All summer long, the Grand Continent will keep moving. Each day, we will bring you ideas wherever you are—ideas you won’t find anywhere else (but that will be everywhere come fall), texts that are hard to find or refreshing with our Sundays. To receive them directly in your inbox and to support this momentum, consider subscribing to the review
Two powerful technological revolutions are underway. The first concerns artificial intelligence and has attracted particularly sustained attention from the press. Some present it as the harbinger of massive changes, even of replacement of human beings by a “super-intelligence.” In 2018, Sundar Pichai, the CEO of Google, did he not declare: “AI is more profound than fire or electricity”?
The second, by contrast, has aroused far less interest: it is a connected ensemble of technologies, including solar energy, batteries, and electric vehicles. We will call this ensemble the “clean technologies triad” or TTP. Its diffusion and adoption today enable fossil fuels to be replaced. It is also what will become the energy foundation of the political economy and will have the greatest impact on the lives of populations worldwide.
Indeed, the TTP directly responds to a large number of challenges. It is already a reality and its adoption is driven by increasingly evident economic interests. By contrast, the scale and impact of AI are still uncertain.
The most crucial technological revolutions for society underpin what Carlota Perez calls a new techno-economic paradigm (PTE). A PTE rests on a key production factor, whose technology must meet several properties: it must be cheap and its price must continually fall; it must be inexhaustible in a foreseeable future; it must have varied applications; finally, it must increase productivity while reducing the cost of capital and labor . To these four criteria, two more must be added: it must meet existing fundamental needs or create new ones (for example, the mobile phone is now a “fundamental” need in most of the world) and enable new firms to develop and become indispensable.
The new techno-economic paradigms are expected to impact all aspects of our socio-economic life. The last three PTEs we have experienced are: the era of steel and electricity, the era of oil and internal combustion engines; and, more recently, the era of information and communication technologies (ICT). In the latter case, the semiconductor was the main “engine,” indispensable for the proper functioning of computers and the software that makes them useful.
The United States has dominated these three periods. If European and Japanese firms also occupied leading positions during the first two, in the ICT realm, the United States imposed its hegemony. In 2026, they are still trying to preserve this global domination in the first two paradigms by controlling energy from fossil fuels and by pushing back the threat now facing their dominance in the third paradigm, hindering China from obtaining semiconductors, advanced equipment, and the software necessary for their production. The objective is to slow its rise in the field of computing, and more specifically of artificial intelligence. American superiority in the oil economy and in PTEs related to semiconductors is indeed being challenged by China: in the first case, via the clean technologies triad; in the second, via the establishment of a parallel computing stack.
The underlying technology of the next PTE could be not AI, but the TTP, which directly addresses some of the current “grand challenges”: climate change and pollution linked to transport and heating, largely caused by the use of fossil fuels; access to cheaper and more sustainable energy to meet fundamental needs such as drinking water, food production, communications, and transport.
AI, the Foundation of the Next Techno-Economic Paradigm?
In the face of this global and seemingly inexhaustible enthusiasm sparked by AI and its large language models, it is crucial to determine whether it is merely a tool or something even more powerful. If widely adopted, could AI transform the global political economy and give rise to a new PTE? To answer this question, let us examine the strategy of the United States, so far dominant, and then that of China, whose development paths for AI are very different, with respect to the PTE criteria mentioned above.
A Cheap Production and a Price That Keeps Falling?
The first criterion of a new PTE is that the technology behind it becomes less and less expensive over time. The true cost of AI remains opaque: American firms initially marketed AI-based computing services at prices well below their production costs and only recently raised usage prices, in order to recoup at least some of the investments. The American AI strategy rests on massive investments in data centers, which require colossal amounts of electricity and, consequently, new production capacities that current consumers would need to subsidize.
The promise of AI is to dramatically raise productivity, to such an extent that even the most ambitious optimists now anticipate the emergence of a “general artificial intelligence.” This objective is pursued not only by the Big Techs but also by other AI-focused companies. Among them are OpenAI, the maker of ChatGPT, which was valued at $852 billion and employed fewer than 7,000 people in March 2026; Anthropic, valued at nearly $1 trillion and employing fewer than 7,200 people in May 2026; and SpaceX, owned by Elon Musk, born from the fusion of the former SpaceX and his company xAI. But to date, neither the Big Techs nor these AI-focused firms have managed to monetize their activities in this field.
