April 13, 2026
FirstFT Asia · email · 5 mins
After 16 years in power, Viktor Orbán’s Fidesz party was routed in Hungarian parliamentary elections, winning just 54 seats — down from 133 in 2022. The opposition Tisza party led by Péter Magyar is projected to win 138 of 199 seats, a two-thirds supermajority sufficient to rewrite the constitution. Orbán conceded, saying his party would “serve our country and the Hungarian nation from opposition.”
Magyar’s victory is a direct blow to both Trump and Putin, who each counted Orbán as their closest ally in Europe. Magyar has promised to repair Hungary’s relationship with the EU and to unblock a €90bn EU loan earmarked for Ukraine that Orbán had obstructed — a move that will be welcomed across European capitals.
Separately, Trump announced a US naval blockade of the Strait of Hormuz after Iran nuclear talks collapsed in Pakistan, raising immediate concerns about oil prices and US-China tensions given China’s heavy reliance on Iranian oil. The IMF is expected to downgrade its global economic outlook this week at its Washington spring meetings, in part due to the widening US-Israeli conflict with Iran.
Quotable:
“Together, we liberated Hungary and got rid of the Orbán regime.” — Péter Magyar, victory speech to crowds in Budapest
The New York Times · email · 8 mins
On Feb. 11, Netanyahu sold Trump on war with Iran by promising regime change was a near-certainty, that a joint U.S.-Israeli campaign could destroy Iran’s missile program, and that Israeli intelligence would foment a popular rebellion. The CIA director immediately called the regime change scenario “farcical” and Secretary of State Rubio called it “bullshit” — but Trump was sold, marking the first time Israel had convinced a U.S. administration to go to war alongside it.
Six weeks later, none of Netanyahu’s promised outcomes have materialized: Iran’s stockpile of 400 kilograms of highly enriched uranium remains in the country, the government shows no signs of falling and appears emboldened, and the war has produced what opposition leader Yair Lapid called “the greatest diplomatic disaster in all of our history.” The war did degrade Iran’s capabilities, but Netanyahu faces elections before October and must somehow persuade the Israeli public that the war was a success.
Israel repeatedly pursued the fight on its own terms — striking Iran’s energy infrastructure against explicit U.S. requests, and opening a parallel war in Lebanon against Hezbollah. Lebanon’s death toll has now exceeded 2,000 (higher than the toll in Iran), approximately one million people have been displaced, and Israel has occupied a strip of southern Lebanon — all while Hezbollah continues firing rockets into northern Israel.
After Trump announced a two-week U.S.-Iran cease-fire, Netanyahu issued a terse supporting statement — then immediately escalated bombing in Lebanon. Netanyahu claims the cease-fire doesn’t cover Lebanon, contradicting both Pakistan (which helped broker the deal) and Iran. Trump has only asked Israel to “scale back” without enforcing anything; Israel responded by halting strikes on Beirut specifically while continuing attacks on southern Lebanon.
Weekend talks in Islamabad between the U.S. and Iran ended without a breakthrough, leaving the cease-fire fragile and Netanyahu room to maneuver. Trump’s blockade-of-Hormuz threat is his main leverage but not a politically viable move — Iran and Israel both know it. Netanyahu gave a 13-minute national address declaring “the battle is not yet over,” signaling he intends to keep fighting for as long as Trump lets him.
Quotable:
“Israel was not even at the table when decisions were made concerning the core of our national security.” — Yair Lapid, leader of Israel’s centrist opposition, on the U.S.-Iran cease-fire negotiations in Islamabad
Bloomberg · email · 4 mins
Trump announced a full US naval blockade of the Strait of Hormuz, set to begin at 10 a.m. Washington time Monday, after 21 hours of US-Iran talks in Islamabad collapsed without agreement on Tehran’s nuclear program or control of the waterway. Iran’s president told Vladimir Putin the country is “fully prepared” for a fair deal, but no progress materialized. Two oil tankers had already aborted transits through the strait as negotiations broke down, with crude prices and broader risk-off sentiment expected to follow.
Japan approved another ¥631.5 billion ($4 billion) for Rapidus, its state-backed chip startup, bringing total government backing to ~$16.3 billion through 2027 with a goal of mass-producing 2-nanometer chips to reduce dependence on TSMC. The competition is escalating: TSMC plans to spend over $50 billion in 2026 alone, and Elon Musk is entering the race with “Terafab,” partnering with Intel to build chips for Tesla, SpaceX, and xAI. Rapidus has IBM licensing its technology and ASML as a partner, with a potential IPO targeted around 2031.
Quotable:
“Wall Street off to the worst start to a year since the US regional banking crisis. The KBW Bank Index is trading at just 12 times forward earnings, a 40% discount to the S&P 500’s multiple of 20.” — Bloomberg, on earnings season kicking off amid Mideast tensions and AI disruption
Bloomberg Weekend · email · 7 mins
The post-American world order is not a jungle of pure predation but a wild ecosystem of shifting, overlapping partnerships — hence “polyamory.” Three forces drive the transition: power redistributing from the West/North to the East/South; non-state actors eclipsing states (Walmart’s revenue exceeds Sweden’s GDP; over 3.5 billion people use a Meta app daily; small militaries must negotiate with Elon Musk’s Starlink for battlefield comms); and globalization persisting despite populism and a pandemic, with world trade still forecast to grow at its prior decade’s pace.
