IBM - Rigetti Computing: A SPAC Implosion Might Be Unfolding

2023-07-26 03:54:09 ET

Summary

• Less than a year after its launch, RGTI has lost its founder, key managers and thrown out its technology roadmap.
• It's technology seems to be a long way behind key competitors, and as yet, there is no plan to catch up.
• Short of cash and customers, the large shareholders might be planning an exit strategy.

This is the second in a series I am writing about Quantum Computing Companies. It is a new industry with the potential for explosive growth and is subject to much hype. I track 7 companies in this area and am writing about the three that have shown enormous share price movements in the last 12 months.

I am hoping to explain the technology and theory as well as assess the prospects for each company. The first article was on D-Wave ( QBTS ) and described their algorithm to solve complex optimization problems. I liked D-Wave; it has a commercial product with numerous customers and offers a real advantage.

Rigetti: A company that lost its way before it found it.

Rigetti Computing Inc ( RGTI ) came to the Market in March 2022 with ambitious plans to beat IBM to the quantum computing market with a comparable technology roadmap.

In less than a year, the original plans are in tatters. The company's four senior managers have left or been dismissed, the technological road map sidelined, and the race toward quantum supremacy with IBM lost.

The new plan is to develop a hybrid classical-quantum computer; it might work for a while but will be redundant by 2027 when IBM achieves quantum advantage, and Rigetti will probably run out of funds in 2024.

The newly appointed managers have experience in orchestrating an out. I suspect that might be the plan for the Supernova investment team.

Quantum Computing The Next Giant Leap

Quantum particles (atoms, electrons, photons, etc.) have properties that mathematicians have shown could be exploited. Algorithms have been developed that solve complex problems requiring infinite time or infinite memory on classical computers in a concise amount of time on a quantum computer. I will try to explain another of the critical properties here in a manner accessible to all. I spent two decades teaching advanced mathematics, so hopefully, I can do this.

Mathematicians started producing algorithms for quantum computers in the 1930s when Alan Turing published his paper " On Computable Numbers, " describing a universal computing machine. When I studied for my undergraduate maths degree in the 1980s, a course on Computability was already compulsory. It covered what problems could be solved if computers that could do it were invented. The mathematics reached a new height in the mid-1990s when Peter Shor published "Shors Algorithm," which shows how to quickly find the prime factors of large numbers. Once running Shors Algorithm will break the encryption most current Internet financial transactions use.

Since then, the amount of mathematics has increased exponentially; when quantum computers arrive, these mathematical algorithms will solve problems that seem intractable in medicine, materials science, climate change, and energy production/storage.

The new mathematics exploits several properties found in quantum mechanics that are not present in the classical computers we use today. These properties are difficult to conceptualize, which has led to them being poorly understood.

Property 1 Superposition

This is one of the two fundamental properties required by the mathematical algorithms. (I will explain entanglement in my next article). Superposition means that the Qubit combines two states, usually called the base states and labeled /1> and /0> in Dirac notation.

I will explain it using an analogy.

Imagine you are the radar operator on a US warship, and the captain tells you an enemy ship is in the ocean. Before you turn on your radar, you don't know where the enemy ship is. That doesn't mean it is everywhere at once or in more than one place, but it will be in some combination of North and East.

You see a Blip like this once you turn on the Radar Machine.

The red dot is the enemy ship. Let us suppose it is 1 mile away. You need to give information to the gunner so he can aim. You could provide the two values marked as blue a and blue b; you could also have given the angle from the East and the distance of 1 mile; both would work for the gunner to aim.

These values are known, in quantum mechanics, as amplitudes (for maths geeks, we are talking about an Argand diagram with a, and b being the real and imaginary parts, and the angle would be theta from Euler's equation).

The enemy position is a superposition of the base states North and East.

Now let's suppose the gunner fires; in the time it took you to write down the coordinates, pass them to the gunner, for him to take aim and fire, it is quite possible that some wind blew or a big wave arrived and moved the enemy ship a hundred yards southeast, and the shell misses its target. A similar problem would arise if the enemy ship knew you had its position. It would get moving pretty quickly.