These colossal investments involve only a small handful of companies. From 2015 to 2026, more than $177.5 billion was invested in OpenAI, at a moment when the AI fever and promises were swelling. In May 2026, rumors suggested OpenAI planned an IPO with a valuation above $1 trillion, now postponed to 2027. This was to be followed by Anthropic, which also planned an IPO in 2026 with a similar $1 trillion valuation . These valuations add to the hundreds of billions already announced by various tech giants such as Amazon, Google, Meta, Microsoft, Nvidia, and Oracle. The phenomenon is striking: the sums at stake are so enormous that even the Big Techs, despite their huge cash flows, turn to Wall Street for borrowing.
The speed, scale, and concentration of AI investments are probably unprecedented in world history and could even surpass those observed during the late-1990s Internet bubble. All of the world’s most valuable companies, with the exception of Saudi Aramco, are investing in AI. For example, in Q1 2026, Google, Amazon, Microsoft, and Meta reported investing over $130 billion in building data centers . Other estimates foresee American companies, notably tech giants and others, investing about $1 trillion in 2027 in AI-related installations and equipment . All these infrastructures will require massive electricity production capacity. The American tech giants are thus, in effect, moving from software‑intensive, low‑asset businesses to asset‑heavy operators of data centers. Their bet rests on the hope of an AI that will deliver such high productivity that these investments in software and data centers—assets whose obsolescence is rapid—will yield extraordinary returns .
The phenomenon is so significant that it tangibly contributes to U.S. GDP growth. For instance, economists at the Federal Reserve Bank of St. Louis calculated that AI investments contributed 1.3 percentage points to real GDP growth in Q1 2025 and 1.16 percentage points in Q2, before easing to 0.48 points in Q3 . According to the Fed: “This contribution helped avoid a sharper contraction in Q1 and accounted for 30% of GDP growth in Q2 and 11% in Q3.” As we write, it seems the scale of this investment has remained in 2026. These figures are clear: the future of the American economy rests on the success of AI. If we consider investment spending as a share of GDP, this marks the largest investment boom in U.S. history.
No other country, not even China, its main AI rival, invests on such a scale or with such intensity, concentrated in such a small number of firms.
The scale of the sums is not the only astonishing aspect: the losses that accompany them are just as striking. For example, OpenAI claims to have generated $13 billion in revenue in 2025 but recorded a loss of $8 billion. This trend is unlikely to end, as the company continues to subsidize AI usage while planning to allocate around $600 billion to infrastructure by 2030 . In May 2026, Anthropic announced its annualized revenue had surpassed $47 billion . Yet, like OpenAI, the company loses billions of dollars every year .
Although the unit cost of using AI (per “token,” i.e., the increment of computing power consumed) has fallen, the trajectory appears to reverse in 2026 and costs are set to rise. This is partly due to a new trend: AI vendors are changing their pricing model. They have begun charging a per-“token” amount, rather than the very low subscription plan originally used as a lure. In response, many of the largest users have reduced their consumption. In June 2026, Uber announced that it would limit its employees’ use of AI tools, even though it had previously encouraged as much usage as possible .
These elements do not allow definite conclusions, but they do suggest that in 2026 AI could continue to be affordable for certain uses thanks to subsidies, while consumers begin to cut back on the use of more advanced and expensive models. The trajectory remains open .
An Exhaustible Resource in the Near Future?
Are the possibilities for AI diffusion inexhaustible?
It is premature to say. It is in the realm of code that applications currently appear most promising. Yet, diagnoses differ from one source to another: some assert that real productivity gains would be modest , while others claim, on the contrary, that programming productivity has advanced significantly. AI is also used in various domains: student work, academic papers and presentations, medical reports, translations and transcriptions, meeting notes, legal briefs, music, art creation, and much more. But do these applications generate a massive demand and change our world to the extent that unprecedented investments justify themselves?
Some AI uses have become unavoidable, as it is now integrated into Google Search or Microsoft products, and users increasingly struggle, or even fail, to disengage from it. If some data suggest that AI applications are boundless, it is not certain that productivity gains will continue to grow at the same pace. As with any new technology, we may, in the future, discover applications that are yet unforeseen.
Finally, the inexhaustible nature of AI also depends on the cost of data centers and electricity. If the real cost per token is charged, its boundless nature will no longer be obvious.
A Broadly Diverse Range of Applications?
In principle, AI should be applicable to any function that uses data and can improve its performance. Could AI use become generalized and in what forms? Will there be a widespread adoption of AI agents, for example?
