India's Rs 6,003.65 Crore Quantum Mission Has Four Hubs, a 1,000 km Key Network and a 64-Qubit Chip, Against a Goal of Up to 1,000 Qubits

The Union Cabinet approved India’s National Quantum Mission on 19 April 2023 at a total cost of Rs 6,003.65 crore for the years 2023-24 to 2030-31. A Lok Sabha reply of 4 February 2026 shows how much had moved by then: Rs 2,330.68 crore sanctioned to the mission’s four research hubs, of which Rs 699.50 crore, about 30 percent, had been released.

The Department of Science and Technology (DST) runs the mission. What follows sets the cabinet’s promises beside what government replies, companies and journals report as of 8 October 2026.

What the cabinet promised

The cabinet release sets a computing goal of “intermediate scale” machines with 50 to 1,000 physical qubits in eight years, on platforms such as superconducting circuits and photonics. A physical qubit is one real quantum device on a chip. DST’s later replies add the stages: 20 to 50 physical qubits in three years, 50 to 100 in five, and 50 to 1,000 in eight.

The other deliverables are specific. They include satellite-based secure communication between two ground stations 2,000 km apart inside India, and inter-city quantum key distribution (QKD, a way of sharing encryption keys using single particles of light) over 2,000 km using trusted nodes on existing optical fibre. A quantum network of two or three nodes with quantum memories is also on the list. So are magnetometers sensitive to 1 femto-Tesla per root hertz in atomic systems (a femto-Tesla is a millionth of a billionth of a tesla) and atomic clocks with a fractional instability of 10 to the power of minus 19.

Four hubs, four project directors

DST opened a call for proposals in January 2024 and received 384, of which 17 were selected, according to a Press Information Bureau (PIB) release of 30 September 2024. That evening the Principal Scientific Adviser, A K Sood, announced the hubs with the minister, Jitendra Singh, joining by video. The minister’s own release of that evening also covers the launch of BharatGen, a government-funded language-model project. India’s wider push to build AI in its own languages is the subject of our piece on it.

The money columns come from the annexure to the 4 February 2026 reply. The directors are as each hub’s own website or kick-off report lists them. The hubs are not-for-profit Section 8 companies run by their host institutions: Foundation for QC Innovation, IITM CDOT Samgnya Technologies Foundation, Qmet Tech Foundation and QMD Foundation.

The two PIB releases of 30 September 2024 count differently. One says 14 technical groups with 17 project teams, the other says 17 technical groups. Later government replies use 14 and 17, involving 152 researchers at 43 institutions across 17 states and two union territories.

What has been built

The first computing stage asks for 20 to 50 physical qubits. The Rajya Sabha reply of 12 February 2026 credits QpiAI, a Bengaluru startup supported by the mission, with what it calls a 64-qubit “scalable, fault-tolerant” quantum processor. QpiAI’s earlier machine, Indus, has 25 superconducting qubits. In its April 2025 release the company gave a single-qubit gate fidelity of 99.7 percent and a two-qubit gate fidelity of 96 percent, with coherence times of 30 and 25 microseconds. We found no independent test of those figures.

The same reply lists a Quantum Diamond Microscope from the PQuest group at the IIT Bombay hub, which it calls India’s first indigenous one, and a 1 km entanglement-based free-space link built by DRDO and IIT Delhi. It also says fabrication and central facilities for quantum devices are being set up at IIT Bombay, IISc Bengaluru, IIT Kanpur and IIT Delhi. The country’s commercial chip plants are a separate programme, covered in our piece on the semiconductor fab race.

Superconducting circuits are the platform behind the 25 and 64-qubit chips. The 2025 physics Nobel prize, covered in our roundup of the science Nobels, went to Clarke, Devoret and Martinis for experiments on a superconducting circuit big enough to hold in the hand.

The 1,000 km network and its fine print

On 8 April 2026 PIB reported that QNu Labs, another supported startup, had demonstrated a 1,000 km quantum communication network with indigenous technology, against the mission’s 2,000 km target. DST Secretary Abhay Karandikar called it “a landmark advancement in secure quantum communication”, PIB said. The release does not say how many relay nodes the network uses. It adds that the mission now supports 17 startups, up from eight in the 12 February reply.

The cabinet target itself specifies trusted nodes, intermediate stations that pass the key along and so have to be trusted themselves. A Chinese team reported a similar design in a 2021 Nature paper: a fibre network of more than 700 QKD links covering over 2,000 km through a trusted relay, joined to two satellite-to-ground links for a total reach of 4,600 km. They wrote that quantum repeaters, the alternative to trusted relays, “cannot be deployed using current technology”.

Free-space tests have lengthened since the 1 km DRDO link. On the night of 27 to 28 September 2026, QNu Labs, BISAG-N and IIT Gandhinagar ran a 5.56 km link between BISAG-N and IIT Gandhinagar. The announcement, from the Ministry of Electronics and IT and carried by DD India, gives a key rate of 230 to 260 bits per second and an error rate below 5 percent. That is a ground link. We found no release reporting a satellite QKD link, which is the mission’s other 2,000 km target.

The Amaravati machine

Amaravati Quantum Valley in Andhra Pradesh is a state project. IBM said on 2 May 2025 that it would anchor the park with an IBM Quantum System Two carrying a 156-qubit Heron processor, which it called the largest quantum computer in India. The hardware would be IBM’s, run with Tata Consultancy Services, not an indigenous machine. IBM’s release gave no commissioning date, and we found no announcement of one by 8 October 2026.

How the money compares abroad

The United States signed its National Quantum Initiative Act in December 2018. On 4 November 2025 the Department of Energy announced $625 million to renew its five national quantum research centres for up to five years, with $125 million in fiscal 2025 dollars and the rest dependent on Congress. That covers five centres, not a whole national programme, so it does not compare directly with India’s total.

The European Commission says its Quantum Technologies Flagship, launched in 2018, has a budget of 1 billion euros. EuroHPC, the EU’s joint supercomputing initiative, is placing six quantum computers at sites in Czechia, Germany, Spain, France, Italy and Poland for 100 million euros, split between the EU and 17 participating countries. The first opened in Poznan in June 2025 and the second in Ostrava in September 2025. A Quantum Act is scheduled for adoption in 2026.

We found no official Chinese figure for national quantum spending to set beside these, and this piece uses no outside estimate. What China has published in a journal is the 2021 network described above.

What a qubit count leaves out

IBM announced Condor, a 1,121-qubit processor, in December 2023 and said its performance was comparable to the 433-qubit Osprey before it. By raw count, one chip had passed the mission’s 1,000-qubit ceiling less than eight months after the cabinet approved it.

A qubit count does not show how often the qubits err. Google’s Willow paper in Nature, published in December 2024, built a distance-7 error-corrected memory from 101 qubits and reported 0.143 percent error per correction cycle. A logical qubit is one error-protected qubit made from many physical ones, and the paper’s formula puts one at distance 7 at 97 physical qubits, so at that ratio 1,000 physical qubits would hold about ten logical ones. That division is our arithmetic, not a claim by the mission.

The deliverables published by the cabinet and DST state counts and no error rate for the computers. QpiAI’s own figure for two-qubit gates on Indus is 96 percent.