Quantum Builders
Measuring, Controlling, and Connecting Quantum Systems
A Live Conversation with Michael Hatridge
A fireside chat with:
Michael Hatridge
This article is based on a conversation between Michael Hatridge and Daniel Rodán Legrain as part of the Quantum Builders series, sponsored by Qblox. Watch the full webinar for more on quantum measurement, parametric amplification, and modular superconducting architectures.
Hatridge does not describe his path as planned. "I cannot claim that I had a very simple or carefully curated way through the field," he said. His PhD work was low-field MRI, studying prostate cancer, breast cancer, and brain function. The qubits came later. "I didn't measure a single qubit until my postdoc," he said. "We didn't do a two qubit gate, like I personally didn't do a two qubit gate until I was a professor."
What carried across every stage was the instrument. "The through line I have is sort of that I was always an instrument builder," he said.
The first parametric amplifier came out of that habit rather than a plan. His group was building a dispersive magnetometer using circuit QED techniques, and it performed better than the physics allowed. "The first paramp I built was definitely on accident," he said. "Very quickly, we learned to stop calling it a magnetometer, but to use the same circuit to amplify qubit signals."
Both of Hatridge's advisors, John Clarke and Michel Devoret, shared the 2025 Nobel Prize in Physics. He recalls looking up an article in 2005 speculating that Clarke might win the Nobel Prize for the macroscopic quantum tunneling experiment, roughly 20 years before it happened.
What he took from both goes beyond technique. "They are also very concerned with not fooling yourself, with careful calibration, careful control experiments, saying what you know, saying what you don't know," he said. He sees that discipline running through the forty years of research that followed the original experiment.
The turning point was a 2011 Les Houches school where Alexander Korotkov presented a theory on the back action of weak quantum measurements. The prevailing intuition, inherited from atomic physics, was that the wavefunction collapses almost immediately, and you rebuild classical knowledge from the few photons you catch. Korotkov argued that if the measurement is weak enough that no observer could have extracted the full information, the quantum features survive.
Hatridge and Devoret decided to test it. Working with theorists Steve Girvin and Mazyar Mirrahimi, and with new parametric amplifiers that pushed quantum efficiency to twenty or thirty percent, they tracked what the qubit does under continuous weak measurement. "In fact, it does this smooth quantum walk across the Bloch sphere, and we could very nicely track it," he said.
The result reset how he views the whole subject. It taught him "how to view quantum measurement as a really interesting quantum game we can play and not where quantum mechanics goes to die."
That framing shows up directly in error correction, where the measurement is designed to stay blind to the thing you are protecting. "You never ask what state you are in," he said. "You ask, did you have an error? And if so, what type?"
Asked what makes an amplifier quantum-limited, Hatridge starts with the unglamorous answer. On a data sheet, the amplifiers his group builds would look mediocre next to commercial parts on bandwidth, saturation power, and power draw. The difference is that the amplifier itself is a quantum object, which means it transforms the light rather than simply converting it to a classical answer.
From there, the limit comes back to Heisenberg. "You can learn one quantity of a microwave field perfectly if you destroy the other," he said. A phase-sensitive amplifier makes that trade and has no quantum limit, but it only works when all your information sits in one quadrature. Refuse to choose, and you accept a minimum uncertainty set by vacuum noise itself. Destroying the other quadrature is not a matter of ignoring it. "You really have to squeeze it away, hide it away," he said.
Hatridge points out that groups like Google routinely cite readout as the worst-performing part of their system, and he has a structural explanation for why. Coherence improves when you seal a qubit off from its environment. Readout is the one operation that requires the opposite.
It also spans the entire chain. There is the qubit interacting with the readout apparatus, the movement of information across the cryostat, and the processing of that information in the quantum regime. "It's the whole system bolted together with all of its warts and defects that come together to make the quantum efficiency," he said. His group can build amplifiers at seventy to eighty percent quantum efficiency, and none of it survives a bad connection. "If you were to put a lossy cable in front of it, your readout won't work for beans. My amplifier cannot perform."
He frames the distinction plainly. Most figures of merit in a quantum computer are closed-system properties, measured when the device is isolated as well as possible. "Readout is the one that says how do you behave when you're interacting in a delicate and controlled way, but as strongly as you can with the right environment."
There is a timing problem stacked on top. Readout must complete tens of thousands of times faster than the qubit lifetime, pushing the system to its limits in the same way gates do.
The word parametric is well established in the field and, by Hatridge's account, misleading. Resonant driving is someone pushing you on a swing. Parametric driving is what a child does alone, standing and squatting to modulate their moment of inertia. "That is the essence of parametric," he said. "There's a parameter, something like the capacitance of something, the inductance of something, the moment of inertia of something that you are driving."
