| |
The scramble for chips is just the beginning.
By: Ronan Wordsworth
In January, China
took the tech world by surprise when it unveiled DeepSeek, an
artificial intelligence company that has proved to be as competitive as
any other but at a much lower cost. The event was a wake-up call to
policy officials of all stripes in Washington, who understand that AI
will soon affect, to some degree or another, all aspects of political
life.
Dominance in this
field requires massive computing power, a big enough energy supply to
power the vast data centers behind AI, and, finally, the human resources
needed to develop new and innovative iterations. It’s little wonder,
then, that governments are racing to finance the very infrastructure on
which AI relies. Within days of the beginning of U.S. President Donald
Trump’s second term, his administration announced the creation of an
enterprise known as Stargate, which plans to invest $500 billion in
private sector AI infrastructure. In February, ahead of a multinational
AI summit in Paris, France announced investments totaling around $112
billion in AI infrastructure and development. This included $20 billion
from Canadian investment firms for new AI projects, up to $50 billion
from the United Arab Emirates for new data centres, and large
investments in Mistral, a European competitor to OpenAI and DeepSeek.
China already has a surplus of data centers, having constructed hundreds
in the western provinces, while large tech firms such as ByteDance are
investing billions in additional facilities.
Government
efforts to take the lead in this nascent geostrategic environment go
beyond investment alone. The Biden administration, for example, imposed
strict chip export controls in 2022 to prevent China from obtaining the
advanced chips presumably needed to run the top AI models. Days before
departing the White House, President Joe Biden added more regulations to
control exports of the most powerful chips, following further
restrictions on chipmaking equipment going to China. This would limit
Beijing’s ability to manufacture its own advanced chips and
semiconductors domestically, or so the thinking went.
Beijing responded
by imposing its own restrictions in a field in which it holds a
distinct advantage: critical minerals. China is responsible for almost
70 percent of the mining of rare earth elements. And with its large
domestic reserves and long-established mining concessions in Africa, it
is responsible for more than 90 percent of all rare earth processing.
For some minerals, China has a near monopoly. The U.S. produces just 12
percent of global supply and relies on Chinese machinery for extraction.
And so, after Washington introduced the chip bans, Beijing placed
restrictions on rare earth extraction and separation technologies.
Later, in 2024, it banned the export of some rare earths required for
the manufacture of semiconductors.
Washington
understands its vulnerabilities in the rare earth supply chain. It’s one
of the reasons Trump has tied negotiations over the Russia-Ukraine war
to mineral rights in Ukraine, and why Biden tried to shore up domestic
extraction. It’s also why Washington is leveraging security guarantees
for Taiwan – the world’s foremost semiconductor superpower – into
investment for new manufacturing centers in the U.S.
Put simply, AI
has the power to transform geopolitics. Traditionally, geopolitical
power is derived from the domination of physical space – air, land and
sea, with space emerging as a fourth domain during the Cold War. (The
cyber realm came not long after the war’s end.) And, traditionally,
geopolitical power is wielded by economic, political and martial means.
AI is unique because not only will it spur an evolution in its own
domain (cyber), but it will also affect the others (economics, politics
and war). All of them will experience upheavals over the next decade as a
direct result of the emergence of faster, better and more capable AI
models. These models also have the potential to disrupt the global
system, widen the gap between the haves and the have-nots, and endow
certain countries with insurmountable leads in the cyber realm.
From an economic
perspective, the benefits of advanced AI are many. It can be used to
optimize supply chains, better predict macroeconomic trends, detect
fraud in banking and financial services, improve financial transactions
and increase workplace productivity. It can assist in the optimization
of energy resources, agricultural production, resource extraction,
disaster relief and management. Crucially, it can also revolutionize
labor in certain industries. Take truck drivers as an example. In the
U.S alone, some 3.6 million people would potentially be out of work if
AI-powered self-driving becomes a reality. Similar transformations are
likely to occur in countless other industries, making competitiveness a
matter of efficiency in AI models rather than a matter of human
productivity. There is a real risk this shift in the labor market will
create political instability.
Meanwhile, some
of the political uses of AI are already here. Russian disinformation
campaigns in Africa, for example, once required thousands of employees
to produce content to flood social media and thus alter the news
environment in target regions. AI-generated and AI-altered images and
videos amplify the effects further, and with the ability to generate
targeted content in hundreds of languages or to specific subsets of
individuals, the prospect of disrupting political movements is already
apparent. Higher powered AI will be even more adept at creating
deepfakes, synthetic media and automated propaganda, making
disinformation campaigns more effective and harder to detect. This is
likely to intensify as a mainstay of hybrid warfare – undermining
adversaries' internal stability through targeted campaigns. The tools
currently available to counter this threat are vastly insufficient.
The risk of
public unrest aside, governments will be able to use AI to expand mass
surveillance, social control and digital repression. Through its ability
to process vast amounts of data, AI will make it easier for governments
to crack down on subversion and dissent. Even for democratic regimes,
the temptation to, say, monitor and combat crime may be too strong to
ignore.
But perhaps the
most important – and most foreboding – use of AI will be its military
applications. In the same way that previous generations of
transformative technologies changed the battlefield, the next generation
will aid and abet combat operations. The number of ways AI will do so
is nearly uncountable, but some examples are instructive: coordinated
drone swarms, automated battlefield strategies and communication
disruption; surveillance and reconnaissance; early warning and evasion
systems; strategic decision-making; increasingly sophisticated
cyberattacks capable of disrupting financial markets, critical
infrastructure and military operations; and stealth detection.
It’s little
wonder, then, that governments are scrambling not only for the requisite
materials needed to power AI but also for the human talent to develop
and operate it. In China, tech companies that had been left out in the
cold by the government’s regulatory crackdown are now becoming major
players. It’s unclear how much money DeepSeek, for example, received
from the government, but Beijing has been and will continue to be
involved in strategic industries such as these. Beijing’s public show of
support serves two purposes. It reassures tech companies that their
investments have not been made in vain, and it sends a message to the
rest of the world that China is not only serious about the AI race but
also uniquely able to lead it.
Washington has
certainly taken notice. Some analysts have called the DeepSeek
announcement this generation's Sputnik moment, which triggered the space
race with the Soviet Union during the Cold War. Whatever lead the U.S.
may have had in the current race is gone. Expect Washington to double
down on export controls covering any remaining aspects of high-end
computing power as it tries to reestablish its position. For its part,
China is likely to intensify the tech war, offering greater support for
domestic players and further restricting rare earths.
It’s unclear
whether AI will continue to be a two-team race. For most, the cost of
admission will be prohibitively expensive, so the disparity between the
haves and the have-nots will only widen. European players are looking to
catch up; the United Kingdom and France are trying to invest in their
own domestic capabilities to avoid being left behind.
The emergence of
DeepSeek was, more than anything, a wake-up call, one that sent a clear
and inscrutable message: that AI is set to revolutionize all forms of
geopolitical power. And just as the space race defined international
relations in years past, the race for more powerful AI may well define
the years to come. |