(This is decoherence and information leakage, the most problematic engineering areas of quantum computing)

If we assume that the distance to the enemy vessel is one mile (in quantum Mechanics, we fix it to be always = 1) then the ship could have been at any point on this circle.

A circle has an infinite number of points, meaning we could have got an endless number of values for a and b; some of these values would have been negative, and blue a would be negative if it were on the left of the North line.

Whenever we measure a qubit, we get these two amplitudes a and b, which can take up an infinite number of values. In classical computing, bits can be only 0 or 1.

Quantum Computing

The original view was that to have a quantum computer, you would need to isolate quantum particles and use some energy (normally microwaves) to manipulate their amplitudes, develop sensors to read the amplitudes, have a method for entangling the particles and then unentangling them when needed.

I will explain entanglement (another crucial property required by the math) in my next article, but for now, it is good enough to understand that qubits can be linked together.

The engineering problems are immense; firstly, the amplitudes of quantum particles are affected by everything; heat, radiation, magnetic fields, and other particles. This is the equivalent of the wind and waves in my radar example. As a result, they must be isolated from the world and run at temperatures so low they may not be achievable.

Secondly, the particles change their behavior if any information leaks, like the enemy ship realizing it has been spotted. (see the double slit experiment for how this works)

D-Wave and IonQ Inc. ( IONQ ) (amongst others) are trying to build systems that use isolated quantum particles and appear to be progressing. I have a position in D-Wave and will write about IonQ next.

It's not quantum, but it is good enough.

In 1999 a paper published by Yashunobu Nakamura suggested a design for an artificial qubit called a charge qubit using superconducting materials. It is an electric circuit, not a quantum particle, often called a Quantum Dot. This initial paper's development led to the Transmon qubit on which all current superconducting computers are based.

The Transmon Qubit is still an electrical circuit, not a quantum particle, this video explains it quite well .

These artificial qubits can provide all the properties needed to run the mathematical algorithms I discussed at the beginning of this piece.

International Business Machines ( IBM ), Alphabet Inc ( GOOGL ), and Baidu Inc (HK:9888) (along with many academic institutions) are investigating these artificial superconducting qubits. The engineering task is less daunting, and excellent progress appears to have been made.

In 2022 Baidu published its plans to build a full-stack superconducting quantum computer.

Google already claims a quantum advantage and is working on error correction .

I will discuss IBM in more depth as it is essential in the analysis of Rigetti.

Chad Rigetti

Chad Rigetti, a Yale Ph.D., was a research scientist working at IBM on their superconducting Quantum computer. Chad has published more than 20 papers on quantum computing, including essential work on the measurement of superconducting qubits and a critical technical advance called the Josephine Gate, without which a superconducting Qubit cannot function.

In 2013 Chad left IBM and founded Rigetti Quantum Computing, presenting an accelerated road map toward Quantum supremacy. (quantum supremacy is the proof that a quantum computer can solve problems that classical computers never will)

Rigetti computing quickly gained funding , Chad is a more than capable scientist, and it was quite likely that working out of the confines of IBM, he could move more quickly using his ideas and beat IBM in the race to supremacy.

In 2022 Rigetti Quantum Computers performed a reverse merger with the blank cheque company Supernova Partners Acquisition Company II Ltd (SNII), initially valuing the company at $1.5 billion. Rigetti Computing received $261 million in gross proceeds and started trading under the symbol RGTI on the Nasdaq on Mar 2, 2022.

The SPAC unravels immediately.

We have seen an alarming number of these SPAC deals implode over the last few years, but the speed at which things went wrong for Rigetti is astonishing. By November 2022, less than nine months after the launch, the concept seemed to unravel. Chad Rigetti resigned, although this line from the FY 2022 report suggests it might not be as simple as that.

Rigetti reimbursed Dr. Rigetti for legal fees paid in the amount of $168,957 in connection with internal matters and the negotiation of his separation from the company in fiscal 2022.

