The Transistor of 2047: Expert Predictions – IEEE Spectrum

IEEE websites place cookies on your device to give you the best user experience. By using our websites, you agree to the placement of these cookies. To learn more, read our Privacy Policy.
What will the device be like on its 100th anniversary?
The luminaries who dared predict the future of the transistor for IEEE Spectrum include: [clockwise from left] Gabriel Loh, Sri Samavedam, Sayeef Salahuddin, Richard Schultz, Suman Datta, Tsu-Jae King Liu, and H.-S. Philip Wong.
The 100th anniversary of the invention of the transistor will happen in 2047. What will transistors be like then? Will they even be the critical computing element they are today? IEEE Spectrum asked experts from around the world for their predictions.
Expect transistors to be even more varied than they are now, says one expert. Just as processors have evolved from CPUs to include GPUs, network processors, AI accelerators, and other specialized computing chips, transistors will evolve to fit a variety of purposes. “Device technology will become application domain–specific in the same way that computing architecture has become application domain–specific,” says H.-S. Philip Wong, an IEEE Fellow, professor of electrical engineering at Stanford University, and former vice president of corporate research at TSMC.
Despite the variety, the fundamental operating principle—the field effect that switches transistors on and off—will likely remain the same, suggests Suman Datta, an IEEE Fellow, professor of electrical and computer at Georgia Tech, and director of the multi-university nanotech research center ASCENT. This device will likely have minimum critical dimensions of 1 nanometer or less, enabling device densities of 10 trillion per square centimeter, says Tsu-Jae King Liu, an IEEE Fellow, dean of the college of engineering at the University of California, Berkeley, and a member of Intel’s board of directors.
"It is safe to assume that the transistor or switch architectures of 2047 have already been demonstrated on a lab scale"—Sri Samavedam
Experts seem to agree that the transistor of 2047 will need new materials and probably a stacked or 3D architecture, expanding on the planned complementary field-effect transistor (CFET, or 3D-stacked CMOS). [For more on the CFET, see “Taking Moore’s Law to New Heights.”] And the transistor channel, which now runs parallel to the plane of the silicon, may need to become vertical in order to continue to increase in density, says Datta.

AMD senior fellow Richard Schultz, suggests that the main aim in developing these new devices will be power. “The focus will be on reducing power and the need for advanced cooling solutions,” he says. “Significant focus on devices that work at lower voltages is required.”
It’s hard to imagine a world where computing is not done with transistors, but, of course, vacuum tubes were once the digital switch of choice. Startup funding for quantum computing, which does not directly rely on transistors, reached US $1.4 billion in 2021, according to McKinsey & Co.
But advances in quantum computing won’t happen fast enough to challenge the transistor by 2047, experts in electron devices say. “Transistors will remain the most important computing element,” says Sayeef Salahuddin, an IEEE Fellow and professor of electrical engineering and computer science at the University of California, Berkeley. “Currently, even with an ideal quantum computer, the potential areas of application seem to be rather limited compared to classical computers.”
Sri Samavedam, senior vice president of CMOS technologies at the European chip R&D center Imec, agrees. “Transistors will still be very important computing elements for a majority of the general-purpose compute applications,” says Samavedam. “One cannot ignore the efficiencies realized from decades of continuous optimization of transistors.”
Twenty-five years is a long time, but in the world of semiconductor R&D, it’s not that long. “In this industry, it usually takes about 20 years from [demonstrating a concept] to introduction into manufacturing,” says Samavedam. “It is safe to assume that the transistor or switch architectures of 2047 have already been demonstrated on a lab scale” even if the materials involved won’t be exactly the same. King Liu, who demonstrated the modern FinFET about 25 years ago with colleagues at Berkeley, agrees.
But the idea that the transistor of 2047 is already sitting in a lab somewhere isn’t universally shared. Salahuddin, for one, doesn’t think it’s been invented yet. “But just like the FinFET in the 1990s, it is possible to make a reasonable prediction for the geometric structure” of future transistors, he says.
AMD’s Schultz says you can glimpse this structure in proposed 3D-stacked devices made of 2D semiconductors or carbon-based semiconductors. “Device materials that have not yet been invented could also be in scope in this time frame,” he adds.
Experts say that the heart of most devices, the transistor channel region, will still be silicon, or possibly silicon-germanium—which is already making inroads—or germanium. But in 2047 many chips may use semiconductors that are considered exotic today. These could include oxide semiconductors like indium gallium zinc oxide; 2D semiconductors, such as the metal dichalcogenide tungsten disulfide; and one-dimensional semiconductors, such as carbon nanotubes. Or even “others yet to be invented,” says Imec’s Samavedam.

