“If there is to be a war [between China and the United States], it will not arise from Western outrage at human rights violations in Xinjiang, or Chinese outrage at Western outrage, or cyber-crime, or technology theft, or currency manipulation, or security crackdowns in Hong Kong, or indignities visited upon the Filipinos or the Vietnamese or the Australians. It will arise from acute economic pain, inflicted on China by actions of the United States to deprive them of the most essential physical resource of the 21st century: semiconductors.” –George Calhoun (“War With China?”)
“The ‘new oil’ in the tech world is semiconductors.”–Kenneth Rapoza
Last week, I was a guest on a nationwide podcast (The McFiles Channel, Aug. 10, 2023), and the moderator asked me what country posed the greatest strategic threat to the United States in today’s world. Without hesitation I replied that it was President Xi Jinping’s China, controlled by his Chinese Communist Party (CCP) minions. However, I quickly qualified my answer, “we are prone to forget that Beijing’s rulers have their own long list of problems.” Among those, I quickly listed a few: a demographic bulge due to the CCP’s disastrous, decades-long “One Child Policy;” a slower-than-expected economic bounce back from the pandemic; restless and underemployed youth; and an economy close to stalling out (slowing retail sales, drying up foreign investment, loss of consumer confidence, and collapse of mega-property developers).
In my answer, however, I forgot to mention what I regard as Beijing’s greatest problem in its strategic rivalry with the US; the struggle to compete with the West in the semiconductor race. In recent months, this gap has been clearly evidenced by developments in the ChatGPT arena.
What is ChatGPT?
It is a viral chatbox—search bot—that has generated feverish excitement in today’s internet world, largely because of the hope it can reinvent search engines. ChatGPT-3 (Chat Generative pre-Trained Transformer) is a third generation, autoregressive language model that uses deep learning to produce human-like text. ChatGPT can provide complex and sophisticated answers to questions by synthesizing information found in billions of words scraped from the web and other sources to train its algorithms. The chatbox is built on top of OpenAI’s GPT-3 family of large language models. Moreover, the chatbox has become a huge success: the AI conversational tool amassed 100 million users just two months after it launched.
Not to mention a flurry of articles predicting that ChatGPT is the first step in an inevitable techno-trajectory leading to the Singularity—that point where machine intelligence surpasses human intelligence and becomes uncontrollable—to be followed by the ultimate destruction of humankind. Perhaps this may be a stretch too far for some of you, but this week, for example, comes word that Google and Google-owned DeepMind AI are working with the latest iteration of generative AI to build a “life advice” tool.
Just what we need, right?
At any rate, it is a vastly different technological world out there.
These advances in large language tools, the engines fueling today’s advances in generative AI, depend on increasingly powerful semiconductor chips produced by companies like Nvidia. I like to talk about Nvidia Corporation because I have—sort of—a personal connection with the company’s President and CEO Jen Hsun “Jensen” Huang. Jensen Huang is a 60-year-old Taiwanese-born American billionaire businessman and electrical engineer who co-founded the company in 1993, (during a meeting at a roadside Denny’s diner in East San Jose). When he was 10 years old, in the early 1970’s, Jensen lived in the boy’s dormitory (with his brother) at tiny Oneida Baptist Institute (OBI) in the remote town of Oneida, in mountainous eastern Kentucky, while attending a local elementary school. I stayed in the same dorm when I taught at OBI a few years later. (I told you it was a “sort of” connection). In recent years, Jensen has generously provided funds for a beautiful new boy’s dorm on the OBI campus.
In the years since, Jensen’s brainchild Nvidia has become the gold standard of the semiconductor world and is a dominant supplier of AI-related hardware and software: the company’s graphics processing units (GPUs), application programming interface (APIs) for data science and high-performance computers, and system-on-a-chip units (SoCs) are in high global demand. Nvidia’s H100 chips, at $40,000 each, are particularly sought after by big tech firms and start-up venture capitalists seeking to build generative AI-driven search engines. In many cases, customers wait more than six months for the prized chips.
Chinese high-tech firms have joined the Nvidia frenzy, with Baidu, ByteDance, Tencent and Alibaba making orders worth one billion dollars to acquire about 100,000 A800 processors and a further four billion dollars’ worth of GPUs to be delivered next year. (Due to US export restrictions, Chinese companies are only allowed to buy A800s which have slower transfer rates than A100s).