( RGTI: 2022 10-K, 2023-3-27 )

The matter clouds further with

Dr. Rigetti prior to his resignation, Mr. Harburn prior to his departure, and Mr. Naidu prior to his termination, was entitled, to the following severance benefits pursuant to his respective employment agreement if his employment is terminated by Rigetti without "Cause" or by the executive for "Good Reason," and provided such executive officer timely executes and does not revoke a release of claims in Rigetti's favor,

( RGTI: 2022 10-K, 2023-3-27 )

Within its first year, Rigetti had lost its four most senior officers.

CEO Chad Rigetti: Resigned Nov 12, 2022

CFO Brian Serada: Employment ended Feb 9, 2023

CTO Michael Harburn: Departed Feb 9, 2023

COO Taryn Naidu: Employment Terminated Apr 18, 2022.

The old management was not the only thing cast aside.

The entire business strategy was changed, the technological roadmap set aside, and 28% of the staff let go. (Q1 earnings call CEO remarks)

Technological Roadmap

IBM has been producing a technological roadmap for Quantum computing that they update periodically. They have kept to the plan and the timeline with impressive resolve.

When Chad started Rigetti Computing, he continued this idea, and the new management continued also. Comparing the four roadmaps of IBM with Rigetti before the SPAC after the SPAC and after the meltdown is worthwhile.

The latest version of the IBM roadmap is

The growth in the number of Qubits is essential; for a complete error-correcting Quantum computer that can run all the developed mathematical algorithms, we need 1,000,000 Qubits. IBM expects to hit 100,000 in 2026 and scale from there. The timeline shows that error mitigation changes to error correction as the scaled 100K cubit machine arrives.

Quantum software applications are expected in 2025. At this point, it should become a revenue generation line for IBM and likely become the most significant one sometime after 2027 when full error correction arrives.

In the Q1 2022 earnings report, Chad Rigetti gave the following roadmap and dates. (this is the post spac roadmap)

2023 84 qubit chip and 336 qubit chip at the end of 2023

2025 1,000 qubit chip

2027 4,000 qubit chip

This was a significant walk back from the pre-SPAC timeline. Chad blamed the slowdown on three things. Higher than expected costs, supply chain constraints, and cash raised from the business combination being less than expected. In retrospect, this was an early warning regarding friction between the Rigetti camp and the Supernova team.

The initial Investor presentation presented the following roadmap (slide 26) and shows that Chad was targeting an entire error-correcting quantum computer by 2027, perhaps a year or two before IBM

The Pre-SPAC roadmap

2021 80+ qubit chip

2024 1,000 Qubit chip

2026 4,000 Qubit Chip

2027+ 1,000,000 Qubit Chip fault tolerant

The original roadmap was very similar to IBM at that time, where IBM met targets and presented an accelerated timeline Rigetti missed targets, removed its senior managers, ditched the timeline, and effectively ended its search for Quantum supremacy.

In the latest earnings call (Q1 2023), management said we could no longer rely on the Q1 2022 roadmap.

The new roadmap in the investor presentation Mar 2023 is a complete departure from chasing numbers of Qubits. Rigetti is now suggesting a hybrid approach towards Quantum Advantage (QA is the point at which Quantum computing offers distinct advantages over classical computing). Rigetti anticipates an 84 Qubit chip at the end of 2023 and will not start to work on the 336 Qubit chip until it has managed to create a Quantum Advantage using its Ankaa-2 84 Qubit chip in a hybrid manner.

Rigetti doesn't think it is about the number of Qubits

In the last earnings call, the CEO set out the new roadmap. He ditched the chase for more qubits and changed the conversation tone towards the quality of qubits; much of the focus appears to be on gate fidelity.

Gate fidelity is a measure of noise; if fidelity is too low, errors will accumulate faster than they can be corrected, and calculations will fail. Gate fidelity will probably need to be over 99%, and in March this year, Rigetti demonstrated 99.5% fidelity.

The CEO said

Our Ankaa-2 84-qubit system with anticipated improvements in design and performance is expected to be deployed and made available to external customers in the fourth quarter of 2023. We remain committed to working to achieve 2-qubit fidelity of 99% with the anticipated Ankaa-2

The problem is IBM have already demonstrated 99.9% fidelity with their two-qubit gates.

