Perhaps US efforts to cut off China’s access to advanced semiconductors will be more successful than analogous restrictions on tech exports to France in the 1960s. But we now have at least one data point – DeepSeek – that suggests otherwise.
BERKELEY – Starting in October 2022, the late, lamented (by some) administration of President Joe Biden implemented restrictions on US exports of advanced semiconductors to China. A classic “dual use” technology, these chips power generative AI and supercomputers used in weapons systems, cyberattacks, and surveillance. Equally, they can enhance the ability of companies to compete internationally – in this case the ability of China’s high-tech firms to compete with their American rivals.
In its final months, the Biden administration doubled down on these restrictions, adding high-bandwidth memory chips and chipmaking tools to the list of prohibited items. Donald Trump’s administration, dominated by China hawks, is poised to ramp up these measures still further.
China has retaliated with export controls on rare earths and related materials used by American producers. But the more serious threat is that China will accelerate the development of its own capacity to design chips and build chipmaking equipment. If China closes the technology gap faster than it otherwise would have, then US export controls will have proven ineffectual or even counterproductive.
It’s hard to predict the success or failure of China’s efforts. But it may be informative to look to the past, where some of the most relevant history comes from, of all places, France in the 1960s.
French politicians and businessmen at the time perceived an economic threat from the United States. This was the era when the journalist and politician Jean-Jacques Servan-Schreiber warned that France could become an economic vassal of the US. Servan-Schreiber was worried about the invasion of Europe by US multinationals, with General Electric’s 1964 acquisition of the French computer company Machines Bull a prime example.
In addition, French President Charles de Gaulle objected to US dominance of NATO and refused to put France’s forces under joint command with other alliance members. Ultimately, he withdrew France from NATO’s integrated command.
It is not too late to watch our AI Action Summit event.
Click the link to watch world leaders, tech experts, and other distinguished speakers – including Justin Trudeau, Petr Pavel, Daron Acemoglu, Reid Hoffman, Marianna Mazzucato, James Manyiga, Audrey Tang, Sylvain Duranton, Celina Lee, Patrick Pouyanné, and others – discuss some of the most important questions raised by the rise of artificial intelligence.
Watch Now
This was the context of America’s decision in 1964 to deny licenses for the export of advanced IBM and Control Data Corporation (CDC) computers to the French Atomic Energy Commission. These giant mainframe computers could have been used for calculations enabling France to build a hydrogen bomb, which only the US and the Soviet Union possessed at the time. A thermonuclear capacity would have allowed France to pursue an independent defense policy in disregard of US wishes. Reverse engineering IBM and CDC machines also might have helped reinvigorate France’s commercial computer industry and overcome the sad dismantling of Bull.
Such was the background of the delightfully named Plan Calcul launched by the French government in 1966. The government envisaged investing more than one billion francs – about €1.5 billion today – in the French computer industry in the program’s first five years. Half would be provided by the state in the form of research subsidies for private firms, and half by the firms themselves.
To help the companies finance their share, the state extended subsidized loans and guarantees. An agency attached to the prime minister’s office coordinated the plan, which entailed merging two French companies to create a national champion, Compagnie internationale pour l’informatique (CII).
Plan Calcul was championed by de Gaulle himself, who saw a world-class military and a world-class economy as equally important for securing France’s status as a global power – for restoring its “grandeur.” One cannot but be reminded of the aspirations of Chinese President Xi Jinping and his personal role in efforts to advance China’s high-tech supremacy.
Ultimately, Plan Calcul failed to make France a leader in high-speed computers. The government’s technocrats, instead of working harmoniously with the private sector, micromanaged CII’s affairs, keeping for themselves the roles of bankers, entrepreneurs, supervisors, and clients. They focused not on commercial markets, where CII would have felt the chill winds of competition, including from IBM, but on government procurement contracts, where CII could stay fat and lazy. Rather than supporting the private sector, the government’s interventions impeded its progress.
Moreover, when forming their national champion, the technocrats excluded Bull, the firm with the most expertise, since it was now in foreign hands. Rather than complementing one another, national security and economic imperatives worked at odds with each other.
But if Plan Calcul was a damp squib, this was not because the US export-control regime was effective. Firms like IBM pushed back against the loss of business and were able to obtain exemptions. Information flowed freely between IBM’s plants in the US and France, where, as the CIA observed, it could be accessed by local scientists. When the latest computers were unavailable, the French were able to substitute them with slightly older versions.
Thus, France acquired the hydrogen bomb in August 1968, successfully testing a device in French Polynesia. In the end, US export controls only briefly delayed the geostrategically inevitable. Perhaps America’s latest resort to export controls will be more effective. But we now have at least one data point – DeepSeek – that suggests otherwise.
To have unlimited access to our content including in-depth commentaries, book reviews, exclusive interviews, PS OnPoint and PS The Big Picture, please subscribe
Daniel Gros
hopes that a Chinese startup's apparent AI breakthrough will provide a much-needed boost to European competitiveness, advises the EU on how to deal with Donald Trump, urges Europe to improve conditions for bottom-up innovation, and more.
