IBM Invests $100 Billion in Quantum Computing and Announces Anderon Fab in the USA
With support from the US Department of Commerce, IBM will build the country's first specialized quantum chip factory, Anderon, backed by $2 billion in investments. The company also announced plans to pour more than $100 billion into the quantum sector over five years.
Industrial Deception at IBM: Why $100 Billion Is Not a Bet on Quantum but the Last Car on a Departing Train
You've seen the headlines. This week IBM announced plans to invest more than $100 billion in quantum computing over five years, while the US Department of Commerce awarded it $1 billion from the CHIPS Act to create the Anderon factory—the nation's first dedicated 300-mm line for quantum chips. It sounds like the moment the "Big Blue" finally seizes the initiative from Microsoft and Google. By 2029, IBM Quantum Starling, 200 logical qubits, 100 million operations—the numbers take your breath away.
But as someone who has spent the last decade advising private-equity funds in deep tech, I see a very different picture. What Microsoft is doing with Majorana 2 is a scientific gamble. What IBM is doing is an industrial trap disguised as national salvation.
Forget the headline investment figures. Let's look at the raw facts that never make it into press releases. This story is not about a breakthrough in physics. It is about a corporation that slept through the AI revolution now trying to buy a seat at the quantum table with public money and an outdated semiconductor mindset. And that strategy has one fatal flaw everyone is missing.
[The Core]: What Is Really Happening
IBM is desperately trying to turn quantum computing into an extension of its semiconductor business, because that is the only thing it knows how to do at industrial scale. It has not invented a new qubit type. It has not found a way around the laws of quantum physics. Instead, it has taken the standard superconducting qubit (transmon) used by everyone from Google to Rigetti and decided to stamp them out on a 300-mm fab exactly the way it stamps laptop chips.
Here lies the fundamental misunderstanding of what a quantum computer actually is. A classical chip is a deterministic system. Print the transistor and it works. A quantum chip is essentially an analog device. Every qubit on the wafer has slightly different frequencies, different relaxation times, and different noise levels. At industrial scale, with thousands of qubits per wafer, parameter variation becomes catastrophic. You cannot simply "print" a million identical qubits because quantum mechanics does not allow two identical macroscopic quantum states.
The inside story no one discusses: Anderon will produce physical qubits, but the real bottleneck in quantum computing today is not fabrication—it is error correction. To obtain one reliable logical qubit from superconducting physical qubits, you need between 1,000 and 10,000 physical qubits. IBM promises 200 logical qubits in Starling. That means 200,000 to 2 million physical qubits inside a single cryostat. The problem is not that they cannot print 2 million transmons. The problem is that no known cooling system can remove the heat generated by a million qubits and their control electronics at 15 millikelvin.
IBM is spending $100 billion not on solving the physics of scaling. It is spending the money on automating the production of something no one has yet learned how to wire into a working system. It is like building the world's most advanced brick factory while trying to erect a skyscraper and no one has invented steel framing yet.
Timeline and Context
We need to rewind five years to understand the desperation behind this announcement. In 2020–2021 IBM was the undisputed leader in the qubit race. It was first to show a 65-qubit processor, then the 127-qubit Eagle. Everyone talked about IBM's "quantum supremacy." And what happened? Nothing. Quantum supremacy never arrived. Google overtook it on qubit quality with Willow, and the market shifted from a numbers race to a quality-and-error-correction race.
Now look at the dates. On 5 May 2026 Microsoft loudly unveiled Majorana 2. It needed the hype and it got it. The market perked up. Literally a month later, in early June, IBM rolled out its response—$100 billion and the Anderon factory. This is not an independent event. It is a panicked reaction to lost attention.
But there is a detail that makes this "panic" especially cynical. Look at the investment structure. Of the $100 billion, only $1 billion is fresh government money for the fab and another $1 billion is IBM's own contribution. Where is the remaining $98 billion? It "will be allocated to R&D, capital expenditures, and the ecosystem." This is the classic corporate trick: announce "investments" you were already planning to make. A large portion of that money is simply operating expenses—salaries, rent, and ongoing research—that IBM would incur anyway, quantum or no quantum.
The key point no one noticed: Anderon will not build a new facility. As IBM itself quietly admits (and this only appears in technical blogs, not glossy press releases), Anderon is simply a tenant in clean rooms at the Albany NanoTech Complex where IBM has already been working for years. It has merely carved out its own capacity into a separate subsidiary to capture federal funds. There is almost no real expansion of manufacturing capacity. This is an accounting maneuver, not an industrial breakthrough.
Winners and Losers
Let's map the interests. Who actually benefits from this "quantum march," and who stands to lose everything.
The biggest winner is New York Governor Kathy Hochul. Seriously. The $2 billion from the CHIPS Act is spread across nine companies, but the lion's share ($1 billion) went to IBM in Albany. That creates jobs and tax revenue for the state. It is a political victory at the local level dressed up as a technological breakthrough. The federal government paid for "national security," and the money landed in a specific congressional district. It is the perfect machine for converting defense budgets into regional subsidies.
