Quantum computing is cool again
Intel has been on quantum computing kick for five years, but recent news from competitor Google has them reaching out to tell their story.
Intel has been showing off its production facilities in Washington County and talking about probability — mainly the probability of anyone making a quantum computer that actually works.
The theory behind quantum computing is that, freed from the limits of binary bits, it can use quantum bits (qubits) to perform massive amounts of calculations in parallel. Qubits don't have to be ones or zeros, they can be in states in between. The ability to be in multiple states simultaneously allows for many more combinations, which, in turn, carries more information.
Google recently ran a benchmark test known as "quantum supremacy," which showed its chip was faster than a "classical" computer. Google ran an analysis in 200 seconds on a small quantum processor that would take the most powerful supercomputer approximately 10,000 years to perform.
That speed and power have commercial entities thinking it would be great for economic modeling or creating new drugs.
Jim Clarke, the director of quantum hardware at Intel Corporation, took the Business Tribune on a tour of the facility where they are working on their chip. At Ronler Acres, next to the gigantic D1X fab, he led us through what looked like a cross between NASA and a Hollywood studio, with staff buzzing around in golf carts and carpenters coming and going.
Clarke used the metaphor of qubits as spinning coins — neither heads nor tails, neither ones nor zeroes, but possibly either — in his pitch to Intel's executive committee in 2014. It worked.
"We're trying to make a quantum analog of a transistor, that for certain applications would be exponentially faster than a classical computer that you might have or I might have today," he explained.
He took us to a spot he calls "the coldest place in Oregon." It's a room with three fridges, shaped like water heaters, in which they test their quantum chips. The chips only operate at extremely low temperatures, in this case, 10 milliKelvins or 100th of a degree above absolute zero.
The sounds in the room are of hisses and clanks from the pumps that cool the fridges. The coolant is a mixture of Helium 4 and Helium 3, the latter being a by-product of nuclear reactors.
Wires run from the fridges to what looks like a rack of servers. These boxes of electronics are marked with brand names such as Keithley (read out scopes, owned by Tektronix), while others have labels made by a label maker showing the name and phone number of whom to call with questions.
A Finnish company called Bluefors make the fridges. They take a weekend to cool down. Inside the onion layers of steel casing, Clarke says there is a device that looks like a chandelier, feeding coolant downwards to the coldest part, the bottom, where the chip sits.
"A tiny bit of light, if it were to get into one of our refrigerators, would warm things up to the point that we would lose all of our information. So, this is very different than a conventional computer."
The coolant room is monitored from a separate space full of computers and screens.
Watch this space
Observing them, like a guard scanning a mall's security camera outputs, is not too stressful a job. Clarke drops by that room frequently and sometimes enters the fridge room.
Observing a quantum particle, or a qubit, however, is more complicated. Quantum computers are hard to produce because the qubits are so fragile. They fail due to excess "noise": electronic, temperature, vibration, radiation.
At Intel, they use just one electron in the circuit, and the only way to observe it is to place another electron nearby and see how that electron behaves. (How they observe the second one is too complicated to explain.)
To complicate things further, Clarke said, "Right now there are probably between eight and 10 different types of quibits or quantum computers. We're actually one of the few that's actually studying this type."
Transistors — those on-off switches that are the building blocks of electronics — are Intel's bread and butter.
"What we do is we modify this transistor slightly so that we actually trap a single electron in the device. And that electron can either have one of two spins: spin up, spin down." Intel picked this path because it resembles their usual business. Others have gone their own way. Other ways of making qubits include lasers and superconductors.
"Google, IBM and others, they're doing superconducting qubits. Intel is doing these qubits that look like transistors. The primary reason is we have tens of billions of dollars of equipment dedicated to transistors (at Ronler Acres). Intel ships 400 quadrillion transistors a year. So, if we can build a quantum technology that's built on the core fundamentals of transistors, we think we're going to be we're going to lead in this technology."
While these Intel chips are different from transistors, they use the same silicon as a regular chip. One way they have the edge over Google? Scale.
"Each superconducting qubit is a million times larger than one of the spin qubits that we're making. Therefore, it's relatively easy to make one of these with a less advanced fabrication facility. And many of these companies are starting to build small fabrication facilities."
Intel can build these in the same fab where they develop their regular transistor-based processors, and Clarke expects they will be able to scale up production faster than the competition.
He points out that Intel is known for its fabs, and for its architecture, the bed on which the chip sits.
"So, we can tap into that network, that primarily happens over in the Jones Farm campus about five miles to the west of here."
How close is a usable quantum computer to being ready?
"We absolutely have a goal to build a complete system. We're working on the quantum processor, the control electronics, the architecture and the algorithm. We are focused on the long term, which is building a quantum computer that's going to change your life or mine."
One of the technologies that motivated quantum computers was cryptography. Peter Shor proposed an algorithm in the mid-1990s (Shor's Algorithm) for breaking the RSA encryption that protects things like e-commerce. That this might be near happening in the 2020s gets the attention of developers.
As well as these computations, which could be done on a classical computer (but would take years), Clarke says quantum computing has another weakness right now.
"We're doing problems where we have to make so many assumptions about the state of our system, our approximations, that the answer at the end is not as accurate as it could be. An example for that would be in the chemical or medical space where we have to make assumptions about how a protein behaves. We get an answer at the very end, but that answer is not as exact as it would be a supercomputer."
Wait 12 years
A quantum computer that doesn't need to be on life support, as it were, in a lab, could be years away.
What we see now is just test systems, with no practical purpose.
"It's not going to be enough to build just a quantum processor. We're going to have to build the infrastructure. So, to build a system that's going to change your life, I am still thinking about 10 years away."
He points out that big changes in electronics come roughly every 12 years, and quantum computing looks like it could be the next big thing.
"At first, I would expect just a few systems used by the government, (the Department of Energy's) National Labs, large banks and hospitals. But as the application space grows, I would expect that to proliferate to far more."
Reporter, The Business Tribune
Follow us on Twitter, Facebook and Instagram
Subscribe to our E-News
You count on us to stay informed and we depend on you to fund our efforts. Quality local journalism takes time and money. Please support us to protect the future of community journalism.