Can AI Increase Productivity While Reducing the Cost of Capital and Labor?
The current vision in Silicon Valley is that most workers will soon be replaced by AI. If realized, capital would replace workers. AI promises to make the remaining workers more productive. Regarding capital, if firms cannot set the price of their offerings high enough to recover their costs, massive investments in data centers might not make capital more productive. This problem is serious because, unlike bandwidth and rail, semiconductors degrade quickly due to rapid technological improvements. Moreover, due to the massive construction of data centers expected to continue in the coming years, electricity demand should rise rapidly, necessitating substantial investments.
For a technology to underpin a PTE, it must not only keep improving in capabilities but also in cost. In 2026, it remains uncertain whether applying AI to the same task, such as coding, will continually improve code quality while reducing costs. Will it enhance production quality or lower costs in a way that is economically significant? In other words, should the returns not diminish?
Does AI Address Fundamental Needs and Can It Do So at a Lower Cost?
Like any software, AI is not directly consumable. Its use aims to create added value or reduce costs by offering numerous services that were not previously available. To date, AI does not seem to directly address identified fundamental needs. However, applying AI to engineering, plant science, medicine, and other domains could yield new perspectives likely to lead to new products. In the future, AI could also meet fundamental needs, such as the desire for “friends” or “partners,” even if digital, through chatbots.
The Emergence of New Key Companies
In the emerging AI sector, OpenAI and Anthropic stand out as emblematic firms. However, unlike the flagship companies of earlier PTEs, these two firms are completely intertwined with the software, semiconductor, and now AI giants—Amazon, Google, Meta, Microsoft, Nvidia, and Tesla/SpaceX (xAI)—and depend on them entirely. As noted earlier, these two companies are in precarious financial situations, SpaceX’s stock having disappointed in July 2026 after its IPO the previous month.
Consequently, the growth of OpenAI and Anthropic seems to support the thesis that a technology can foster the rise of new firms, but they are not yet profitable. It is also possible that existing tech giants will eventually absorb this technology. This possibility suggests that AI may not be transformative enough to constitute the basis for a new PTE.
China’s AI Strategy
China’s strategy, very different from that of the United States, could it be more capable of making AI the foundation of a new PTE?
The development of AI in China has accelerated in an unexpected way, propelling its models to the level of American frontier innovations . China does not simply follow the American example: Chinese firms have indeed embraced an open-source strategy, making their AI models available to all users. These models can be downloaded, modified, and run freely. In 2026, these models spread globally due to their low cost, even within American companies that seek to cut expenses, and they turn away from expensive American proprietary models. This success is spreading even as the U.S. administration seeks to curb it, or even to halt it, not only in the United States but worldwide .
According to Goldman Sachs analysts, Chinese developers managed in 2026 to create world-class models while investing far less capital than American firms. This is the competitive edge for China: offering high-quality models at lower cost to capture market share and thus secure users worldwide.
As in the United States and Europe, Chinese companies apply AI to research, online advertising, or software production. What distinguishes China, however, is the concept of embodied AI, i.e., integrating AI into physical agents such as robots or drones . Beijing asserts that these tools are not designed to replace workers but to assist them.
Returning to the conditions necessary for a new PTE, China’s development model can reduce the cost of AI, enabling diffusion and widespread adoption. The choice of open source will allow many more users to innovate, creating new applications and new uses for AI while lowering the cost of adoption and use. Yet, truly transformative use cases remain to be identified in China as in the United States.
An Alternative PTE: Electric Batteries and Clean Technologies
Let us focus more precisely on the “clean technologies triad” or TTP, certainly much less in the limelight and at first glance less seductive, but which could serve more plausibly as a foundation for the birth of a new PTE.
For two hundred years, fossil fuels—transformed into energy by combustion—have been the world’s primary energy source. They are present in most aspects of our lives, whether in the transport of goods, the mobility of people, or heating. Replacing such a resource would trigger profound changes at several levels: in daily life, in the flow of energy, in pollution sources, and also in the distribution of political and financial power worldwide. In 2025, investments in the energy transition reached $2.3 trillion, and are expected to rise further given the current geopolitical climate and as demand for electric vehicles, batteries, and solar installations increases .
The promises of this clean energy initially sparked interest as a solution to pollution and climate change, partly caused by the use of internal combustion vehicles. Until recently, it was expensive and inefficient, especially in transport. It was only economically viable in niche markets or with substantial subsidies. However, the advances made in the last decade, both technologically and in production, have not only improved performance rapidly but also reduced prices. A better refinement of batteries is the key factor behind the growing success of clean energies and electric vehicles.