As the Nvidia story indicates, China is constantly playing a game of catch-up when it comes to today’s most advanced semiconductors. The Chinese high-tech corporation Baidu, for example, belatedly entered the generative artificial intelligence fray with an announcement in early February 2023, revealing concrete plans to launch a chatbox to rival ChatGPT. The announcement prompted a jump in the premarket trading of 15%. Baidu’s AI chatbox was called “ERNIE Bot,” for English users, with plans to complete internal testing by March 2023, before making it available to the public. Baidu, the Chinese company which most resembles Google, has a core business in online search and advertising. Another Chinese tech giant, Alibaba, announced its own AI-powered chatbot. Other Chinese companies have since followed suit.
Nevertheless, the lack of latest generation semiconductors (chips and microchips) in China prevents the country from competing on equal terms with the West in areas such as large-scale training models. And the US appears determined to keep this lead. In early August 2023, for example, the Biden Administration announced new proposed regulations aimed at curbing US investments in key Chinese technology sectors—namely semiconductors, microelectronics, quantum information technologies, and AI systems. Such regulations are hardly new; beginning with former President Trump—and continued thus far by the Biden administration—there has been a steady tightening of the semiconductor noose, starting with Huawei and ZTE. Why? A long history of unfair Chinese trade policies, technology theft, increasing national security risks and geopolitical rivalry.
In this vein, as early as late September 2021, the Biden White House hosted a meeting of leading American and foreign business leaders to discuss the global shortage of semiconductors and the importance of the semiconductor industry to the US economy, our country’s national security interests and our future role as the leader of global technology. The three-decade domination of US companies in the semiconductor industry—that’s why we call it “Silicon Valley” after all—is slipping: in 1990 we controlled 37 percent of global semiconductor manufacturing, but last year (2020) only 12 percent, according to statistics provided by the Semiconductor Industry Association. Much of the ground has been lost to companies in Taiwan, Japan, South Korea and China. In the auto industry, for example, the number of semiconductors in a modern vehicle can exceed 1,000.
But its more than automobiles and EVs. In today’s world, microchips power just about everything we use, and semiconductors are essential to everything from cellphones, to sensors, to 5G, to self-driving vehicles, to the IoT, to satellites, as well as to drones and other modern weapons of warfare. That’s why China’s five-year plan for building the chip industry gets the same strategic attention Beijing once gave to building its atomic bomb.
And that brings us to the crux of this article.
Let’s assume for a second the unimaginable, but for some the inevitable: a future war between the United States and China. The question of how such a conflict would begin has been the subject of several recent books, including two that have become popular with Pentagon planners (as I described in my recent missive on AI and the coming war). The two books–2034: A Novel of the Next World War, and Ghost Fleet: A Novel of the Next World War by P. W. Singer and August Cole (2015)—are both techno-thrillers portraying a future war pitting a declining United States against a rising China. The two books are similar in portraying the US conventional warfare advantage being neutralized by Chinese technological breakthroughs (in 2034 a new, but never defined, algorithmic-cyber capability enables Beijing’s leaders to blackout portions of the US, intercept naval encrypted communications, and shut down defensive weapons’ systems; in Ghost Fleet a secret space weapon enables a new Chinese military leadership to launch a preemptive strike in the Pacific after the US Defense Department’s constellation of satellites over Asia are blinded).
Why do I mention the books? Both assume what has become conventional wisdom among military planners around the world—that China will overtake the United States over the near-term on all major technological fronts, leaving the US behind as Beijing’s leaders move toward global domination. With this comes a corollary assumption: China’s top-down authoritarianism gives it a future advantage in earmarking large amounts of money for targeted technology areas such as artificial intelligence, robotics, quantum computers, space exploration, and subatomic research.
What is the problem with this logic?
It ignores one basic fact: China’s future technological trajectory has a very real Achilles heel.
Simply put, The West still has a stranglehold over top-end global semiconductor (chips and microchips) design, fabrication (“fab”) and production. As Chinese Vice Premier Liu He told an assembly of China’s top scientists in May 2021: “For our country, this technology [semiconductors] is not just for growth. It’s a matter of survival.”
In an excellent recent article, George Calhoun asserts this weakness will become the most likely cause of a possible future US-Sino war. How does he arrive at this conclusion? Calhoun suggests that China in 2023 finds itself in much the same bind that Imperial Japanese leaders found themselves in 1941.
