At the G7 summit in Evian last week, the leaders issued their first joint declaration since President Donald Trump returned to office, and German Chancellor Friedrich Merz called it “a real success.”
After a year of trade friction, a drawdown of American troops from Germany, and open strain between the two men, the change in tone was itself the news. But a communique does not hold an industrial base in place. The harder question, the one Trump and Merz left Evian without answering, is what economic architecture replaces the bargain that is clearly ending, the one in which America underwrote Europe’s choices without asking for partnership in return.
The answer should be the most ambitious U.S.-Germany economic partnership in a generation, because the two countries face the same strategic problem, and neither can solve it alone. For two decades, Beijing pursued industrial dominance with discipline while much of the democratic world outsourced strategic industries for short-term efficiency. Germany is feeling the result acutely, as China is again its largest trading partner.
In the first nine months of 2025, German imports from China rose 8.5% even as exports to China fell 12.3%, and German carmakers have lost roughly one-third of their share inside China since 2022. German power has recently traded at as much as four times the level in nuclear-powered France. This is the predictable harvest of a strategy the West was slow to name, calling it “interdependence” long after it had curdled into “dependence.”
The encouraging news is that Berlin now sees it. Merz has called Germany’s nuclear phase-out a “huge mistake.” His economy minister, Katherina Reiche, says natural gas is “the only baseload supply I have left.” These are German leaders speaking with unusual candor, and that candor creates room, for the first time in years, to do serious business with Washington.
Three things should happen before the end of 2026.
First, build a real energy partnership, because Germany cannot lead in artificial intelligence or hold onto its industry without one. The United States is now the world’s largest liquefied natural gas exporter and home to the democratic world’s most advanced small modular reactor programs. Germany is the natural anchor customer. A serious package would tie long-term U.S. LNG to German industrial offtake at predictable prices, pair American small modular reactor designs with Siemens Energy’s manufacturing base to seed a product the rest of Europe will want, and combine American grid software, proven at hyperscale, with Germany’s deep grid engineering.
The urgency is concrete. In Frankfurt, grid operators have effectively barred new data-center connections until at least 2030, and across Europe’s main hubs, grid connection waits now run as long as seven years, pushing American hyperscalers like Amazon to reconsider where they build. No ally can host the computers that frontier AI requires, or keep heavy industry at home, on those terms.
Second, fix the regulatory asymmetry on AI. In May, EU lawmakers agreed to postpone the AI Act’s hardest high-risk deadlines toward 2027. That delay is not a solution. It is a window.
As written, the Act’s burden falls hardest on the Western firms that actually sell frontier models into Europe, while Chinese developers, who ship far less into the market, absorb a fraction of it. The result is that European citizens get throttled access to Western AI while Beijing scales its own at home and exports it abroad. Washington and Berlin should use the reprieve to make a joint demand: one framework that applies equally to anyone selling into Western markets, Beijing included, rather than disproportionately to the firms that share Western values.
Third, build an industrial alternative to China together, each side bringing what it has. De-risking is a slogan until someone provides the risk-off destination. Germany brings advanced manufacturing depth that the U.S. needs and cannot quickly replicate. America brings frontier AI, capital, and energy. Manufacturing AI is something the two can build rather than buy from Beijing, and deployed inside allied factories, it makes the West’s industrial base more productive than China’s. Germany’s rearmament, its largest since reunification, should be coupled to American autonomous-defense firms rather than slow consortia rebuilding yesterday’s capabilities. And firms like Zeiss and Trumpf, whose optics and lasers make the West’s most advanced lithography possible, should be drawn deeper into the American semiconductor ecosystem, not left exposed.
None of this is charity. The U.S. is the world’s largest economy, and Germany is Europe’s. The old transatlantic bargain, in which America subsidized Europe’s choices indefinitely, could not continue, and Evian showed that both sides now know it. A joint declaration is a start, not a structure. The conditions for the real thing exist because Berlin has reached the diagnosis Washington reached years ago. What remains is execution. Make the offer, make it concrete, and make it now.
Noosheen Hashemi is co-founder and CEO of January AI. She helped scale Oracle during its hyper-growth years. A member of the Council on Foreign Relations, she also sits on the advisory councils of the Stanford Institutes for Economic Policy Research and for Human-Centered AI. Through her family office she is an active investor in American technology companies, including frontier-AI firms affected by the AI-regulation question discussed above.