EP04 Updates on AI, Semiconductors, future power demand and Quantum Computing

Summary:

In this episode, we discuss the growing need for energy to support advancements in artificial intelligence (AI) and semiconductors. We expresses concern about the potential energy shortage that could arise as the U.S. increases its domestic production of semiconductors and AI technology. This is compounded by the ongoing transition to less energy-dense, intermittent renewable energy sources. We highlight quantum computing (QC) as a promising solution to reduce energy consumption in AI, emphasizing its potential to revolutionize machine learning and significantly increase computational efficiency. While QC technology is still in its early stages, we expresses optimism about its rapid development and potential to alleviate the energy challenge.

 Questions to consider as you read/listen:

 How will the increasing energy demands of AI and semiconductors impact the transition to renewable energy?

1. What are the potential benefits and challenges of quantum computing for accelerating the development of AI and solving complex problems?
2. What is the current state of quantum computing, and what are the key factors driving its development and potential for widespread adoption?

 

Long format:

  

Updates on AI, Semiconductors, future energy demands and Quantum Computing 

 Over the last few days we have seen a positive uptake in all things having to do with reshoring America’s semiconductor and AI efforts. 

 Between 

 1 the funding of WolfSpeed (see my earlier post yesterday on SiC technology)

 2 continued big Ai user deals with nuclear (see my post last week on Microsoft and yesterday on Google)

 3 the Spruce Pine facility not to be terminally destroyed

 We can add two more events

 4 DOE announces the opening of applications for up to $900 million in funding to support the initial domestic deployment of Generation III+ (Gen III+) small modular reactor (SMR) technologies (https://www.energy.gov/articles/biden-harris-administration-announces-900-million-build-and-deploy-next-generation-nuclear#:~:text=WASHINGTON%2C%20D.C.%20%E2%80%94%20As%20part%20of,modular%20reactor%20(SMR)%20technologies)

 5 And more advancements in the technology involved in the quest for quantum computing and the energy sector’s acknowledgment of how much energy AI will take and how investing in QC will potentially drive that down (up to 100 times more efficient). https://oilprice.com/Energy/Energy-General/Can-Quantum-Computing-Solve-AIs-Energy-Crisis.html 

 According to this article and quoting from it…

 By 2030, AI is expected to represent 3.5 percent of the global electricity consumption, and 9 percent of electricity generation in the United States (a sharp increase from the country’s current rate of around 3.5 percent  – already a hefty amount). Put together, electric vehicles and AI are on track to add 290 terawatt hours of electricity demand to the United States energy grid by the end of the decade according to projections by Rystad Energy.

 End quote

 It is refreshing to see folks start to wake up and realize that we are in a near emergent situation because where do we get all of our future energy from as AI and semiconductors and ancillary technology will gobble up and enormous amount of energy especially considering our growth and unprecedented reshoring efforts due to the US’s retreat from globalization and further massively complicated by our political commitment to move from power dense sources to intermittent very low energy/power density (yet clean) sources. 

 Folks, we simply cannot continue on this path. There’s not enough energy. 

 We have a very long way in QC but the pace of development is encouraging. 

 For background Quantum Computers will be a great accelerator of AI, and quantum computing has the potential to revolutionize machine learning and solve problems that were once viewed as impossible. 

 A quantum computer (QC) is a new computer architecture that uses quantum mechanics to perform certain kinds of computation much more efficiently than a classical computer can. 

 Oversimplified, classical computers are based on “bits“. A bit is like a switch. It could be either zero (if off) or one (if on). Instead of bits, QC use quantum bits, or qubits, which are typically subatomic particles such as electrons or photons. Qubits follow principles of quantum mechanics regarding how atomic and subatomic particles behave, which include unusual properties that give them super processing capabilities.

 The first property is superposition, or the capability for each qubit to be in multiple states at any given time. This allows multiple qubits in superposition to process a vast number of outcomes simultaneously. 

 If you ask AI on a classical computer to figure out how to win a game, very oversimplified and not entirely 100% accurate but close, it will try various moves and take them back in its “head” until it finds a winning path. But an AI built on QC will try all moves, extremely efficiently, holding uncertainty in its head, resulting in an exponential reduction of complexity. 

