In political debate, the claim that France is suffering from “austerity” is one of the most absurd that can be made. France is a structurally socialist country. No single centre-right government can change that. Fifty-seven per cent of the country’s GDP is consumed by public expenditure. Fifty-seven. The scale of the French welfare state is enormous - even by Nordic standards. The tragedy is that France finances most of this through debt. That debt will one day have to be paid by the young people now marching in the streets. They will wake up to a brutal reality. Virtually every affliction France suffers from is the product of socialism - not of cuts.

Stuff that does NOT happen if there is an oversupply of compute The Dot com analogy is so incredibly off

GPUs are aging like fine wine, with rental rates up across the board since January 1st Approximate change YTD: - H200: +160% - B200: +120% - RTX PRO 6000: +65% - H100: +35% Even the 6-year-old A100 rents for more than it did at the start of the year Charts of the Week: a16z.news/p/charts-of-the-we…

Denmark spent decades building nearly 5,000 wind turbines Together, they produced 21.6 TWh in the last 12 months One nuclear plant in Sweden made 23.3 TWh Oh, and nuclear runs 24/7 — wind doesn’t There's one big reason Denmark has some of the most expensive electricity in the world transparency.entsoe.eu/gener… api.energidataservice.dk/dat… eia.gov/electricity/monthly/… iaea.org/publications/16059

Did anyone even read the details of Anthropic’s Q2 numbers? Excluding fair-value changes (which are non-cash items), they’re approaching break-even fast. Tremendous momentum, and a good sign that they can cover their compute costs. They can safely commit to more capacity.

New massive record in Korean exports over night. Here is why it matters. The earnings cycle still accelerates.

I do not know how things are in other countries. But in Denmark, when I follow the coverage of the unrest in France on our two public-service television broadcasters, the explanation I keep hearing is this: France has spent years cutting back on upper secondary education. Classes are overcrowded. Add teacher shortages and excessive centralisation, and young people take to the streets. That is only half the truth. And when you peddle half the truth, you are, as we know, also peddling half a lie. Here is what can be pieced together from the French quality press and the official figures. France’s public spending per upper-secondary pupil is above the OECD average. That should complicate any account which treats a lack of money as the whole explanation. There is some truth in the criticism: some lycées have large classes, and the French education system is centralised and bureaucratic. One example worth mentioning is Parcoursup, the higher education admissions platform. I suspect it would be a strong contender in any international competition for bureaucratic opacity and unnecessary stress. But, as I have already said, this is only half the truth. The other half concerns what teachers and school leaders are being asked to endure. I admit that this is anecdotal. But recently, a math teacher from Metz described in a long and serious post on X how the teachers simply burn out at the most problematic high schools. We need to dig deeper into this constant attrition of skilled high school teachers. Why are schools being left to carry so much of the burden? French politicians are no different from politicians elsewhere in Europe. They want as many young people as possible to complete upper secondary education and go on to higher education. For the usual reasons: education is supposed to break the cycle of inherited disadvantage, while higher education is supposed to raise productivity. I find those arguments only partly convincing when these benefits are treated as automatic consequences of expanding access. In my view, France faces a problem we also know in Denmark: schools are being asked to compensate for failures of integration and basic socialisation that extend far beyond the classroom. The danger is that, instead of institutions civilising their pupils, disruptive pupils end up barbarising the institutions. No school can function properly if it is expected to tolerate young people who turn up unprepared, have no interest in learning and spend their time causing trouble and destroying the learning environment. It is hardly surprising that such conditions can wear teachers down. They cannot reasonably be expected to shoulder the entire responsibility for correcting mistakes that the French political elite has spent decades making. La France Insoumise and other radical organisations have encouraged the protests and blockades. Mélenchon has called for non-violence. However, this formal appeal for moderation appears symbolic when the LFI mayor of Saint-Denis simultaneously allows himself to be photographed among demonstrators and calls their actions justified. Le Point has meanwhile reported on mobilisation by Étudiants musulmans de France and its cooperation with RED Jeunes. The former is linked to the Muslim Brotherhood milieu in a French Interior Ministry report; the two organisations helped establish the anti-Islamophobia collective Génération EDR, whose links to the antifascist group La Jeune Garde are also described by Le Point. That is the kind of convergence I mean by Islamogauchism in practice. The flags also deserve attention. Palestinian, Moroccan and Algerian flags have appeared at some of the protests, including scenes of disorder. To me, this is another indication that, for some participants, the agenda extends beyond class sizes and school budgets. We do not yet know the full picture. But what concerns me is the apparent convergence, within parts of this movement, of several forms of radicalism: Antifa activism. Decolonial ideology. Islamogauchism. Islamist networks that have learnt from the left to speak fluent woke. That is why I do not think it entirely misplaced to speak of an “Islamogauchist intifada” when describing this radical current within the French protests. The wider protest wave has already spread to Belgium and Italy. The Danish public-service coverage I have seen gives me far too little of this side of the story. Two questions: Is it the same elsewhere in the Western world? And if so, why?