But since very few people study history anymore, his reasoning may seem far-fetched for some.
Let’s go back to the 1940s. Pressure on Japan’s totalitarian military leaders by then-US President Franklin D. Roosevelt (in the form of an oil embargo—then the most important strategic commodity) prevented Tokyo’s leaders from achieving their geostrategic objectives in Asia. Indeed, after the war, Emperor Hirohito was quoted by one of his wartime aides as saying that Japan went to war with the United States because of oil—and lost to the United States—because of oil. One can argue, perhaps, that Calhoun’s exclusive focus on the economic aspects of this one strategic commodity may be an oversimplification of a complex strategic military decision-making process. But it makes for an interesting comparison.
Calhoun also notes that Japanese efforts to break the stranglehold by developing a homegrown synthetic oil industry—converting coal to oil—failed (before FDR’s embargo, Japan relied on the US for 88% of its oil consumption). That left Tokyo’s leaders with the option of military expansionism to control oil and other raw materials in formerly French possessions in Indochina, the Dutch East Indies (Indonesia today) and coastal China.
The point: in 1941, Japan was in a position of acute strategic vulnerability.
Calhoun argues that China is in a similar bind today and this is the most likely path to future war. Today, semiconductors occupy the place oil held some eight decades ago. In 2020, China spent $350 billion on chips based largely on Western technology and, again, China’s modern tech economy runs on silicon. (This amount, by the way, is more than China spent on imported oil.) As Calhoun observes, today China buys 60% of the world’s chip production, 90% of which is sourced from outside China or produced domestically by foreign manufacturers such as Intel and Nvidia.
Chinese leaders know the problem is growing and the gap regarding this strategic commodity is widening. There is also a growing qualitative gap, especially in the area of the “fabless” IC sector, which drives chip design, the innovative sector pushing forward the digital economy. As a result, many of China’s semiconductor fabrication facilities lag four to five technology generations behind the leaders in the field.
At the same time, despite China’s much-ballyhooed central decision-making—and the seeming success of its “moonshot” approach by using massive direct public investments in other technology areas—it has been unable to compete in this strategic commodity. Beijing’s leaders have repeatedly announced new policy initiatives to catch up in the semiconductor race. China’s more than 20-year track record of chipmaking has been littered with “unfulfilled promises, stillborn projects and government waste.” As one expert notes, the semiconductor industry is very market oriented and isn’t like launching a space station or building a giant atom smasher.
How serious is this getting? In February 2021, key executives from leading American companies such as Intel, Nvidia, Broadcom, Western Digital, IBM and Qualcom met with President Biden and urged him to invest in the future of semiconductors. In their collective view, it was a technology race we couldn’t afford to lose. In Bloomington, MN, the new SkyWater Technology Foundry was built—the first semiconductor “fab” to open in the US in a generation.The US Senate has earmarked $52 billion—under the Chips for America Act—to encourage domestic semiconductor investment. Two years ago, the leaders of the United States, Japan, India and Australia agreed to take steps to build secure semiconductor supply chains when they met in Washington, according to the Nikkei Business Daily. The agreement clearly targets China’s most obvious weakness. Squeezing the semiconductor supply chain is the latest wrinkle in efforts to decouple the supply chains linking the world’s two biggest economies. Indeed, technology-related foreign direct investment (FDI) between China and the US dropped by 96 per cent from 2016 to 2020, posing a significant challenge to Beijing which has invested heavily in technology and supply chain independence, according to Boston-headquartered consultancy Bain & Company.
At the same time, the unique nature of profit-driven private business models in the United States and elsewhere in the West will make it difficult to maintain a united front on the matter of controlling semiconductor sales to China. Recently, for example, representatives of four key government security entities could not agree on whether to place a Chinese-linked telecommunications company on the Commerce Department’s entity list, which bars exports of US technology to the sanctioned firm without a department license. The company, Honor, is a recent spin-off of sanctioned Chinese firm Huawei. Soon after Huawei sold Honor, several US companies began striking deals to sell chips to the new company. Qualcomm, based in California, is selling Honor a high-tech chip that the Chinese company is using in its newly launched 5G smartphones. Qualcomm overall is supplying technology for three new Honor phones, Qualcomm’s chief executive said during a July conference call with investors. The Department of Treasury, for example, reportedly declined to put several PRC technical surveillance firms on the SDN list.