 The second unusual properties that give them super processing capabilities and superiority over classical computers is entanglement, which means two qubits remain connected so the actions performed on one affect the other, even one separated by great distances. Thanks to entanglement every qubit added to a quantum machine EXPONENTIALLY increases its computer power. To double a $100 million classical super computer you’d have to spend another $100 million. To double your quantum computing, you just need to add one more qubit.

 <<<QC is very fragile in the sense that even slight vibrations, electrical interferences, temperature changes, or magnetic waves can cause super position to decay or even disappear. To make a workable and scalable QC, researchers have to invent new technologies and build unprecedented vacuum Chambers, super conductors, and super cooling refrigerators to minimize these losses in quantum coherence, or decoherence, caused by environment.>>>

 Because of these challenges, it is taking a long time for scientists to increase the number qubits in QC – from two in 1998 to 65 and 2020, which is still too few to do anything useful. However, even on a few dozen, some computing task can be accomplished with QC over 1 million times faster than classical computers. IBM’s roadmap shows the number of qubits more than doubling every year for the next three years. 

 Finally to know, quantum computers are programmed differently from computers, algorithms will need to be invented, and new software tools will need to be built.

 As of today, quantum computing is considered to be in the "noisy intermediate-scale quantum (NISQ)" era, meaning that while significant progress has been made, current quantum computers are still prone to errors, relatively small in scale, and not yet capable of performing complex calculations without significant error correction, making them largely experimental and not yet ready for widespread practical applications.

 As of today, the state-of-the-art in qubit technology involves processors with several hundred qubits, with companies like IBM recently achieving over 400 qubits on their "Osprey" chip, but significant challenges remain in terms of error correction and scaling up to truly useful quantum computing applications, often requiring thousands of high-quality qubits to create a single reliable "logical qubit" through error correction techniques; further research is needed to address these limitations.

 

——//Opinion time//——-

 Again, in my opinion (and that’s all it is) and for what it is worth, this semiconductor/AI war is the conflict that must be won. We must reshore every component from silicon crucibles to fabrications of chips to design and production to lithography (very very tough nut to crack due to Dutch dominance) to then to actual use. 

 We can’t be on the final stretch and run out of energy/power!!!! That is my fear. 

 PZ beats the drum way way before it was popular that China was going to decline and is destined to have a supreme reversal of fortunes. I’m beating the drum of where are we getting the energy given our three competing interests (AI/Chips, general reshoring, moving to less energy dense intermittents). 

 Events like WolfSpeed and SiC and QC give me hope that it’s not going to be as bad. 

 All of these recent announcements and articles are very encouraging to me. 

 To do otherwise than to do what we are doing makes it so America runs the risk of losing its position as top of the Global Value Chain (GVC) in my opinion. 

 

Sources:

 https://medium.com/edge-elections/the-state-of-the-art-in-quantum-computing-cffd654c363f#:~:text=According%20to%20this%20roadmap%2C%20IBM,scaling%20without%20physics%20limitations'%E2%80%9D.

 

https://www.orfonline.org/research/quantum-computing-current-scenario-and-future-prospects#:~:text=According%20to%20recent%20estimates%2C%20each%20functional%20or,progress%20being%20made%20on%20a%20continual%20basis.&text=In%202021%2C%20IBM%20developed%20a%20127%2Dqubit%20chip,followed%20by%20the%20433%2Dqubit%20Osprey%20in%202022

 

https://www.nature.com/articles/d41586-023-01692-9#:~:text=Some%20firms%20are%20so%20optimistic,at%20the%20University%20of%20Helsinki

 

https://link.springer.com/article/10.1007/s13222-024-00467-4#:~:text=The%20current%20state%20of%20quantum,of%20continuous%20quantum%20error%20correction

 

https://theconversation.com/quantum-computers-in-2023-how-they-work-what-they-do-and-where-theyre-heading-215804#:~:text=What%20is%20the%20current%20landscape,still%20some%20way%20from%20perfection.&text=Today's%20machines%20are%20of%20modest,goal%20via%20diverse%20technological%20approaches

 

https://crsreports.congress.gov/product/pdf/R/R47685#:~:text=Since%20the%20enactment%20of%20the%20NQI%20Act%20in%202018%2C%20researchers,support%20under%20the%20NQI%20Act