PodBrowser
Impact Theory

AI Boom Just Hit a Wall, Eight European Countries Are ON FIRE Right Now, Poland and Hungary Just Sent Europe a Warning | Weekly Recap

Sunday, 25 October 2026 · 4 min read · Listen to the episode ↗

AI’s infrastructure boom is meeting power and financing limits: Amazon reportedly has Nvidia racks sitting unused, while Anthropic’s rapid revenue growth comes with heavy compute costs and long-term commitments. The episode surveys unrest in France, Italy, Belgium, Albania and elsewhere, discussing economic pressures and competing explanations. Poland’s Donald Tusk and Hungary’s Péter Magyar reaffirmed hard-line migration policies, which the discussion connects to national cohesion and historical experience.

Protests and violent unrest were reported in at least eight countries: France, Italy, Belgium, Albania, Spain, Argentina, Brazil and India. In France, the speaker said more than 500,000 people marched on October 17, 1,200 were arrested and 150 police officers injured; looting, fires and property damage were reported across cities, with mayors seeking partial curfews. It remained unclear whether the looting was random. Riots were also described in Milan, Brussels and Liège, while Albania saw five consecutive days of street battles, including the burning of its Supreme Court, as protesters demanded Prime Minister Edi Rama resign. Spain had a march of 20,000, and student clashes in Buenos Aires were linked to university cuts.

The episode framed economic distress as a major source of unrest, citing inflation and the effects of COVID-era decisions. Eurozone inflation reached 3.8% in September, a three-year high; the speaker attributed nearly half of it to the Iran War energy shock and said average EU diesel prices were up 39% since January. In France, student protests over education funding, teacher shortages and pay, including school blockades, preceded other groups joining. A respondent cautioned that students’ grievances were legitimate but that describing the wider unrest as “student riots” was misleading. The host also cited energy problems and immigration, arguing that immigration strains schools and the social safety net.

France’s political structure was presented as making conflict especially combustible: centralized authority, high public expectations and low trust leave local and regional governments with limited power, while the national government becomes both the target of anger and the expected provider of benefits. The speaker cited public spending at 57% of GDP, civil servants accounting for one in five workers, and pensions consuming nearly 14% of economic output. Organized groups, the argument went, can mobilize against reforms more easily than dispersed young workers and future taxpayers. Polls cited in the episode showed 70% of French adults saying there were too many immigrants; a September 2025 CSA poll found 79% wanted tougher immigration policy, 63% supported suspending legal immigration for three years, and 83% backed sending home foreigners convicted of crimes or long-term unemployed. These figures were presented alongside a claim that anti-immigration sentiment is boosting Marine Le Pen’s popularity.

Hungary’s Peter Magyar said the country would keep its toughest line on migration and would not loosen border policy under pressure. Poland’s Donald Tusk said Poland would accept no asylum applications, would not repeat or pay for other countries’ mistakes, and would deport anyone breaking Polish law. An unnamed Polish lawmaker warned migrants not to approach the border, saying they risked being shot. The host and a cited argument by Branko Milanovic connected these positions partly to national cohesion and historical experience. The host said Poland went from 66% Polish in 1939 to nearly 99% in 1950 after the Holocaust, Stalin’s border shift and the expulsion of Germans; Hungary’s homogeneity was linked to the 1920 Treaty of Trianon and the murder of Hungarian Jews in 1944. The account also noted that Poland issued nearly one million residence permits in 2021, mostly to Ukrainians, while Hungary under Viktor Orbán brought in tens of thousands of guest workers from the Philippines, Vietnam and India. Whether those workers assimilate was left unresolved.

The AI infrastructure boom was described as running into power and financing constraints. Amazon is reportedly taking delivery of about $5 billion in Nvidia GB300 server racks in August and roughly $3 billion more each month, but some equipment is being stored because data centers lack power, with no clear forecast for when all of it can be installed. Ed Zitron estimates $200–300 billion of AI chips industry-wide are sitting unused. The host argued that power shortages could leave the West behind China and that AI investment competes for capital while contributing to more expensive US government debt. Oracle reportedly faces power-related delays at a New Mexico data center; its rising credit default swaps could also make financing more expensive. Nvidia’s new DSX pod in Berlin was described as a data center in a standard shipping container, but smaller or remote facilities still cannot solve a shortage of electricity.

Anthropic’s reported financial figures showed rapid growth alongside large losses and commitments. A draft IPO prospectus reportedly showed an $80.7 billion loss in the first half of 2026, versus $13 billion a year earlier, with about $72 billion attributed to marking ownership interests to market rather than current cash losses. Six-month revenue was $16.3 billion, up from $1.1 billion, while compute spending rose to about $9 billion from $2.3 billion. Third-quarter revenue was reportedly $16.8 billion, but quarter-over-quarter growth slowed to 45% from 146% between the first and second quarters. The host cited roughly $450 billion in non-cancelable cloud agreements over seven to ten years, up to $86 billion in Nvidia-chip rentals and more than $160 billion in Google-chip leasing obligations, potentially around $750 billion in total commitments. He estimated annual commitments of $60–75 billion and warned they could absorb current revenue before payroll, research and development, or other costs; slower growth, customer concentration and cheaper Chinese open-source models add exposure.

Ray Dalio was cited as calling AI a classic bubble that might have about 20% more room to inflate but was nearing a pop.

This summary was generated from the episode transcript and can contain mistakes.