Hybrid Quantum Computing

Rigetti explains its approach in this slide.

Rigetti suggests this system can run in the cloud and is a solution its commercial partners will use.

Microsoft Corporation ( MSFT ) offers Rigetti Hybrid and solutions from IONQ, Quantinuum, Pasqual, and Quantum Circuits Inc on its Azure platform, allowing customers to test ideas and begin planning for Quantum computing.

The Hybrid concept is a little misleading; all quantum computers are linked to classical computers, they control the configuration of the quantum computer, input the algorithms and receive and then interpret the results. Quantum Hybrid attempts to increase this relationship and have the two machines working in parallel on the same problem splitting the task into parts best suited to each device.

At present, it is the only way to take advantage of quantum computing and is an excellent way for companies and organizations to begin to prepare for the quantum revolution.

Commercial Progress

Can it work? Do they have the customers and applications to make money with this hybrid approach?

The first thing to note is that once IBM releases their 2027 scaled error corrected +100,000 Qubit system, the hybrid will likely be obsolete, along with any other supercomputing system. A fully error-corrected Quantum Computer will make any system relying on classical computing look slow and cumbersome.

Rigetti only has 1 Commercial Customer.

Despite the many logos and icons in the 2023 Roadmap analysis of Q1 2023, 10K suggests they only have one commercial customer.

Customer C would appear to be Rigetti's only significant commercial customer (adding customers A, B, D in 2023 and 2022 = the percentage of revenue from government entities). Revenue from government organizations is currently development grants and milestone payments, not the sale of products. (CFO Q1 remarks) The four customers account for 92% of their income, some of the other 8% must come from the Azure platform)

Revenue from Customer C was $396,000 in Q1 2023 and $484,000 in Q1 2022, a decrease of 18%. The initial Investor presentation (page 41), lists leading customers as Amazon Web Services, NASA, DARPA, and the US Airforce. I will assume that AWS is Customer C; Amazon provides Quantum computing through its bracket interface that allows customers to experiment with the different types of quantum computers. Rigetti is one of two superconducting Qubit suppliers on the AWS Bracket. The other is the Harvard/MIT linked QuEra and it has a 256 Qubit chip named Aquila already operating, putting Rigetti a long way behind on this platform.

Rigetti hardware seems to be quite away behind all of its major competitors which might account for the recent change in emphasis.

The New management team

After cleaning out the old management who were the Quantum computing heritage of Rigetti computing, a new team has swept into place. They have experience selling companies to bigger entities.

New CEO appointed Dec 2022

Subodh Kulkarni was formerly CEO of CyberOptics from 2014 until Nordson acquired it in November 2022.

New CFO appointed Dec 2022

Jeff Bertlesen, formerly CFO CyberOptics from 2005 appointed COO in 2014 until Nordson acquired it in November 2022.

A new CTO was appointed from within the company, and the COO has not been replaced.

Financials

The balance sheet summary gives a solid impression (figures Last reported quarter)

However, in the Q1 2023 earnings call, the CFO said they must raise more funds towards the end of 2024. To raise money in 2024, Rigetti must make significant progress and develop some kind of competitive advantage.

They may have an advantage in the new hybrid approach that, as yet, I am unaware of, or they may have a use case that has not been made public. They may be about to sign a new large customer or have developed a quantum hybrid application that will change the landscape.

I will be watching carefully to see if that is the case.

Conclusion

With the founder gone, the technological roadmap sidelined, and the path to quantum supremacy closed, I find it difficult to see the value in Rigetti.

The new Hybrid computing plan would appear to have a very short shelf life, and it isn't easy to see how it will ever generate enough cash to fund this business.

The value in Rigetti may be in its remaining staff; they have many highly qualified quantum computing research scientists who have developed intellectual property that other companies will value. The rest of the company has been streamlined with a 28% cut in headcount.

Perhaps the roadmap is now a Hireover with IBM, Google and Baidu seen as future employees, not current competitors.

I will not be buying Rigetti, and if I held shares, I would offload now to take advantage of the recent share price appreciation.

For further details see:

Rigetti Computing: A SPAC Implosion Might Be Unfolding