Google’s Septillion-Year Challenge

Here, from Forbes, is a representative sample of the breathless reporting inspired by the recent announcement of Google’s “Willow” chip for quantum computing.

“The Willow team’s claim that it can perform a complex calculation in just five minutes that would take a supercomputer 10 septillion years is mind-boggling. To put this into perspective, 10 septillion years is roughly 700 billion times the current estimated age of the universe of about 13.8 billion years.”

It certainly boggles our minds. In fact, it boggles our minds how much it’s boggling our minds (as some Joseph Heller character once exclaimed). We never expected to be so boggled.

One good boggle, however, deserves another. So, boggle us this: How would you set up a problem that would take a classical computer 700 billion times the current estimated age of the universe to solve? Heck, who would type it in?

Yes, we get that we are talking about an entirely different type of computing machine. But that does not get us past the issue of how one asks a question requiring septillions of years to answer. Even if the operation is a relatively straightforward mathematical problem, such as factoring the key to a heretofore unbreakable code, would it take 10 years to enter the number? A thousand. A million? Even if the number were spit out by a random number generator from a classical computer, generating billions of digits per second, how long would it take?

This is not a frivolous question. Quantum computing has great potential for solving highly specific problems. But in computation, the actual calculations are never the whole of the challenge. First the problem–the data–must be entered (input) and then the results of the calculation must be released to the user in a comprehensible form (output).

Input/Output or I/O systems are modest helpmates to classical computers’ powerful combinations of logic and memory. But there is a type of computer for which I/O is most of the game, and for which setting up the game takes longer than playing it.

Analog computers are I/O dominated machines. Consider the old record player, pumping out tunes from a vinyl disk. The computation is the process of interpreting the indented grooves on the vinyl as sound played out of the speaker. As with most analog devices, the “computation” is relatively straightforward as the indentations are “analogous” to the frequencies and amplitudes they actuate. The work is mostly in the I/O.

• Find an orchestra
• Have the musicians play Beethoven’s Fifth.
• Record the music into an analog device.
• Make a vinyl disk
• Represent the recording on the physical indentations in that vinyl disk.
• Play the record on the turntable.

Then do the computation, i.e., turn the indentations into sounds.

• Then reverse the process through an assemblage of preamplifiers, amplifiers, speakers and all else to produce an output worthy of Mr. B.

As with a digital computer, you can use the record player again and again. You can compute the indentations of many different vinyl disks into all sorts of music. But though the computer remains unchanged, for each new piece of music the Input stage–representing the bulk of the effort–must be repeated and even the output adjusted to best deliver, say, heavy metal rather than baroque.

The quantum computer is analogous to the analog computer: most of the work goes into setting the problem. And then, also like analog computers, a great deal of energy goes into the correction of errors that would otherwise accumulate catastrophically, a problem digital computers in their simple world of on and off, zero and one, do not share.

This is not to say the quantum computer has no future. It has not only a future but a present. Though most of the quantum burden is now being carried by giants of the industry–Google, IBM, Amazon being especially active–there are also several start-ups with revenues, credible business plans, and distinct technologies. At this point, however, most of the revenues–and most of the investment–are from government or from research projects.

At this early stage, all projections are hazardous. Yet we know enough to make some tentative projections for the quantum market for this decade.

Quantum computing chips are the backbone of quantum systems. In 2023, the global quantum computing chip market was valued at $480 million and is projected to grow to $5.6 billion by 2030 at a compound annual growth rate (CAGR) of 33%. That’s an impressive pace, but keep in mind that sales of traditional digital chips will exceed $600 billion this year.

Table 1: Global Quantum Computing Chip Market Size and Growth Projections

Source: The Information Network

Quantum chips are slowly transitioning from niche research tools to practical hardware in industries with especially daunting computational challenges such as AI, cryptography, optimization, material science, and drug discovery. (Optimization simply means taking a task already ongoing and rendering more efficient and effective. Logistics companies like DHL, for instance, are testing quantum algorithms to simulate and improve delivery routes.)

Table 2: Quantum Chip Market by Application Area

Source: The Information Network

Investment in the field is rising quickly, but again dominated by government. Last year, the proportion of government funding to venture investment stood at 4.5 to 1, though down from 7 to 1 at the beginning of the decade.

Table 3: Investment in Quantum Chip Technologies

Source: The Information Network

For all this impressive news, given the cost and complexity of deploying it, quantum computing still faces fundamental questions. The biggest question may still be “what can we use it for?”

To help answer that question, Amazon Web Services (AWS) on Nov. 22 of this year, launched the Quantum Embark program to help businesses identify relevant quantum applications, develop technical proficiency, and establish a strategic roadmap for quantum adoption. The very announcement of Embark sent stocks of several quantum start-ups soaring.

• Shares of Quantum Computing Inc. (QUBT) surged by 28% by day’s end.
• D-Wave Quantum Inc. (QBTS) was up 48% by the end of the trading day.
• Rigetti computing is up 470% since the AWS announcement.

The graph below shows the astonishing rise in the fortunes of some truly tiny quantum stocks this year. The current valuation of every one of these companies is orders of magnitude greater than their annual revenue. IONQ, with a market valuation of more than $7 billion, brought in only $37.5 million in revenues over the past 12 months. QUBT has not yet passed half a million dollars in sales but is valued at $800 million. It’s price-to-sales ratio stands at over 2,000, as we write.

None of this is sustainable without exponential growth in real revenues, not government grants or even private research projects.

Quantum computers are still in their infancy, temperamental and fragile. They need environments colder than outer space to keep their delicate qubits stable, and they’re prone to errors that make even the steeliest engineers sweat. Yet their potential to solve admittedly highly specialized problems is undeniable.

We write about technology companies. We love what we do. But we must always remember that the real money goes not to those who make the machines but those who use them to transform industries and make life more abundant for the rest of us. Global GDP now exceeds $110 trillion, most of it dependent on semiconductors. Yet the semiconductor industry itself accounts for less than 0.6% of all those trillions.

That said, in next month’s issue of the Gilder Technology Report, we will put all the companies mentioned here and more under our investment microscope to search out specks of gold.