IBM announced on June 2 that it will invest more than $10 billion in quantum computing over the next five years, a commitment it tied directly to a specific promise: delivery in 2029 of the world’s first large-scale, fault-tolerant quantum computer, a machine it calls IBM Quantum Starling. The investment spans research and development, capital expenditure, manufacturing scaling, ecosystem partnerships and strategic acquisitions, a breadth that reflects how much of the technology’s future depends on things other than the machines themselves.
The Starling system is the centerpiece of the roadmap. IBM says it will be capable of running 100 million quantum operations on 200 logical qubits, roughly 20,000 times more operations than today’s quantum computers can execute. The step-change is the point: today’s machines are noisy, error-prone research instruments, while a fault-tolerant system corrects its own errors in real time and can run long computations that would otherwise be destroyed by decoherence. Fault tolerance is the industry’s hard problem, and IBM is putting its money behind a date.
The announcement is as much about manufacturing as it is about physics. IBM plans to establish Anderon, a standalone quantum wafer foundry in New York, supported by a Letter of Intent with the U.S. Department of Commerce and proposed CHIPS Act incentives. Quantum processors require fabrication techniques that differ from conventional chips, and IBM has concluded it cannot rely on third-party capacity for a technology that is central to its future. The foundry gives the company control over the supply chain that its roadmap depends on.
The scale of the existing business is larger than most people realize. IBM has deployed more than 90 quantum systems worldwide, the largest fleet in the industry, and runs the IBM Quantum Network with more than 340 organizations across financial services, healthcare, materials science, academia and government. Its Qiskit software platform is used by nearly 70 percent of quantum developers, according to the company. The $10 billion is not a bet on an unproven lab experiment; it is the expansion of an installed base that already generates revenue.
The roadmap beyond Starling shows where the investment is heading. IBM Quantum Blue Jay, the next machine in the sequence, is designed to run one billion quantum operations across 2,000 qubits. The company also described accelerated research on a quantum computing internet and networked quantum computing, the longer-term vision in which quantum processors are linked the way classical computers are linked today. The announcement framed the $10 billion as funding progress beyond the current roadmap, not merely toward it.
The commitment lands in a competitive field that is consolidating around deadlines. Google has pursued fault tolerance with its own hardware, Microsoft has charted a topological approach, and a cluster of startups, Quantinuum, PsiQuantum and others, have raised billions on similar promises. IBM’s differentiator has been its insistence on a full-stack approach, building everything from the chips to the software to the applications, and the new investment doubles down on that strategy rather than departing from it.
The economics of quantum have shifted in the company’s favor. For years, quantum computing was a research expense with no clear customer. IBM says it has signed more than $1 billion in quantum-related commitments, and its network includes banks testing portfolio optimization, pharmaceutical companies modeling molecules and materials firms simulating new alloys. The customers are early and the workloads are narrow, but they are real, and they give IBM something its competitors lack: evidence that the machines can be sold.
The 2029 date is the risk. Fault tolerance has a history of being five years away, and IBM itself has pushed targets back before. The company says its path has been de-risked by advances in error correction, including a quantum low-density parity check code that the company says cuts the overhead required for error correction by roughly 90 percent compared with other leading codes, and by the modular architecture it plans to scale from chip to chip. Whether that is engineering confidence or corporate optimism will be tested by the intermediate systems on the roadmap, including Kookaburra, its first modular processor designed to store and process encoded information, due this year, and the chip-linking architecture that follows it.
The investment also carries a strategic message. IBM is a company that missed the AI wave’s first act, watching NVIDIA and the hyperscalers capture the value it had once claimed for enterprise computing. Quantum is its answer to the question of what comes after: a technology where the barriers to entry are manufacturing, integration and trust, all of which favor an incumbent with a balance sheet. The $10 billion says IBM intends to be the leader of the post-AI era rather than a follower of it.
For now, the money is the message. Quantum computing has been a story of promises for three decades; IBM has now attached a five-year budget and a named machine with a delivery date to its version of the promise. Whether Starling arrives on schedule, in a working form, will determine whether this commitment looks like prudent positioning or the most expensive research grant in corporate history. The company’s customers, and its competitors, will be watching the calendar.