𝟭𝟬𝟬𝟬 𝗬𝗘𝗔𝗥𝗦 𝗢𝗙 𝗠𝗔𝗖𝗛𝗜𝗡𝗘𝗦 𝗧𝗔𝗞𝗜𝗡𝗚 𝗢𝗨𝗥 𝗝𝗢𝗕𝗦. 𝗔𝗡𝗗 𝗛𝗘𝗥𝗘 𝗪𝗘 𝗔𝗥𝗘. In 1086, William the Conqueror surveyed England's land, property, and livestock. Back then, ~60% of the population worked in agriculture and fishing, and labour’s share of national income was roughly 60–75%. How much is labour’s share today? 𝟱𝟵.𝟳% in Q1 2026. I have been looking at the record from 1260 onward. During this time, humanity introduced an endless wave of automation: steam engines, mechanised farming, cars, computers, the internet, and much more. People certainly lost jobs, and entire trades virtually disappeared (farming fell from 60% to under 1% of the workforce). But labour’s share of income didn’t dwindle to zero. It has been wobbling, not trending down. AI certainly seems unusually interested in everyone’s occupation. But historically, we spend centuries saving labour while continuing to find new things for labour to do. It will abolish some jobs, change many more, and create roles that currently sound ridiculous. Describing a "Social Media Manager" to William the Conqueror would have required a considerable afternoon. History gives us reasons for optimism.

This is the diffusion of CPI inflation, measured similarly to Kevin Warsh's preferred PCE inflation diffusion metric. It's down. our full reaction @TheTerminal blinks.bloomberg.com/news/st…

Why should Hyperscalers slow their investments, when the IRR is currently improving by the month? I know funding markets are stressed, but the rest is telling the execs to march on

J.P. Morgan expects US earnings to grow 24% this year, and 16 percentage points of that come from tech. Europe is forecast to grow 19%. Tech’s contribution? About 0.2 points. Strong showing from the people who regulated it first.

@larstvede: more intelligence will lead to better choices and a brighter future. Throughout the history of mankind some variation of “the world is ending” has been alive and well always. Being optimistic about the future can be almost contrarian. The future is bright, and many great innovations are making it brighter still