"Transistors will remain the most important computing element"—Sayeef Salahuddin
Silicon-based chips may be integrated in the same package with chips that rely on newer materials, just as processor makers are today integrating chips using different silicon manufacturing technologies into the same package, notes IEEE Fellow Gabriel Loh, a senior fellow at AMD.
Which semiconductor material is at the heart of the device may not even be the central issue in 2047. “The choice of channel material will essentially be dictated by which material is the most compatible with many other materials that form other parts of the device,” says Salahuddin. And we know a lot about integrating materials with silicon.
Everywhere. No, seriously. Experts really do expect some amount of intelligence and sensing to creep into every aspect of our lives. That means devices will be attached to our bodies and implanted inside them; embedded in all kinds of infrastructure, including roads, walls, and houses; woven into our clothing; stuck to our food; swaying in the breeze in grain fields; watching just about every step in every supply chain; and doing many other things in places nobody has thought of yet.
Transistors will be “everywhere that needs computation, command and control, communications, data collection, storage and analysis, intelligence, sensing and actuation, interaction with humans, or an entrance portal to the virtual and mixed reality world,” sums up Stanford’s Wong.
This article appears in the December 2022 print issue as “The Transistor of 2047.”
Samuel K. Moore is the senior editor at IEEE Spectrum in charge of semiconductors coverage. An IEEE member, he has a bachelor's degree in biomedical engineering from Brown University and a master's degree in journalism from New York University.
Batteries expose supply-chain and skills gaps
Robert N. Charette is a Contributing Editor to IEEE Spectrum and an acknowledged international authority on information technology and systems risk management. A self-described “risk ecologist,” he is interested in the intersections of business, political, technological, and societal risks. Charette is an award-winning author of multiple books and numerous articles on the subjects of risk management, project and program management, innovation, and entrepreneurship. A Life Senior Member of the IEEE, Charette was a recipient of the IEEE Computer Society’s Golden Core Award in 2008.
A General Motors Hummer EV chassis sits in front of an Hummer EV outside an event where GM CEO Mary Barra announced a US $7 billion investment in EV and battery production in Michigan in January 2022.
“Energy and information are two basic currencies of organic and social systems,” the economics Nobelist Herb Simon once observed. A new technology that alters the terms on which one or the other of these is available to a system can work on it the most profound changes.”

