Episode 178 · Deep tech · 54 min

Start with the use case, not the physics

Prof. Arindam Ghosh of IISc builds light detectors from atom-thin materials. He would happily see RSA encryption broken if it meant a working quantum computer, and he hopes quantum sensors will one day put brain imaging in a helmet.

AG
Arindam Ghosh
Professor of Physics, IISc Bengaluru · with Shreyas Mishra
Start with the use case, not the physics: episode thumbnail
54:14
The brief

The argument in sixty seconds

Quantum technology is engineering on physics nobody fully understands, Ghosh says, and that is no obstacle. He maps it as three verticals — computing, sensing and communication — on a horizontal of materials and devices. A useful quantum computer was once thought to need a billion qubits; by his count it is now about 10,000. What remains is engineering, but engineering that still needs basic research. Sensing gets less attention, though almost anything measured today could be measured more finely with a quantum sensor, he says. His advice to founders: pick the use case first, then learn only the quantum it needs.

Disclosure: Prof. Ghosh is a co-founder of Quan2D Technologies, an IISc-incubated startup working on 2D materials and quantum-sensing devices. He does not mention it on air, though he discusses quantum sensing and quantum startups at length.

The field, drawn

Ten charts, fifty-four minutes

Figures are as said on air; field strengths and milestone years are typical published values. Sources: National Quantum Mission (PIB), post-quantum standards (NIST), prize citation (Infosys Science Foundation). Click ▶ to hear it.

Kolkata, IISc, Cambridge and back

his path, by year · and the device behind his Infosys Prize
His path
Kolkata · BScIISc · straight to a PhDCambridge · Cavendish LabIISc · facultyBhatnagar Prize 2012Infosys Prize 20201990200020102020
The work behind the prize
light in: a few photons2D semiconductor absorbs lightgraphene carries the chargecurrent out~10¹⁰× the sensitivity of silicondetectors, per the prize jury
In 1991 IISc physics began admitting BSc graduates directly to a PhD; he was one of six picked nationally, he recalls. At Cambridge he tried to crystallize electrons. He came back as shared clean rooms reached Indian labs.▶ 2:28▶ 3:25▶ 5:09▶ 11:44

Three verticals on one horizontal

how he maps quantum technology · hubs of India’s National Quantum Mission (NQM)
ComputingNQM hubIIScBengaluruSensingNQM hubIIT BombayCommunicationNQM hubIIT Madras+ C-DOTQuantum materials and devicesunder all three · NQM hub IIT Delhi
Materials and devices rarely come up when people discuss quantum technology, he says, though they are “almost implicit” in all three. The ₹6,003.65 crore NQM, approved in 2023, is organized the same way.▶ 34:26

From a billion qubits to 10,000

estimated qubits for a useful quantum computer, log scale · his figures; he doesn’t say physical or logical
early 2000s1,000,000,000today10,00011K1M1B100,000× fewerplatforms named on airSuperconductingPhotonic · room tempTrapped ionsNeutral atomsSemiconductorMajorana · proposed
Ten thousand is “well within our reach”, he says. Unlike silicon in classical computing, almost every quantum platform may become useful on its own track; which will win, “I don’t think anybody knows”.▶ 18:04▶ 18:26▶ 19:05

He’d trade RSA for a quantum computer

NIST’s timetable for retiring RSA (2030 and 2035 steps are a draft plan) · he puts an RSA-breaker at ~1,000–4,000 logical qubits
the defence is already scheduled202020242028203220363 PQC standardsAug 2024RSA-2048deprecatedall RSAdisallowedtodayRSA-breaker: date unknownit would also buy: materials, optimization,climate modelling, “so many things”
Breaking encryption was once pitched as a reason to fund quantum computing. He calls it “the least thing that you have to worry about”: he would happily lose RSA for a machine that can do so much more.▶ 21:59▶ 22:52▶ 23:34

Listening in leaves a trace

quantum key distribution, as he describes it
ground or satellite linkSendersingle photonsReceiversees errorsEavesdroppermeasures a photonmeasuring disturbs the photons:the intrusion shows up
“Fundamentally not hackable”, he says, so countries and companies want it for strategic links. A link needs a single-photon source and detector; the photons themselves carry the key.▶ 31:11▶ 31:43▶ 33:05

Sensing fields far fainter than Earth’s

magnetic field strength in tesla, log scale · typical values; bars start at the faintest field each sensor reads
fTpTnTµTmTTbrain ~100 fTheart ~50 pTEarth ~50 µTMRI magnet 3 Tdiamond sensor: from ~1 pTphone: from ~0.1 µTdashed: his ~10 fT
Where he would use them
Healthcare
Brain imaging in a helmet
wearable scanners, no big magnet
Navigation
GPS-free navigation
maps Earth’s magnetic field
Geoscience
Earthquake and tsunami warning
atomic sensors track the crust; also oil and gas
Radar
LiDAR through noise
entanglement lifts buried signals
Most sensors are quantum at some level, silicon ones included; what sets quantum sensors apart, he says, is that they exploit coherence.▶ 26:42▶ 28:27▶ 30:15