𝐇𝐀𝐒 𝐈𝐑𝐀𝐍 𝐋𝐎𝐒𝐓 𝐓𝐇𝐄 𝐁𝐀𝐓𝐓𝐋𝐄 𝐅𝐎𝐑 𝐇𝐎𝐑𝐌𝐔𝐙? A critical story from the Middle East is being largely overlooked: 𝐭𝐡𝐞 𝐒𝐭𝐫𝐚𝐢𝐭 𝐨𝐟 𝐇𝐨𝐫𝐦𝐮𝐳 𝐢𝐬 𝐞𝐬𝐬𝐞𝐧𝐭𝐢𝐚𝐥𝐥𝐲 𝐨𝐩𝐞𝐧 𝐚𝐠𝐚𝐢𝐧—albeit via an extreme logistical workaround. 𝐊𝐞𝐲 𝐡𝐢𝐠𝐡𝐥𝐢𝐠𝐡𝐭𝐬: • JP Morgan estimates regional crude flows are back to ~98% of pre-war levels (~13m bpd transiting Hormuz). Goldman Sachs reports similar recoveries. • Transit is far from safe: 𝐭𝐚𝐧𝐤𝐞𝐫 𝐫𝐚𝐭𝐞𝐬 𝐡𝐚𝐯𝐞 𝐬𝐮𝐫𝐠𝐞𝐝 𝐮𝐩 𝐭𝐨 $𝟑𝟎/𝐛𝐚𝐫𝐫𝐞𝐥 to price in the risk. • The workaround: older, depreciated shuttle tankers move oil through under U.S. escort near Oman's coast, conduct ship-to-ship (STS) transfers in the Gulf of Oman to modern carriers, and head right back. 𝐓𝐡𝐞 𝐬𝐭𝐫𝐚𝐭𝐞𝐠𝐢𝐜 𝐫𝐞𝐚𝐥𝐢𝐭𝐲: Iran can still add friction and cost, but 𝐓𝐞𝐡𝐞𝐫𝐚𝐧 𝐡𝐚𝐬 𝐥𝐨𝐬𝐭 𝐭𝐡𝐞 𝐚𝐛𝐢𝐥𝐢𝐭𝐲 𝐭𝐨 𝐬𝐡𝐮𝐭 𝐝𝐨𝐰𝐧 𝐭𝐡𝐞 𝐜𝐨𝐫𝐫𝐢𝐝𝐨𝐫. U.S. forces have consistently neutralized attempted interdictions. Iran has lost its primary geo-economic lever, while its own export revenue is choking under sanctions. Meanwhile, Trump wants pump prices kept low heading into the midterms. If Teheran wants sanctions relief, 𝐭𝐡𝐞𝐢𝐫 𝐰𝐢𝐧𝐝𝐨𝐰 𝐭𝐨 𝐜𝐮𝐭 𝐚 𝐝𝐞𝐚𝐥 𝐢𝐬 𝐛𝐞𝐟𝐨𝐫𝐞 𝐭𝐡𝐞 𝐦𝐢𝐝𝐭𝐞𝐫𝐦𝐬, while the political incentive in Washington is highest. After that, U.S. willingness to compromise drops significantly.

When the machine first started replacing the horse, how big did people expect the machine’s output to be in the future? All horses? 10x the horses? 100x? @larstvede references the steam engine to explain how one must think about the future scope of AI. Hint: it’s not 1x human output.