Electric vehicles at scale alter the terms of both basic currencies concurrently. Reliable, secure supplies of minerals and software are core elements for EVs, which represent a “shift from a fuel-intensive to a material-intensive energy system,” according to a report by the International Energy Agency (IEA). For example, the mineral requirements for an EV’s batteries and electric motors are six times that of an internal-combustion-engine (ICE) vehicle, which can increase the average weight of an EV by 340 kilograms (750 pounds). For something like the Ford Lightning, the weight can be more than twice that amount.
EVs also create a shift from an electromechanical-intensive to an information-intensive vehicle. EVs offer a virtual clean slate from which to accelerate the design of safe, software-defined vehicles, with computing and supporting electronics being the prime enabler of a vehicle’s features, functions, and value. Software also allows for the decoupling of the internal mechanical connections needed in an ICE vehicle, permitting an EV to be controlled remotely or autonomously. An added benefit is that the loss of the ICE power train not only reduces the components a vehicle requires but also frees up space for increased passenger comfort and storage.
The effects of Simon’s profound changes” are readily apparent, forcing a 120-year-old industry to fundamentally reinvent itself. EVs require automakers to design new manufacturing processes and build plants to make both EVs and their batteries. Ramping up the battery supply chain is the automakers’ current “most challenging topic,” according to VW chief financial officer Arno Antlitz.
It can take five or more years to get a lithium mine up and going, but operations can start only after it has secured the required permits, a process that itself can take years.
These plants are also very expensive. Ford and its Korean battery supplier SK Innovation are spending US $5.6 billion to produce F-Series EVs and batteries in Stanton, Tenn., for example, while GM is spending $2 billion to produce its new Cadillac Lyriq EVs in Spring Hill, Tenn. As automakers expand their lines of EVs, tens of billions more will need to be invested in both manufacturing and battery plants. It is little wonder that Tesla CEO Elon Musk calls EV factories “gigantic money furnaces.”
Furthermore, Dziczek adds, there are scores of new global EV competitors actively seeking to replace the legacy automakers. The “simplicity” of EVs in comparison with ICE vehicles allows these disruptors to compete virtually from scratch with legacy automakers, not only in the car market itself but for the material and labor inputs as well.
Another critical question is whether all the planned battery-plant output will support expected EV production demands. For instance, the United States will require 8 million EV batteries annually by 2030 if its target to make EVs half of all new-vehicle sales is met, with that number rising each year after. As IEA executive director Fatih Birolobserves, “Today, the data shows a looming mismatch between the world’s strengthened climate ambitions and the availability of critical minerals that are essential to realizing those ambitions.”
This mismatch worries automakers. GM, Ford, Tesla, and others have moved to secure batteries through 2025, but it could be tricky after that. Rivian Automotive chief executive RJ Scaringe was recently quoted in the Wall Street Journal as saying that “90 to 95 percent of the (battery) supply chain does not exist,” and that the current semiconductor chip shortage is “a small appetizer to what we are about to feel on battery cells over the next two decades.”
The competition for securing raw materials, along with the increased consumer demand, has caused EV prices to spike. Ford has raised the price of the Lightning $6,000 to $8,500, and CEO Jim Farley bluntly states that in regard to material shortages in the foreseeable future, “I don’t think we should be confident in any other outcomes than an increase in prices.”
One critical area of resource competition is over the limited supply of software and systems engineers with the mechatronics and robotics expertise needed for EVs. Major automakers have moved aggressively to bring more software and systems-engineering expertise on board, rather than have it reside at their suppliers, as they have traditionally done. Automakers feel that if they're not in control of the software, they're not in control of their product.

Volvo’s CEO Jim Rowan stated earlier this year that increasing the computing power in EVs will be harder and more altering of the automotive industry than switching from ICE vehicles to EVs. This means that EV winners and losers will in great part be separated by their “relative strength in their cyberphysical systems engineering,” states Clemson’s Paredis.
Even for the large auto suppliers, the transition to EVs will not be an easy road. For instance, automakers are demanding these suppliers absorb more cost cuts because automakers are finding EVs so expensive to build. Not only do automakers want to bring cutting-edge software expertise in-house, they want greater inside expertise in critical EV supply-chain components, especially batteries.
Automakers, including Tesla, are all scrambling for battery talent, with bidding wars reportedly breaking out to acquire top candidates. With automakers planning to spend more than $13 billion to build at least 13 new EV battery plants in North America within the next five to seven years, experienced management and production line talent will likely be in extremely short supply. Tesla’s Texas Gigafactory needs some 10,000 workers alone, for example. With at least 60 new battery plants planned to be in operation globally by 2030, and scores needed soon afterward, major battery makers are already highlighting their expected skill shortages.