Any discipline can found a quantum startup

four founders he sketched on air
The use case
Where quantum comes in
Software engineer or mathematician
▶ 36:39
Traffic optimization
A quantum optimization algorithm
Geologist
▶ 36:56
Climate models for weather services
Quantum code run over decades of data
Atmospheric modeller
▶ 38:04
The ground-to-satellite laser link
Serves QKD, yet “has got nothing to do with quantum”
Biologist
▶ 38:56
Gene editing
Quantum computing alongside CRISPR
He pictures this “replicated in thousands of different applications”. The first step is to stop the word “quantum” intimidating you, he says; short courses for newcomers are multiplying.▶ 35:39▶ 38:38▶ 40:06

Students: foundations first, then add AI

how he says to learn quantum, by career stage
students: foundations, while there is timelinearalgebrauncertaintyprinciplequantumgatesexecutives: higher up; hire the experteveryone: pair quantum with AIquantumsensorAI modeldenoisesproduct:navigation
Quantum will sit inside AI, cybersecurity, healthcare, grid management and robotics, he says; knowing where quantum and AI complement each other will raise a young professional’s value.▶ 41:58▶ 42:48▶ 45:39

Graphene is young for a material

years from the first device to everyday use
the transistor · 1947 → IBM PC 198134 years to everyday computinggraphene · 2004 → today22 yearssingle-layer MoS₂ transistor 2011010203040years after the first device
Graphene was expected to replace silicon and hasn’t. Instead it opened the way to atom-thin semiconductors such as molybdenum disulfide, which he says do silicon’s job on less energy, and to a push for post-silicon electronics.▶ 47:11▶ 48:15▶ 49:02

Noise is the signal

subtract the average; the fluctuations carry the physics · illustrative trace, size on a log scale
measured signalaverage: what most keepsubtract, then magnifyfluctuations: what he studies
how small1 nm1 µm1 mmquantum dota few nmhuman hair~100 µm~25,000× smallerand small in all three dimensions
Studying noise began as his PhD problem. In materials as small as quantum dots, the fluctuations open “a whole different area of science”: how the material behaves and what new uses it might have.▶ 51:07▶ 52:14▶ 52:53
Worth keeping

Lines that stay

Not knowing the precise mechanism didn’t stop us from engineering it.

Arindam Ghosh, on quantum entanglement ▶ 16:26

I really felt that evenings in a cricket ground or a football ground was actually helping my research than taking time away from it.

Arindam Ghosh, on cricket and research ▶ 13:39
Chapters

Where the conversation travels

Every block is a chapter, coloured by what it's about. Click any of it to jump straight to that minute on YouTube.

Glossary

The jargon, unpacked

Logical qubit
An error-corrected qubit assembled from many physical qubits, which on their own are too error-prone for long calculations.
Post-quantum cryptography
Encryption for ordinary computers, built on maths problems quantum computers are not expected to crack. It is meant to replace schemes such as RSA.
Quantum coherence
A quantum system’s ability to hold a well-defined superposition. It is easily disturbed, which is what makes it useful for sensing.
No-cloning theorem
An unknown quantum state cannot be copied exactly, so an eavesdropper cannot read a message and pass on an undisturbed duplicate.
2D material
A material one or a few atoms thick, such as graphene or molybdenum disulfide.
Femtotesla
A quadrillionth of a tesla. Earth’s magnetic field is about 50 billion femtotesla.
Connections

If this resonated, go here next

Full transcript

The whole conversation, searchable

202 segments
AGProf. Arindam GhoshSMShreyas Mishra, UpStreamVoice

Machine transcript of the interview audio, corrected for mishearings and lightly edited for readability; [unclear] marks words that could not be recovered. Click a timestamp to open that moment on YouTube.

Checked against the record: the award is the Infosys Prize 2020 in Physical Sciences; the introduction calls it the “Infosys Laureate Prize”. MRI does not measure the brain’s own magnetic field; it reads radio signals from hydrogen nuclei inside a strong magnet. The brain’s field, about 100 fT, is measured by magnetoencephalography (MEG), and the wearable helmets now in trials are MEG scanners built on optically pumped magnetometers, a quantum sensor (said: “the MRI machines measure magnetic field of the brain”, “a handheld MRI”). In quantum key distribution the photons set up the shared secret; the two sides use an ordinary authenticated channel only to compare settings, so no public key is exchanged (said: “a public key that is exchanged”). Real QKD systems have been attacked through hardware flaws. NIST asks organizations to move to post-quantum standards as soon as possible, because data encrypted today can be stored now and decrypted later; its draft transition plan deprecates RSA after 2030 and disallows it after 2035. The first transistor (germanium, Bell Labs) dates from 1947 and the first microprocessor, Intel’s 4004, from 1971, 24 years later; personal computers spread in the 1980s (said: “40 years from 1950s”).