All of this reminds me of a famous quote by Russian Bolshevik leader Vladimir Lenin (and one which Communist leaders in China and elsewhere are fond of quoting): “When the time comes to hang the Capitalists, the Capitalists themselves will sell the rope for that.”
Anything for a profit.
Even semiconductors.
Moreover, scientists are also getting edgy. A recent article, for example, warned that Washington’s efforts to cut semiconductor and research ties with Beijing over security concerns threatens “American progress in critical areas,” and lament that the landmark agreement to cooperate on science and technology—signed in 1979 and routinely renewed—may expire later this month.
So, what is the future likely to hold? It doesn’t take a crystal ball to predict that pressure on China’s leaders will steadily mount due to this acute strategic vulnerability. Just as important, control of the design and production of increasingly sophisticated microchips is one of the few remaining technology cards remaining in the West’s playing hand these days.
Likewise, you need no gift of prophecy to see that Beijing’s leaders will do everything possible to close the semiconductor gap.
Calhoun suggests the tensions growing out of the West’s effort to capitalize on this techno-vulnerability may lead to future war. Such a trajectory, of course, will not happen in a vacuum. There will be, no doubt, several opportunities to avert open conflict. Moreover, even now a myriad of other factors are swirling in the background that will influence any future decision to go to war. Chief among these are the nature and intensity of increasing trade friction with the US and the growing technological and geopolitical rivalry between the two global superpowers. Then there is the ongoing strategic dance with Taiwan highlighted by Beijing’s steady military buildup and overflights in the Strait. (Perhaps it is worth noting that Taiwan is home to several of the world’s most robust and modern semiconductor companies.) Some argue that a future invasion of Taiwan moved a step closer by the recent removal of a threat on Beijing’s flank by its strong-arm tactics in Hong Kong.
Perhaps most important of all, Beijing’s quest to achieve semiconductor self-sufficiency has become entangled in President Xi Jinping’s domestic political agenda after he secured an unprecedented third term as paramount leader in 2022. One highly publicized prong of Xi’s power grab is his “common prosperity” campaign that includes a crackdown on China’s private sector high-tech titans. According to at least one prominent China watcher, Xi’s efforts are being driven by three basic factors: ideology, demographics and decoupling from the West. The bottom line: Xi can’t be sure of the loyalty of China’s high tech titans—who’ve become rich and famous in their own rights—because they sit on large stores of data (which may be eventually used against him). At the same time, Xi seems to be realigning the country’s tech sector to favor state-led developments in the areas he cares about, such as creating breakthrough technologies in AI and semiconductors.
But it is a risky proposition for Xi. In 2020, China’s private companies accounted for 76.6% of China’s overall R&D expenditures (followed by government-run research institutions and universities—both hit hard by the pandemic). So, while Xi’s moves may be cementing his control among the Party cadres, he is also crippling the very entities within the technology sector most capable of breakthrough innovations in such critical areas as semiconductors.
So far, at least, China’s ChatGPT contenders have promised to develop generative AI training systems that toe the Party line.
Finally, is there a technological breakthrough on the horizon that will render the West’s semiconductor stranglehold a moot matter? Tomorrow’s world will feature exponentially increasing costs of design and verification costs versus semiconductor scaling, FinFETs, 2.5D/3D assembly, chiplets, interposers, and encryption PQC algorithms. Moreover, future replacements for silicon are already showing some promise: such as graphene (a 2D conductive material—microchips using graphene can sustain many more transistors than silicon), gallium nitrate (better performance in high energy and high heat applications than silicon, but cumbersome and expensive to produce), and perovskites. And these developments do not include the possibility of game-changing spillover discoveries in technologies such as biogenetics, superconductors or post-digital quantum breakthroughs.
Perhaps it is worth noting that when ChatGPT burst suddenly on the scene in November 2021, it surprised almost everyone.
Are we on the cusp of another breakthrough that will make today’s generative AI, large language training models and ChatGPTs (and their future technological offspring) pale by comparison? Will a growing generation of Chinese scientists, computer technicians, AI specialists and entrepreneurs spearhead such advances? I formerly made the case in my college history classes that we are long overdue for such a “Black Swan” breakthrough.
Then the question becomes: would a technological replacement for semiconductors make a future global war more or less likely?
For a footnoted version of this article, please contact Braeburn Road Books via the contact page.