrication. Photonic qubits leverage semiconductor manufacturing for integrated photonics components. Intel’s Tunnel Falls represents the closest approach to standard semiconductor fab compatibility. Scaling to volume production across any modality will require significant manufacturing innovation.
What role will quantum chips play as classical computing improves?
Quantum computing does not replace classical computing but augments it. Quantum chips are expected to serve specialized roles: optimization problems with exponential solution spaces, quantum simulation for materials and molecules, machine learning with quantum speedups, and cryptanalysis. Classical computers handle most computational tasks more efficiently. The future involves hybrid classical-quantum systems where quantum processors tackle specific subproblems while classical systems orchestrate overall computation. This symbiotic relationship means quantum and classical computing coexist, with quantum chips becoming specialized infrastructure comparable to GPUs or tensor processors today.
For readers looking to go further into the quantum computing landscape, TQI has recent coverage across several related areas – quantum computing jobs and salaries in 2026 covers what roles are in demand and what they pay across the hardware and software stack; quantum computing use cases across industries maps where each modality is most likely to deliver practical value first; and Chinese quantum computing companies covers the parallel hardware ecosystem developing outside the Western companies profiled in this article.
