Delian Asparouhov: Eastern Europe is the new tech frontier, zombie acquisitions are a temporary aberration, and Trump's AI executive orders are coming

Summary

Delian Asparouhov, co-founder of Varda and partner at Founders Fund, returned from a multi-week tour of Eastern and Western Europe with a sharply bullish view on the Eastern European tech corridor and a correspondingly bearish read on France and Western Europe.

Eastern Europe vs. Western Europe

The contrast he describes is stark. In Eastern Europe, a founder showed him a construction zone that would take six to eight months to finish in the US, with a projected move-in date of three and a half weeks, enabled by round-the-clock labor and essentially zero regulatory friction. The founder's view on equipment permits: government officials don't understand the technology well enough to regulate it.

In France, Asparouhov met a founder running a profitable robotics-enabled consumer business with strong unit economics who was forced to shut the company down entirely because French labor law made it impossible to execute a 30% reduction in force. Severance liabilities exceeded the company's entire balance sheet. The French government, when approached directly, refused to allow a partial layoff and demanded full severance for all employees. The business no longer exists.

On macro indicators, Poland's GDP per capita is on track to cross the United Kingdom's within two years. The UK has been essentially flat since 2008. Eastern European talent is repatriating home as salary growth and cost-of-living dynamics increasingly favor Warsaw over London, where real estate has been bid up by foreign capital. Asparouhov sees the Poland-Ukraine-Romania-Bulgaria belt as the likely production and engineering hub for the European defense build-up, with Western European offices handling marketing and small-scale R&D. Dedicated funds are already forming around the Ukrainian defense tech ecosystem.

Mistral and Apple

On Mark Gurman's report that Apple is seriously considering acquiring Mistral, Asparouhov is skeptical on both sides of the deal. Mistral, he argues, has quietly pivoted away from being an OpenAI competitor into something closer to a French government-anointed AI implementation consultant, deploying into enterprises like LVMH, Airbus, and Thales at the direction of Paris. President Macron has personally championed the company. That positioning makes a French regulatory approval of an Apple acquisition implausible and also makes Mistral the wrong asset for what Apple actually needs. Apple's problem is the consumer application layer, and Mistral is now an enterprise services business. Asparouhov frames Apple's failure to establish itself in consumer AI as its first major missed platform shift, following dominant positions in consumer PC and consumer mobile.

More broadly, he argues that sovereign AI strategies focused on domestic data centers and domestically trained models miss the point if citizens are accessing AI through chat.com or ChatGPT. The only area where sovereign AI efforts have real soft-power teeth, in his view, is open-source models, where weights embed values at the distribution layer. DeepSeek's refusal to acknowledge Tiananmen Square in its open-source weights is his primary example.

Zombie Acquisitions

Asparouhov views the current wave of acqui-hire structures, where a major acquirer takes the founding team and leaves a hollowed-out shell company, as a temporary aberration rather than a structural shift. He expects the gap to close from two directions: new legal terms appearing in AI company term sheets, and the FTC eventually asserting that minority acquisitions above a certain transaction size require regulatory review.

He distinguishes the Windsurf situation as particularly egregious relative to comparable deals, citing poor treatment of remaining employees on vesting acceleration and liquidity. The remaining Windsurf team's decision to join Cognition he describes as a credible competitive threat to the team Google acquired, meaning Google may have inadvertently funded its own opposition. He also notes the backdrop of the Adobe-Figma deal, where Adobe paid a $1 billion breakup fee after the FTC under Lena Khan blocked it, as the context in which acquirers now view acqui-hire structures as a regulatory shortcut. That window, he believes, is closing.

Trump AI Executive Orders

Asparouhov is co-hosting an event with the All-In team and Hill and Valley at which President Trump is expected to sign a new set of AI executive orders live, going beyond the administration's initial orders on winning the AI race and economic growth from AI. Confirmed speakers include Lisa Su of AMD on chips. The event will also feature companies focused on AI factories, data centers, energy infrastructure, and AI-powered industrial robots, framing the AI story around physical-world deployment and domestic job creation rather than software platforms.

David Sacks in his role as AI and crypto czar and Michael Kratsios, head of science and technology at the White House, are the primary policy architects behind the event. Asparouhov describes the bipartisan case for AI as durable, noting Obama's recent public alignment with the abundance caucus, where Obama's Democratic approval rating sits at approximately 96%, compared to the next highest figure of around 65% for Hillary Clinton. California Representatives Ted Lieu and George Whitesides are cited as Democratic voices on AI and aerospace industrialization appearing at the event.

On the question of whether Trump's data center investment announcements reflect genuine government leverage or simply market activity receiving political branding, Asparouhov points to concrete government lending programs in defense manufacturing and natural resources as proof of concept. In both sectors, the administration has created program offices offering guaranteed lending at below-market cost of capital for projects aligned with policy priorities, analogous to Fannie Mae and Freddie Mac in the residential mortgage market. He does not have specific data on equivalent programs for data centers but says he would not be surprised if they exist.