The underlying reason for the worry: Supplying sufficient raw materials to existing and planned battery plants as well as to the manufacturers of other renewable energy sources and military systems—who are competing for the same materials—has several complications to overcome. Among them is the need for more mines to provide the metals required, which have spiked in price as demand has increased. For example, while demand for lithium is growing rapidly, investment in mines has significantly lagged that which has been aimed toward EVs and battery plants. It can take five or more years to get a lithium mine up and going, but operations can only start after it has secured the required permits, a process that itself can take years.
Mining the raw materials, of course, assumes that there is sufficient refining capability to process them, which outside of China, is limited. This is especially true in the US, which according to a Biden Administration special supply chain investigative report, has “limited raw material production capacity and virtually no processing capacity.” Consequently, the report states that the US “exports the limited raw materials produced today to foreign markets.” For example, output from the only nickel mine in the US, the Eagle mine in Minnesota, is sent to Canada for smelting.
“Energy and information are two basic currencies of organic and social systems. A new technology that alters the terms on which one or the other of these is available to a system can work on it the most profound changes.” —Herb Simon
One possible solution is to move away from lithium-ion batteries and nickel-metal hydrides batteries to other battery chemistries such as lithium-iron phosphate, lithium-ion phosphate, lithium-sulfur, lithium-metal, and sodium-ionamong many others, not to mention solid-state batteries, as a way to alleviate some of the material supply and cost problems. Tesla is moving towards the use of lithium-iron phosphate batteries, as is Ford for some of its vehicles. These batteries are cobalt free, which alleviates several sourcing issues.
Another solution may be recycling both EV batteries as well as the waste and rejects from battery manufacturing, which can run between 5 to 10 percent of production. Effective recycling of EV batteries “has the potential to reduce primary demand compared to total demand in 2040, by approximately 25% for lithium, 35% for cobalt and nickel and 55% for copper,” according to a report (pdf) by the University of Sidney’sInstitute for Sustainable Futures.

While investments into creating EV battery recycling facilities have started, there is a looming question of whether there will be enough battery factory scrap and other lithium-ion battery waste for them to remain operational while they wait for sufficient numbers of batteries to make them profitable. Lithium-ion battery pack recycling is very time-consuming and expensive, making mining lithium often cheaper than recycling it, for example. Recycling low or no-cobalt lithium batteries which is the direction many automakers are taking may also make it unprofitable to recycle them.
An additional concern is that EV batteries, once no longer useful for propelling the EV, have years of life left in them. They can be refurbished, rebuilt and reused in EVs, or repurposed into storage devices for homes, businesses or the grid. Whether it will make economic sense to do either at scale versus recycling them, remains to be seen.
Howard Nusbaum, the administrator of the National Salvage Vehicle Reporting Program (NSVRP), succinctly puts it, “There is no recycling, and no EV recycling industry, if there is no economic basis for one.”
In the next article in the series, we will look at whether the grid can handle tens of millions of EVs.
A copyright storm may be brewing for GitHub Copilot
Rina Diane Caballar is a journalist and former software engineer based in Wellington, New Zealand.
GitHub Copilot dubs itself as an “AI pair programmer” for software developers, automatically suggesting code in real time. According to GitHub, Copilot is “powered by Codex, a generative pretrained AI model created by OpenAI” and has been trained on “natural language text and source code from publicly available sources, including code in public repositories on GitHub.”
However, a class-action lawsuit filed against GitHub Copilot, its parent company Microsoft, and OpenAI claims open-source software piracy and violations of open-source licenses. Specifically, the lawsuit states that code generated by Copilot does not include any attribution of the original author of the code, copyright notices, and a copy of the license, which most open-source licenses require.
“The spirit of open source is not just a space where people want to keep it open,” says Sal Kimmich, an open-source developer advocate at Sonatype, machine learning engineer, and open source contributor and maintainer. “We have developed processes in order to keep open source secure, and that requires traceability, observability, and verification. Copilot is obscuring the original provenance of those [code] snippets.”
“I very much hope that what comes out of this lawsuit will be something I can rely on when making decisions about training models in the future.”
—Stella Biderman, EleutherAI
In an attempt to address the issues with open-source licensing, GitHub plans to introduce a new Copilot feature that will “provide a reference for suggestions that resemble public code on GitHub so that you can make a more informed decision about whether and how to use that code,” including “providing attribution where appropriate.” GitHub also has a configurable filter to block suggestions matching public code.