Despite Donald Trump's promise to boost fossil-fuel production, the economic and technological forces driving the clean-energy revolution cannot be stopped. The global transition will power forward, even if America has abandoned climate leadership, and even if the road ahead includes a few more bumps.
assuages fears that the global energy transition will be thrown into reverse by the new US administration.
Perhaps US efforts to cut off China’s access to advanced semiconductors will be more successful than analogous restrictions on tech exports to France in the 1960s. But we now have at least one data point – DeepSeek – that suggests otherwise.
questions the effectiveness of efforts to limit China’s access to advanced semiconductors.
BERKELEY – Starting in October 2022, the late, lamented (by some) administration of President Joe Biden implemented restrictions on US exports of advanced semiconductors to China. A classic “dual use” technology, these chips power generative AI and supercomputers used in weapons systems, cyberattacks, and surveillance. Equally, they can enhance the ability of companies to compete internationally – in this case the ability of China’s high-tech firms to compete with their American rivals.
In its final months, the Biden administration doubled down on these restrictions, adding high-bandwidth memory chips and chipmaking tools to the list of prohibited items. Donald Trump’s administration, dominated by China hawks, is poised to ramp up these measures still further.
China has retaliated with export controls on rare earths and related materials used by American producers. But the more serious threat is that China will accelerate the development of its own capacity to design chips and build chipmaking equipment. If China closes the technology gap faster than it otherwise would have, then US export controls will have proven ineffectual or even counterproductive.
It’s hard to predict the success or failure of China’s efforts. But it may be informative to look to the past, where some of the most relevant history comes from, of all places, France in the 1960s.
French politicians and businessmen at the time perceived an economic threat from the United States. This was the era when the journalist and politician Jean-Jacques Servan-Schreiber warned that France could become an economic vassal of the US. Servan-Schreiber was worried about the invasion of Europe by US multinationals, with General Electric’s 1964 acquisition of the French computer company Machines Bull a prime example.
In addition, French President Charles de Gaulle objected to US dominance of NATO and refused to put France’s forces under joint command with other alliance members. Ultimately, he withdrew France from NATO’s integrated command.
PS Events: AI Action Summit 2025
It is not too late to watch our AI Action Summit event.
Click the link to watch world leaders, tech experts, and other distinguished speakers – including Justin Trudeau, Petr Pavel, Daron Acemoglu, Reid Hoffman, Marianna Mazzucato, James Manyiga, Audrey Tang, Sylvain Duranton, Celina Lee, Patrick Pouyanné, and others – discuss some of the most important questions raised by the rise of artificial intelligence.
Watch Now
This was the context of America’s decision in 1964 to deny licenses for the export of advanced IBM and Control Data Corporation (CDC) computers to the French Atomic Energy Commission. These giant mainframe computers could have been used for calculations enabling France to build a hydrogen bomb, which only the US and the Soviet Union possessed at the time. A thermonuclear capacity would have allowed France to pursue an independent defense policy in disregard of US wishes. Reverse engineering IBM and CDC machines also might have helped reinvigorate France’s commercial computer industry and overcome the sad dismantling of Bull.
Such was the background of the delightfully named Plan Calcul launched by the French government in 1966. The government envisaged investing more than one billion francs – about €1.5 billion today – in the French computer industry in the program’s first five years. Half would be provided by the state in the form of research subsidies for private firms, and half by the firms themselves.
To help the companies finance their share, the state extended subsidized loans and guarantees. An agency attached to the prime minister’s office coordinated the plan, which entailed merging two French companies to create a national champion, Compagnie internationale pour l’informatique (CII).
Plan Calcul was championed by de Gaulle himself, who saw a world-class military and a world-class economy as equally important for securing France’s status as a global power – for restoring its “grandeur.” One cannot but be reminded of the aspirations of Chinese President Xi Jinping and his personal role in efforts to advance China’s high-tech supremacy.
Ultimately, Plan Calcul failed to make France a leader in high-speed computers. The government’s technocrats, instead of working harmoniously with the private sector, micromanaged CII’s affairs, keeping for themselves the roles of bankers, entrepreneurs, supervisors, and clients. They focused not on commercial markets, where CII would have felt the chill winds of competition, including from IBM, but on government procurement contracts, where CII could stay fat and lazy. Rather than supporting the private sector, the government’s interventions impeded its progress.
Moreover, when forming their national champion, the technocrats excluded Bull, the firm with the most expertise, since it was now in foreign hands. Rather than complementing one another, national security and economic imperatives worked at odds with each other.
But if Plan Calcul was a damp squib, this was not because the US export-control regime was effective. Firms like IBM pushed back against the loss of business and were able to obtain exemptions. Information flowed freely between IBM’s plants in the US and France, where, as the CIA observed, it could be accessed by local scientists. When the latest computers were unavailable, the French were able to substitute them with slightly older versions.
Thus, France acquired the hydrogen bomb in August 1968, successfully testing a device in French Polynesia. In the end, US export controls only briefly delayed the geostrategically inevitable. Perhaps America’s latest resort to export controls will be more effective. But we now have at least one data point – DeepSeek – that suggests otherwise.