The first real breakthrough in battery use came with the introduction of the Prius in Japan in 1997, Toyota’s hybrid vehicle, which for two decades stood as the showcase of automotive electrification. This innovation significantly boosted demand for higher-performance batteries, since the Prius still relied on fossil fuels. The second step in this rise was the use of lithium batteries, initially employed in consumer electronics, which underwent rapid technological advances and a dramatic drop in costs.
The considerable progress achieved in this field has enabled fossil fuels to be replaced in a growing number of applications. Batteries have two characteristics that ease this transition. First, they economically justify the time lag between energy production and consumption. This was crucial for enabling renewable energy to be exploited, whereas fossil fuels can store energy naturally. Second, their ability to store enough electricity without economic loss allows energy to be used without being connected to a grid. Batteries can thus store energy and make it available anywhere. This is what now makes electric vehicles a viable alternative to internal-combustion vehicles.
By 2026, solar electricity combined with battery storage is the cheapest form of energy production and supply in most regions of the world. The advantages of this technology continue to be highlighted, thanks to steady progress in both photovoltaic production and the profitability of batteries and electric motors. These advances are on the verge of making the TTP the dominant energy source, replacing fossil fuels.
A Low-Cost Production and a Price That Keeps Falling
Electricity can now replace gasoline and diesel in transportation, and natural gas in cooking and heating, under favorable economic conditions. This is made possible by the constant improvement of batteries in terms of energy density, charging time, and safety, while their costs keep falling. It becomes increasingly economical to shift electricity production in time, which helps smooth the intermittency of renewables. Batteries that are light and high in energy density allow access to energy without a physical connection to a source. The ability to generate electricity anywhere, as long as there is sun or wind, and then store it for later use, has profound implications: it challenges the very foundations of the twentieth‑century economy centered on fossil fuels.
In 2025, Ember, a consultancy focused on energy research and the TTP, observed that renewable energies were increasingly preferred for new energy projects . Their cost advantages drove this choice. For example, the levelized cost of energy for a solar-battery hybrid project is $59 per megawatt-hour, versus $102 per MWh for a natural gas turbine project. Even accounting for storage, renewables are now the cheapest option, and prices continue to fall rapidly.
As 40% of the cost of an electric vehicle comes from its batteries, the price of EVs is also dropping. The same is true for other components, notably electric motors. Finally, cars are redesigned to be optimized with the new powertrains, which will improve efficiency and reduce costs as well.
An Exhaustible Resource in the Near Future
The energy produced from renewable sources is inexhaustible. It is now possible to install photovoltaic panels anywhere. Spaces that could be described as “idle,” such as building facades, roofs, and even car roofs, can be repurposed to generate energy. Parking lots, for example, can be covered with solar panels. In California, the aqueducts have begun to be covered with solar panels that produce energy and reduce evaporation, thereby saving water. Initially, electric vehicle batteries relied on nickel-manganese-cobalt technology, but they are gradually being replaced by far cheaper lithium-iron-phosphate batteries that are also highly effective. Sodium batteries, even cheaper, are also making inroads. The decline in the cost of sodium batteries will make their supply essentially inexhaustible. Energy storage will thus be as abundant as the production of clean energy itself.
A Broadly Diverse Range of Applications
Energy is a fundamental input for all economic activities, and electricity is the most technically portable form of energy. In transportation, few anticipated that batteries could replace kerosene, but regional air transport and various battery-powered inland water transport are under development. Fossil fuels are already being displaced in all forms of transport. Finally, the TTP is increasingly competitive for energy uses beyond transport as well.
Southern countries depend more on fossil fuels than developed ones, due to weaker electrical grids. However, solar energy and batteries make electricity accessible even in the most isolated areas, provided there is sufficient sun. This has created new markets for electric vehicles, particularly scooters and motorcycles. Access to inexpensive energy for those who currently rely solely on costly fossil fuels broadens the applications of the TTP.
Moreover, the continual improvement of batteries enables new applications once unimaginable. As higher energy density batteries appear, humanoid robots will become even more capable. Paradoxically, AI development requires enormous amounts of energy, which will come from not only renewables but also costly fossil fuels and nuclear energy.
The TTP Increases Productivity While Reducing the Cost of Capital and Labor
The TTP will have a significant impact on the cost of energy-related investments.