In his lab the parameter has become abstract. Rather than modulating a physical inductance, they treat the circuit's nonlinearity as a design tool and use microwave drives to switch interactions on and off. "Parametric driving is a way to turn those tools into lower order interactions that I can turn on and off with microwave driving," he said. The picture underneath stays the same. "The overall idea still is the kid on a swing."
Once his group started looking for these susceptibilities, they found there was very little they could not build with them. Gates became parametric. Couplers became parametric. It became something of a badge of honor in the lab.
The argument holds even for teams that want nothing to do with the approach. Avoiding parametric interactions means verifying that no drive frequency in your system accidentally lands on one. Hatridge points to the well-known readout problem in transmon qubits, where the cavity drive hits a parametric resonance of the transmon. "Even if you don't believe in them, they don't know that. And so you will turn them on by accident."
On scaling, Hatridge starts with a number. In Google's fifty-three qubit result, one qubit was dead. Transistors can be manufactured with a vanishingly small failure rate. Superconducting qubits are fabricated objects with individual character, and he is not confident that changes.
If flawed components are a given, the architecture has to contain the damage. "You need to truncate the size of your system and figure out a way to build it out of pieces," he said. He does not treat this as specific to his lab or even to superconducting circuits. Neutral atom systems hit their own constraints from optics and array size. "It's really easy to build an argument for modularity. It's much harder to imagine pretty much in any quantum system why a monolithic thing is not going to eat your lunch."
Where he departs from the ion trap approach is the boundary itself. Trapped ion modularity draws a hard line, with deterministic operations inside a module and measurement-based entanglement between them, because collecting photons from an atom efficiently is hard. Superconducting circuits have the opposite problem. "Our atoms are sort of gigantic," he said, and coupling one to a transmission line is so easy it barely counts as a design parameter. His group is testing whether the line needs to exist at all, routing photons between modules efficiently enough that an error correction code could plan around the weaker links.
The 3D circuits his lab works with are not built for volume manufacturing, which is exactly the point. "We really can test them in pieces, unhook them, pull out bad qubits," he said. Without a multi-billion-dollar fab, staged assembly is the advantage a university lab has.
Hatridge describes the lab's internal motto as building circuits big and strange enough to push the field. In practice, it is a sizing question. "If we have a new idea, what's the minimum size circuit we can put together that gives a fair test of that idea? If it's five qubits, we should do that. If it's like 500, probably not."
He is clear about which academic groups should be left alone. "We're not gonna scale the surface code," he said, noting that well-funded teams have that locked down. The interesting question is whether the surface code is the right thing to build, and making that argument credible means building a piece of the alternative.
The same logic shapes how he sees the split with industry. A quantum engineer's job, as he puts it, is to squeeze the exciting physics out of the qubits so they behave the same way a million times. The academic's job is to work on the ones that are completely out of control.
That extends to his own amplifier work. His group has stayed away from TWPAs, which use thousands of junctions. "Our amplifiers these days are sort of dead simple. They have like 10 or 15 junctions in them," he said, and he will still take the edge on quantum efficiency.
Hatridge sees the student pipeline changing shape rather than shrinking. Today's graduate students arrive with far more computer science than his cohort had and less low-temperature physics. Pitt now runs a physics and quantum computing undergraduate major built around that shift, and he expects joint CS and physics PhDs to become routine.
Graduate recruiting has gotten easier because industry exists. "There was not an industry that was gonna hire me to do ultra low field MRI," he said of his own student years. The pressure has moved to postdocs and junior faculty, where companies can offer money and a clearer path.
His answer for prospective students is deliberately open. There is no checklist. Some of his students love nanofabrication and some cannot stand it. "We're looking for folks who are curious, who are driven, and can show some evidence of doing pretty hard, random looking tasks."
Hatridge declined to name anything overhyped, preferring the freedom of not answering to a stock market. On what the field underestimates, he keeps returning to accumulation. "It's the question of how much complexity we can really master," he said, and whether each step toward fault tolerance gets easier or harder than the one before it.
His five-year wish list has one item. "I would like to see experiments that really attack fault tolerance directly," he said, pointing to a multi-logical qubit processor that checks all the boxes rather than a single error corrected qubit at breakeven.
He also has a useful measuring stick for how far the field has moved. "We were pretty excited to have a nine-hundred-nanosecond transmon when I was a postdoc," he said. "Now if you walk up and tell me you have a nine-hundred-microsecond transmon, I'm like, probably."
What keeps him optimistic is the part of academic research that industry cannot replicate. Every year brings students who are not beholden to what happened thirty years ago and are seeing the problem for the first time.
Hatridge's point about readout, that quantum efficiency belongs to the whole chain rather than any single component, is the assumption Qblox designs around. Our control and readout stack is built so the electronics are not the weak link between a good qubit and a good measurement, with multiplexed readout and real-time feedback that holds up as systems grow. If your group is working through readout fidelity or scaling a modular system, we would be glad to talk.
Contact us to learn how Qblox supports high-fidelity superconducting qubit readout.