The next winner is GlobalFoundries. It received $375 million to create a "multimodal" quantum fab. Note that GlobalFoundries long ago lost the process-technology race to TSMC and Samsung in classical chips. The quantum fab is its ticket back into the game. IBM is effectively admitting it cannot scale its own production and is bringing in a competitor. That is a sign of weakness, not strength.
The biggest losers are small quantum startups. Look at the grant list: D-Wave, Rigetti, PsiQuantum, Quantinuum. They received crumbs compared with the giants. The federal government just told every venture-capital investor: "The big players will get the main contracts—don't bet on startups." That kills innovation. Money will flow to IBM and GlobalFoundries, not to IonQ or QuEra, which may have superior technology but lack lobbyists in Washington.
China loses, but not the way the White House thinks. This package will force Beijing to double its own subsidies. The US and China are entering a subsidy-escalation spiral in quantum. Yet China holds a critical advantage: it controls supply chains for rare-earth elements and specialty materials, including lithium niobate for photonics and helium-3 for cryogenics. Five years from now, US fabs built on subsidies may discover they have nothing to make chips from because China has cut off raw-material supplies. A gilded cage for American industry—that is what the CHIPS Act is building.
What the Media Is Not Saying
Journalists write about the "quantum moment" while missing three glaring facts that turn IBM's rosy forecasts into an investor nightmare.
Inside Story #1: The interconnect problem no one is solving.
IBM promises 10,000 physical qubits for 200 logical qubits by 2029. Suppose the Anderon fab can print them. But how do you connect 10,000 qubits to one another inside a room-sized cryostat? Today's standard quantum processors have dozens or hundreds of wires running from each qubit to control electronics. For 10,000 qubits you would need 10,000 physical coaxial cables entering the cryostat. Every cable adds heat. Every cable takes up space. Physics says it is impossible—you will hit a thermal-load wall and melt the system before it ever runs. IBM is silent because it has no solution. It is hoping for a "quantum internet" or photonic interconnects that do not yet exist commercially.
Inside Story #2: The "free" CHIPS Act money is a Trojan horse.
Few people read the fine print in the Commerce Department agreements. It states that the federal government will receive passive, non-controlling minority equity stakes in each recipient company proportional to the size of the subsidy. In other words, the US government becomes a shareholder in Anderon. This is not a grant. It is the nationalization of future profits under the guise of support. IBM is handing the government a piece of its creation in exchange for cash today. And when (and if) the quantum industry takes off, taxpayers will get a return through their stakes—but at what price? The price that PsiQuantum and others that took the money must now answer not only to shareholders but also to officials at the Department of Defense. That destroys flexibility and speed of decision-making.
Inside Story #3: AI is the quantum killer.
Why does IBM need this announcement so desperately right now? Because its classical business is dying. Nvidia is worth $3 trillion; IBM is around $150 billion. It slept through the AI-accelerator revolution. Now it is trying to sell investors the story that "quantum will be the next AI." But the entire quantum-computing race could become irrelevant if generative AI solves molecular-structure modeling with classical methods. Neural networks already predict protein folding with accuracy once promised only by quantum computers. By 2029, when IBM finally plans to show Starling, classical AI will have advanced so far that any "quantum advantage" may be purely academic rather than commercial.
Outlook: The Next 30 Days and 90 Days
Turn off emotion. Turn on cold calculation. Here is what will actually happen to the quantum industry after this news.
Next 30 days: A wave of skepticism from physicists.
Within a month, at least three serious papers will appear on arXiv showing that a 300-mm fab for superconducting qubits does not solve the scaling problem because of Coulomb interactions between closely spaced qubits on the wafer. IBM's stock will not drop—investors do not read arXiv. But venture funds will stop writing checks to startups working on superconducting qubits and shift to photonic and neutral-atom platforms (PsiQuantum, QuEra) that have no heat-dissipation or interconnect problems. It will be a quiet but fatal reallocation of capital.
Next 90 days: Legal complaints and fights with the Commerce Department.
Diraq, Atom Computing, and Rigetti (which also received funding, but less) will file complaints with the GAO (Government Accountability Office) over the uneven distribution of subsidies. They will demand a review, arguing that their technologies (silicon spins at Diraq or neutral atoms at Atom) have a better chance of scaling than IBM's outdated transmon qubits. Lengthy congressional hearings will begin. A temporary injunction could halt final funding tranches for Anderon.
In the longer term (two to three years) IBM will get its factory. But it will prove strategically meaningless, because the world will by then understand: the winners will not be those who can stamp out the most qubits, but those who figure out how to make them work together with minimal errors. That requires new physics, not a fab—and IBM does not have it. Its $100 billion is simply the most expensive monument in history to its own inability to change its paradigm of thinking. IBM engineers still think in terms of transistors and process nodes. The quantum world does not work that way. And no amount of money will change that.
— Editorial Team
No comments yet.