Fossil fuel energy costs comprise two elements: investment costs tied to large-scale drilling, maritime transport, refining, and delivery; and the variable cost of the fuel itself, which cannot be recycled after use. Solar cells and lithium batteries likewise require a massive extraction, transformation, and production infrastructure, but their variable costs are virtually zero. On the construction scale alone, a fossil-fuel plant is now more expensive than a solar plant; and this comparison ignores the vast upstream extraction and transport infrastructure that the former entails. The costs of the TTP, finally, continue to fall thanks to predictable technical innovations and scale economies in mass production.
The impact on labor costs, however, is harder to quantify. In vehicle production, electric cars have fewer components and are therefore less complex than internal-combustion vehicles. They require less labor and lower capital investments in factories, both for production and design. Similarly, photovoltaic cell production is highly automated and does not require substantial labor. By contrast, installing solar systems will require a sizable workforce. In general, electricity is simpler and easier to manage than fossil-fuel energy. From a broader perspective, the TTP reduces energy costs and should thus lead to a lower cost of living.
The TTP Addresses Existing Fundamental Needs or Creates New Ones
The TTP directly meets the fundamental human need for energy, which is crucial in any society. As the cost of renewable energies falls, the prices of fertilizers, seawater desalination, etc., should also drop. Advances in batteries will reduce the need for a direct connection to a power source for means of transport, such as trains and subways. While humanoid robots, drones, and many other products existed before powerful and affordable batteries, their development has become possible thanks to ongoing improvements in these technologies.
The Development of New Key Companies
The electric-vehicle economy has given rise to several emblematic firms, with Tesla standing out as a pioneer in the global electric-vehicle industry and offering cutting-edge models. Tesla also produces batteries for electric vehicles but sources others from various suppliers, many of whom are Chinese. All other industry leaders are Chinese, the most notable being BYD, which now sells more electric and plug-in hybrid vehicles than any other company in the world. BYD is also the second-largest battery manufacturer globally. Besides BYD and Tesla, other Chinese EV makers, such as Xpeng, Nio, and Geely, have grown rapidly. There is also the less-known Chinese company CATL, which accounts for about 40% of the world’s EV batteries, along with many other smaller manufacturers. China now produces around 90% of global photovoltaic cells and nearly all wafers. No traditional automaker or battery maker reaches 10% of the world market; the only close contender is LG Energy Solutions in Korea.
It thus seems increasingly likely that Chinese companies will lead and guarantee the global transition to clean technologies.
Clean Technologies as the New Techno-Economic Paradigm
While AI has drawn remarkable interest, the TTP benefits from ongoing technological improvements and its adoption is accelerating worldwide. As for which of these two technologies will be at the heart of the next PTE, the TTP seems more likely to become the pivot, given current trends. That said, AI remains a premier new technology that could further enhance the TTP.
Artificial intelligence has the potential to become a powerful general-use technology, and the Chinese open-source model helps reduce its usage cost, but for now it does not directly address any of the world’s grand challenges nor does it seem to meet society’s fundamental needs. It could, however, play a very important role in improving the chemistry of batteries, solar panels, and many other aspects of the TTP.
The development of electric vehicles directly addresses the climate challenge while reinforcing national energy self-sufficiency and delivering significant lifetime savings. Add to this the reduction of local pollution from aerosols in brakes, the decrease in antifreeze use, and the elimination of oil leaks from internal-combustion engines. While these benefits do not always show up in possession costs, they have real environmental and social impacts. Certainly, the TTP also has environmental impacts, particularly during the extraction, processing, and fabrication of minerals, but they are likely to be far less than those associated with internal-combustion vehicles. Moreover, batteries, solar panels, and EVs are recyclable; this is not the case for combustion engines, and fossil fuels are far from being recyclable. Their combustion produces heat and chemical pollution. More importantly, the TTP is cheaper and more resilient.
The rise of the TTP has geopolitical implications. For example, one pillar of the United States’ global power is the petrodollar, which adoption of EVs and batteries could threaten, even without the war with Iran. For many countries, one of the most constraining sanctions has long been the denial of access to fossil fuels—a situation Germany and Japan experienced during World War II.
In 2026 worldwide, outside the United States, the TTP attracts the greatest volume of investment. For most people on the planet, access to reliable and affordable energy and mobility matters far more than AI-generated code, news summaries, advertising, music, or art. While the TTP is already well established and presents clear environmental and economic benefits, AI is promising as a tool. A real transformation or simply another instrument—whether the answer is yet to be written.