𝗜𝗦 𝗔𝗜 𝗚𝗥𝗢𝗪𝗜𝗡𝗚 𝟱–𝟲𝗫 𝗜𝗡 𝟮𝟬𝟮𝟲? Technology futurists such as Ray Kurzweil have long assumed that global AI token production—the units of data AI models process—could grow roughly fivefold annually through 2050 and beyond. So what is happening in 2026? So far this year, global token production has risen to around 3–3.5x its level at the start of the year. And growth is accelerating. 𝗔𝘁 𝘁𝗵𝗲 𝗰𝘂𝗿𝗿𝗲𝗻𝘁 𝗽𝗮𝗰𝗲, 𝘄𝗲 𝗰𝗼𝘂𝗹𝗱 𝗿𝗲𝗮𝗰𝗵 𝟱–𝟲𝘅 𝗯𝘆 𝘆𝗲𝗮𝗿-𝗲𝗻𝗱. Three forces are driving this 1) “Reasoning” models consume far more tokens per task. An AI solving a complex problem may perform thousands of invisible reasoning steps before producing an answer. 2) Autonomous agents are rapidly taking off. They perform tasks through many interconnected steps—and 𝗮𝗴𝗲𝗻𝘁 𝘁𝗼𝗸𝗲𝗻 𝘂𝘀𝗮𝗴𝗲 𝗵𝗮𝘀 𝗿𝗶𝘀𝗲𝗻 𝗿𝗼𝘂𝗴𝗵𝗹𝘆 𝟯𝟬𝘅 𝘀𝗶𝗻𝗰𝗲 𝗝𝗮𝗻𝘂𝗮𝗿𝘆 𝟭. 3) Open-weight AI is exploding. At the beginning of 2026, these models accounted for around 26% of token consumption in the US and 16% in Europe. Today, the shares are around 60% and 65%—while centralized, closed models have continued growing too. Meanwhile, the marginal production cost on the newest AI chips is falling by around 80% annually as chips, architectures and algorithms improve. This creates an interesting economic paradox. If AI volume grows ~5x while cost per token falls ~80%, shouldn’t total AI operating costs stay roughly unchanged? No. Total operating costs are still growing by 60–100% annually. New chips only gradually replace the installed server base. AI revenues are therefore also rising rapidly as demand increasingly pushes against available capacity. 𝗘𝗮𝗰𝗵 𝘂𝗻𝗶𝘁 𝗼𝗳 𝗔𝗜 𝗶𝘀 𝗯𝗲𝗰𝗼𝗺𝗶𝗻𝗴 𝗰𝗵𝗲𝗮𝗽𝗲𝗿 𝘁𝗼 𝗽𝗿𝗼𝗱𝘂𝗰𝗲. 𝗕𝘂𝘁 𝗼𝘂𝗿 𝗮𝗽𝗽𝗲𝘁𝗶𝘁𝗲 𝗳𝗼𝗿 𝗺𝗮𝗰𝗵𝗶𝗻𝗲 𝗶𝗻𝘁𝗲𝗹𝗹𝗶𝗴𝗲𝗻𝗰𝗲 𝗶𝘀 𝗴𝗿𝗼𝘄𝗶𝗻𝗴 𝗲𝘃𝗲𝗻 𝗳𝗮𝘀𝘁𝗲𝗿.

UPDATE: AIs have achieved the highest *possible* score on the Mensa Norway IQ test - 151 3 years ago: cognitively impaired human (64 IQ) 2 years ago: average human 1 year ago: genius human Today: literally off the charts Next year?

𝗗𝗢𝗢𝗠 In 1995, Jeremy Rifkin warned us that software and automation were about to permanently eliminate the global workforce. Fast forward three decades: global employment hit record highs, labor markets ran critically tight, and companies spent years complaining they couldn't hire fast enough. Every major technological leap brings the exact same existential panic: "Machines take over." "Humans become obsolete." Humans invent hundreds of new industries, work more comfortably, and wonder what the fuss was about.

The presumption that the Fed raising short-term rates reduces inflation is predicated on the belief that higher rates reduce demand and investment. But what if higher rates don’t reduce demand and investment because the demand for intelligence and energy is unaffected by higher rates because winning the race for super intelligence has a near infinite ROI and the demand for compute will remain incalculable. Why won’t higher rates at this unique moment in history therefore lead to more inflation as interest costs are embedded in everything? And the problem is compounded as the more the Fed raises rates, the more inflation we will have and the more the Fed will need to raise rates further and so on. But what if the old models don’t apply to the current paradigm and the Fed is wrong? I think the Fed might have just made a mistake. Am I right or am I wrong?

Kevin Warsh: We're raising rates Trump:

Leveraging AI to enhance information digestion is top of mind for many investors today. @larstvede explains their process: “we use five different LLMs to identify the top 30 venture funds and top 50 liquid asset allocators in the world, […] what would their ideal portfolio look like?”. Full episode below.