The onus, however, still falls on developers, as GitHub states in Copilot’s terms and conditions: “GitHub does not claim any rights in Suggestions, and you retain ownership of and responsibility for Your Code, including Suggestions you include in Your Code.”
In addition to open-source licensing issues, Copilot raises concerns in terms of the legality of training the system on publicly available code, as well as whether generated code could result in copyright infringement.
Kimmich points out the Google v. Oracle case, wherein “taking the names of methods, but not the functional implementation is OK. You’re replacing the functional content but still keeping some of the template.” In the case of Copilot, it might generate copyrighted code verbatim. (See related tweet below from Tim Davis, computer science professor at Texas A&M University, as an illustration of Copilot generating copyrighted code.)
Kit Walsh, a senior staff attorney at the Electronic Frontier Foundation, argues that training Copilot on public repositories is fair use. “Fair use protects analytical uses of copyrighted work. Copilot is ingesting code and creating associations in its own neural net about what tends to follow and appear in what contexts, and that factual analysis of the underlying works is the kind of fair use that cases involving video game consoles, search engines, and APIs have supported.”

But when it comes to generated code, Walsh says it boils down to “how much [Copilot] is reproducing from any given element of the training data” and if it encompasses creative expression that is copyrightable. “If so, there could be infringement happening,” she says.
The lawsuit against GitHub Copilot is the first of its kind to challenge generative AI. “It’s setting a legal precedent that has implications for other generative tools,” Walsh says. “It’s the type of work that if a person authored [it, they] could qualify for copyright protection, and it could embody someone else’s copyrighted work like snippets of code.”
“If I as an engineer would like to use Copilot, I will need to be able to restrict what it provides me to code that’s attributed to the license.”
—Sal Kimmich, Sonatype
For Stella Biderman, an AI researcher at Booz Allen Hamilton and EleutherAI, the lawsuit is a welcome development. “It’s going to, I hope, provide clarity and guidance as to what is actually legal, which is one of the big issues for those working on open-source AI,” she says. “I very much hope that what comes out of this lawsuit will be something I can rely on when making decisions about training models in the future.”
The open-source community seems divided on the lawsuit and GitHub Copilot itself. For instance, the Software Freedom Conservancy has been vocal about its concerns with Copilot—even calling for a boycott of GitHub—but is cautious about joining the class-action lawsuit. Kimmich says they know of open-source developers taking an ethical stance in choosing not to use Copilot, but also others who are enjoying it. “They’re learning while developing and executing code on the fly.”

Kimmich themself is on a waitlist for Copilot and recognizes the benefits it offers developers. “The neural network behind it is using more than just code to help you—it’s providing much more contextual information,” they said. “It means I as a developer now have an extended intelligence, which is giving me a contextualized recommendation. I think that’s excellent. It’s the most powerful generative intelligence that we’ve had so far for this application.”
Yet unless the open-source licensing issue is solved, Kimmich envisions using GitHub Copilot only for pet projects and exploring new packages. “It stops short of production code because of the licensing issue,” they said. “If I as an engineer would like to use Copilot, I will need to be able to restrict what it provides me to code that’s attributed to the license, or have a license which states that it was codeveloped. If I can’t locate the provenance of the original licenses or the original intellectual property, then I need to be able to know if I want to avoid it.”
Learn about the latest generation high-performance data acquisition boards from Teledyne
In this webinar, we explain the design principles and operation of our fourth-generation digitizers with a focus on the application programming interface (API).

Register now for this free webinar!

Topics covered in this webinar:
Who should attend? Developers that want to learn more about Teledyne SP Device's latest generation data acqusition boards (digitizers).
What attendees will learn? An overview of existing digitizer products and their specifications as well as planned upcoming models. Details about the application programming interface (API) and how to operate these devices.

Presenter: Thomas Elter, Senior Field Applications Engineer


Note that any programming tips and code writing requires some knowledge of computer programming. Please, be careful if you do not know what you are doing…

Post expires at 5:36am on Wednesday February 22nd, 2023

Leave a Reply

Next Post

Bearer Launches Data-First Security Solution - GlobeNewswire

Tue Nov 22 , 2022
November 15, 2022 04:00 ET | Source: Bearer Bearer San Francisco, California, UNITED STATES SAN FRANCISCO, Nov. 15, 2022 (GLOBE NEWSWIRE) — Bearer, the data-first security software company, today announced the general availability of the Bearer Data Security Platform. Based on extensive interviews with more than 130 enterprise CISOs at […]
%d bloggers like this: