The Tech-Bro Presidency

Trump’s second term will be remembered, largely with horror, for many things, but one of the most significant has been the outsized role played by large US tech companies and their founders. Whether at his inauguration, at state dinners or at parties, the tech bros have been well represented. Elon Musk, during his time running DOGE, the highly ineffective cost-cutting operation, acted almost as Trump’s political equal.

The fawning praise they lavish on Trump, who they must surely realise is a complete idiot when it comes to tech, and the hundreds of millions of dollars they have spent on Trump’s ballroom and on political donations have ruined their reputations with much of the electorate in America and abroad. Yet it is technology stocks that have largely powered the stock market rally of the past two years. Investment in AI is sucking in vast amounts of domestic and international funding, to the detriment of almost every other sector of the economy. And the vast resources poured into AI have produced huge leaps in performance over the past two years. AI models have made major breakthroughs in mathematics and biology. They are no longer interesting and sometimes laughably wrong chatbots, but genuinely useful tools for a whole range of jobs and activities.

In the past few weeks, far more worrying issues surrounding the US tech giants have taken centre stage and could well form the basis of a plan of action and legislation for the new Congress, which will be seated after the upcoming midterms. The first issue is that the White House and Elon Musk between them agreed to limit Ukraine’s access to Starlink and to Starshield, its military counterpart. They did this by cutting off connections to drones flying faster than 100 miles per hour. Above that speed the link failed and the drones were effectively blind. Because Ukraine uses these systems to intercept incoming Russian missiles and drones, any break in the connection leads directly to the deaths of innocent Ukrainians. When President Zelensky posted publicly about this, the White House and Musk each blamed the other, although further reporting indicates that Musk initiated the restriction. It amounted to direct help to Russia, and came only days after a minor but sanctions-breaking diesel deal that Trump had agreed with Putin.

The second worrying development concerns AI. In early September, Jacob Coxon, a young AI researcher who had worked at both OpenAI and Anthropic, went public with his concerns about the two firms’ race towards self-improving models, warning that AI could wipe out humanity by the end of the decade. He cited the already well-known Hugging Face attack, in which a non-public OpenAI model deliberately deceived its human controllers, broke out of its “sandbox” and attacked another company’s infrastructure. Rather than denying Coxon’s concerns, many senior figures in the industry echoed them. Since then, other AI-driven hacking concerns have come to the fore: a vulnerability was exposed in PHP, an old but widely used programming language, and numerous government-linked websites were taken down until a patch was installed.

October ended with news of possibly the biggest “bank” heist in history. Binance admitted that hackers, whom others attributed to North Korea’s TraderTraitor unit, had stolen $6.3 billion of crypto assets. The company froze $3 billion of those assets and promised to make good all customers’ losses. Whether it can do so remains to be seen. Crypto has no industry-wide deposit insurance, so the ability to make clients whole depends on each individual company. The crypto industry is somewhat distinct from the traditional US tech giants, but it is a technology Trump has embraced, having discovered that it is a marvellous way to facilitate corruption and untraceable transactions. A sceptic could argue that this is crypto’s primary use: an asset ideally suited to laundering money and criminal proceeds.

The risks are already here

Each of these examples illustrates, in different ways and to different degrees, a technology sector that is largely out of control or, more precisely, out of state control. For more than 40 years, the dominant narrative in the US and much of the developed world has been that government control is bad, and that private companies and markets are the best way to deliver products and services. That ideology now appears to be reaching its limits within a democracy.

National security and defence are the first concern of any government. A state that cannot defend and maintain its borders can hardly be called a state. Successive Republican and Democratic administrations have allowed a single company, SpaceX, not only a near-monopoly on launching rockets into orbit but also the deployment of a highly effective low-Earth-orbit satellite communications system, Starlink, which has global reach and has become a crucial part of several countries’ national defence. Put simply, such capabilities cannot be under the control of a single company, still less a single person. SpaceX’s highly unusual shareholding structure gives Musk voting power out of all proportion to his economic stake. He alone controls the company and, by extension, the technologies it provides. Even if Musk were not such an erratic individual, neither the US government nor any other can allow one person such power over critical systems.

You would like to think that AI’s potential threat to humanity alone would push Trump to consider regulating the industry, yet he has ruled that out. The industry cannot regulate itself, because no company can stop developing models for fear of losing out to a competitor, whether in the US or in China. And the risks of AI are already present. One area where AI has proved extremely effective is hacking computer systems. It is entirely conceivable that AI-backed hacking tools could disable large parts of the internet or national utility systems. A recent hack (admittedly not attributed to AI at this time) in France saw refrigeration units turned up to 50°C, spoiling all the food inside. A few failed fridges are an inconvenience; the same thing happening across entire supply chains, or across every internet-enabled device in our homes, would be a national emergency. There has long been concern that Chinese IoT modules embedded in a whole range of products could allow Chinese spying and remote control. But AI’s capabilities potentially allow any number of actors, state or otherwise, to cause mayhem.

The risks and influence of crypto are somewhat different. But when so much value has been attached to what remains a solution in search of a problem, the knock-on effects on the traditional financial system become more significant. As more and more institutions have bought into cryptocurrencies and exchanges have provided crypto-linked products over the past decade, the linkages have become substantial. Falling crypto prices are not the concern of the Federal Reserve, other central banks or regulators such as the SEC, but the health and integrity of financial systems most certainly are. If a crypto collapse spills over into traditional financial markets, then that is a problem. A major selling point of crypto was its hard-coded software constraints, but what could AI hacking do to undermine the code that underpins the tokens?

As if on cue, the Nasdaq 100 fell 3.5% on the last trading day of October and was briefly down 10% from the high reached only on 6 October. The fall was driven primarily by the huge AI build-out costs and by doubts about whether future revenue potential could justify the expenditure.

What does China do?

The risks from the tech industry are real and present today. They will increase in the future, and the worst-case scenarios are not yet present, but the current situation demands attention and greater oversight from governments around the world. Since AI development is so often presented as an all-or-nothing race between the US and China, it is worth understanding how China has approached all the sectors now causing such concern in the West.

In each of these three cases China has taken a very different approach. The primary goal of the Communist Party is to remain in power. That has driven policies that try to reduce dependence on overseas systems and also limit non-Party power centres domestically. Starting in 2000, China began to develop a rival to the US-operated GPS system. After 20 years of development, China has a global alternative called BeiDou. Its performance does not match that of GPS, and it is used primarily by PRC entities and nationals, but Chinese leaders realised they did not want to be dependent on US systems for something as basic as geolocation. It is no surprise, then, that the Chinese authorities would not allow a private company, or a private individual, to control an orbital launch system and a private satellite communications system on the scale of Starlink. Private companies do operate in this sphere in China, but the lines between private and state companies are often very blurry. Private companies simply do not have the same room for manoeuvre as they do in the US, nor are private property rights held in such reverence. China has a long history of jailing and detaining private-sector founders when the state has considered their businesses a threat to the state-owned sector. Jack Ma, the founder of Alibaba, is probably the most famous such billionaire to have been detained, but he is not the only one.

AI presents a more complicated picture. The state is very supportive of AI adoption across the economy and sees it as a key tool to maintain growth levels and increase productivity. The surprising success of a number of private companies that have developed open-source models nearly on a par with some of the top US models has been a source of pride for the Chinese and a catalyst for more work in the field. China does suffer from a lack of compute capacity and from funding limitations, as many companies have relied on equity financing to grow their businesses, unlike in the US, where retained earnings and debt have been the key drivers. US export controls mean that China does not have easy access to the latest AI chips from Nvidia, and that has allowed the US to maintain a lead over its Chinese counterparts. But the risks AI poses affect the Chinese economy and the Communist Party just as much as they affect the West. The Chinese have floated ideas on regulating AI. There has been a general distrust that China would follow such rules, but again the state is not afraid to act swiftly and harshly if it feels threatened.

For many years much of bitcoin mining was conducted in China due to the plentiful supply of cheap power. That has now ceased, and China has basically banned cryptocurrencies and stated that they are not legal tender. The use of cryptocurrencies to facilitate money laundering and capital flight has been a key driver in this regard. The capital controls that China has in place are not perfect, and there is leakage, but they are seen as essential for the Party to maintain economic control.

It is nonsensical to think that Chinese political control can easily be layered onto open democratic societies. But in the desire to ensure its primacy, the Party has realised that billionaire power centres and control of technology can pose a real threat, to the Party and to the broader society. It is hardly a winning political slogan in the US, or anywhere else, to champion the idea that our regulation should be more like China’s, but when we are faced with the threats of technology in the hands of a limited number of billionaire founders, it is really time to take action. Brad Smith, the president of Microsoft, recently said that the milk industry is more heavily regulated than AI, and this should highlight how exposed we all are to risks that are present today. Congress, following the upcoming midterm elections, needs to start building a new regulatory framework in which some technologies come directly under the oversight and control of the government and elected representatives. No one would consider the building and sale of nuclear warheads just another technology. Private firms build them, but neither their chief executives nor their boards decide how they are used or who they are sold to; that sits with the state. Yet the aerospace and communications technologies of SpaceX and the frontier models of AI can be just as significant. Chinese state control over companies and the economy is seldom worth following, but perhaps there are some lessons to be learnt when national security has become beholden to those whose primary concern is not the security of the nation.