A fortnight ago Trump flew to Beijing and brought Silicon Valley with him: Jensen Huang, Tim Cook, Elon Musk, plus the Wall Street contingent. A few days later Washington quietly cleared Nvidia to ship H200s to Chinese firms. The same export controls the US spent two years building were loosened in the space of a state visit. The following is a compilation of what I’ve been reading on the US-China-EU triangle, the Draghi report a year and a half on, and the question underneath all of it: who actually controls the technology stack, and what is Europe doing while the other two decide?

The Beijing trip and what it signalled

To start, the optics. The South China Morning Post and PBS both ran the manifest: the most valuable companies in America, in one delegation, in Beijing. The White House framing, per Euronews, was about getting China to “open up” to American firms on trade, AI and market access. Xi’s line back, reported by CNBC, was that China’s door would “open wider.”

The substance is more interesting than the choreography. The genuinely new thing is the Nvidia H200 clearance: for two years the US treated advanced chips as a national-security chokepoint to deny China. Now it’s a bargaining chip to be traded for soybeans and rare-earth access. That’s not a tweak; it’s a reframing of what the chip is. Less has changed structurally than the headlines suggest (the rivalry is intact), but the logic shifted from containment to transaction. Worth holding onto that, because it’s exactly the lever the US has over Europe too.

The thing Europe can’t escape: rare earths

Here’s the asymmetry that frames everything. The US can loosen chips because it controls them. China can dangle market access because it controls something Europe can’t substitute: rare earths and processed critical minerals.

ECFR’s “Beijing hold’em” is the sharpest thing I’ve read on this. The argument: dependency is vulnerability, and in 2025 Beijing proved it, using its rare-earth monopoly as a geoeconomic weapon, with export-licensing measures that left European automakers, wind-turbine and defence producers triaging which products to build based on material shortages rather than demand. Their companion piece, “Don’t look down”, maps the clean-tech version of the same trap. MERICS notes Beijing explicitly limited rare-earth flows to pressure the EU on EV tariffs ahead of the summit: coercion as routine negotiating tactic.

So Europe sits between two giants who both have a hand on its throat: the US with the chips and the trade relationship, China with the minerals and the manufacturing. Von der Leyen calls the China relationship “one of the most defining of the 21st century” (€2bn of trade a day, twice the EU-Switzerland volume), and her three priorities (rebalance, de-risk, diversify to Vietnam/Brazil/India) are sensible. The problem an MEP put bluntly in the same piece: diversification targets mostly lack China’s refining capacity. De-risking is a decade-long industrial project dressed up as a near-term strategy.

Draghi, a year and a half on

Which brings me to the document everyone in Brussels cites and nobody implements. A bit of background for anyone new to it: the Draghi report (September 2024) diagnosed Europe’s competitiveness decline and proposed three pillars: close the innovation gap with the US and China, decarbonise without deindustrialising, and reduce strategic dependencies. The financing ask was ~€800bn a year in new investment, partly through joint debt.

The one-year verdicts are brutal and worth reading in sequence:

  • ECIPE, “The Draghi Report Turns One”. The cleanest metaphor: at one year old most babies stand and wobble forward; Draghi’s report “remains largely in its crib, frequently cited, politely praised, but rarely acted upon.”
  • Saraceno, “Europe Still at the Starting Blocks”. The report is “gathering dust,” and unable to compete on innovation, European firms default to price competition by compressing margins, which starves the very investment that would let them innovate. A vicious cycle stated plainly.
  • CaixaBank Research, “How far has the EU progressed on the Competitiveness Compass?”. The institutional view: positive on direction and debate, limited on delivery, with key decisions still pending and NGEU money running out in 2027.

And the number that should end any optimism: per the ECB, the strategic spend needed for the green, digital and defence transitions is now ~€1,200bn a year over 2025–31, up from €800bn a year earlier. The diagnosis didn’t just go unactioned; the bill grew while Europe stalled. Draghi himself said the lessons are “even more urgent” now, partly because, as Science|Business reported, Brussels signed a trade deal committing to buy ~$40bn of US AI chips. Europe’s answer to technological dependence on Washington was, in effect, to buy more from Washington.

The finance angle: even the spenders are flinching

Here’s the twist that makes me doubt the “just spend €1.2tn” framing is sufficient on its own. Even in the US (where the capital exists, the chips are domestic, and the spending has been enormous), the returns are now in question.

Axios’s “Corporate America enters its AI reckoning” is the piece I keep coming back to. The reporting: Microsoft cancelled most of its Claude Code licences partly over cost; Uber’s COO says AI spend is “harder to justify”; one consultant’s client burned through half a billion dollars in a single month after failing to cap employee usage. The diagnosis is that enterprises confused activity with value (“tokenmaxxing,” burning tokens to look modern) while one model-training CEO admits the uncomfortable truth that “the reality of AI right now is that it only works for coding.” Use cases default to automating annoying tasks rather than revenue-driving ones; data access is hoarded, so agents underperform.

Why this matters for the European debate: the case for massive EU tech investment implicitly assumes the returns are there to be captured if only Europe spends. But if the country that’s already spent (with cheaper capital and sovereign chips) is hitting an ROI wall, then “out-invest the gap” is necessary but nowhere near sufficient. The harder question is allocation and absorption: can you actually convert capex into productivity? That’s the unglamorous, governance-and-implementation question, and it’s the one Draghi’s critics (ECIPE, Saraceno) keep circling back to.

Outlook

So where does the triangle settle? My read: the US has demonstrated it will weaponise and trade away its chip advantage depending on what Trump wants that week, which makes it an unreliable anchor for European technology policy. China has demonstrated it will use rare earths as routine leverage, not just a crisis tool. And Europe has demonstrated, a year and a half after Draghi handed it the playbook, that it can produce world-class diagnosis and near-zero execution.

The optimistic case is that coercion is clarifying: nothing concentrates the mind like a chokehold, and the de-risking communications, the Industrial Accelerator Act, and the critical-raw-materials push are at least pointed the right way. The pessimistic case (the one the reading leans toward) is that Europe keeps mistaking strategy documents for strategy, and that the €1.2tn gap is the kind of number that gets cited at conferences rather than financed.

What I’m watching: whether the next EU budget negotiation (with NGEU expiring and limited fiscal space) actually puts money behind the Compass, or whether 2027 becomes the year the Draghi report turns three and still can’t walk. And whether the AI ROI reckoning in the US makes European caution look prudent rather than slow, for once.

Disclaimer: most of the above was read across the past few weeks and synthesised here; the through-line is mine and the sources don’t always agree with each other, which is rather the point.