Middle powers are the biggest winners in this structure. Greater agency and more options make countries fluid and opportunistic — willing to pivot between partners based on interest rather than ideology. Intensifying great-power coercion (US, China) actually accelerates middle-power coalition-building: they “hang together to avoid hanging separately.” Progress will come fastest on technical-cooperation issues — trade, public health, transnational crime, AI standards — while security deteriorates, with more countries in active conflict today than at any point since World War II.
Anthropic’s Claude Mythos model found thousands of high-severity software vulnerabilities — some reportedly decades old — across every major OS and browser, and was judged too dangerous for public release. Instead, Anthropic gave early access to 50+ organizations including Microsoft, Apple, Google, and CrowdStrike to patch their own systems first. Counterintuitively, AI will likely benefit cyber defenders more than attackers: offense has always needed only one path in, but defenders have always held more information about their own systems — AI finally lets them act on that informational advantage fast enough to matter. The catch: only well-resourced organizations with compute budgets will benefit immediately; smaller companies are exposed.
A US-Iran ceasefire took effect on Wednesday, but traffic through the Strait of Hormuz remains “severely constrained.” Kalshi prediction markets (as of Saturday evening) put the most likely return-to-normal date sometime in May. Deutsche Bank identifies China as the Iran war’s economic winner: its Iranian oil imports face a “severe test,” but its status as the world’s largest clean-tech producer positions it to help governments now desperate to reduce dependence on Middle East energy.
Quotable:
“Before AI, offense held a structural advantage in cybersecurity: Attackers need only one path in, while defenders have to secure everything… Defenders have always held more information about their own systems, but they couldn’t act on it fast enough to fend off many attacks. AI has the potential to close that gap.” — Michael Deng, Bloomberg Economics, on why Anthropic’s Mythos model may help defenders more than hackers
Joel Suss · email · 5 mins
Orbánomics has centralized economic power, rigged markets and public tendering against “opposition” firms, and dismantled independent institutions — central bank, judiciary, media, universities. The early years after Orbán’s return to power in 2010 produced growth fuelled by cheap Russian energy, foreign capital, and EU funds, but when those dried up they “sharply exposed pre-existing structural weaknesses.” Hungary is now stuck assembling German cars with chronically low productivity, under-investment in education and R&D, and stagnant GDP — made worse by oligarchic restructuring that kills competitive forces.
Hungary has the highest cumulative inflation in the EU since 2020: prices up 57%, nearly double the bloc-wide average of 28%. The proximate cause is domestic policy — pre-election fiscal stimulus in 2022 worth 2.2% of GDP to “buy” votes, ineffective government price caps, and a progressively neutered central bank that let annual inflation peak above 25%. With a budget deficit running at ~5% of GDP (matched almost exactly by family-policy spending alone) and the highest debt-servicing costs in Europe — lenders demand an explicit “Orbán premium” — there is little fiscal room to fix anything.
Orbán’s flagship pro-natalist programme — tax breaks and interest-free loans estimated at ~5% of GDP — has failed on its own terms. The fertility rate briefly recovered from its 2011 low to 1.6 births per woman, then collapsed back to 1.3 in 2025, below neighbouring Bulgaria and Slovakia, which spent nothing comparable. Macroeconomic stability, not policy incentives, appears to drive procreation: Hungary’s recessions since 2022 wiped out any demographic gains. A hard anti-immigration stance on top of sub-replacement fertility has reduced Hungary’s population by 500,000 people since 2011 — a 4.5% decline — with severe labour shortages now hitting healthcare and education.
Quotable:
“Orbánomics is a textbook example of right-wing populism with Hungarian characteristics.” — Joel Suss, summarising the model’s structural damage ahead of Hungary’s April 2026 election
Kyle Poyar · email · 11 mins
AI-native companies are scaling from $1M to $10M ARR in a median of 9 months — faster than IPO-caliber SaaS companies historically took to reach their first $1M. HappyRobot crossed $10M ARR 10 months after reaching $1M; GrowthX hit $12M ARR within a year of founding. The standouts at the extremes: Bolt went from $1M to $10M in under 2 months, HeyGen in 5 months, Gamma in 6 months.
Late starts don’t disqualify you — PMF for an AI product triggers the same explosive ramp regardless of company age. Clay, Gamma, and HeyGen each spent 2+ years before having a usable product; StackBlitz existed for 7 years before Bolt; Intercom was a 12-year-old SaaS company before Fin. Once AI-native PMF clicked, the growth curve was identical to greenfield startups.
Delay hiring account executives until $2M–$5M ARR. SaaS companies typically hired AEs at ~$1M ARR; AI companies deliberately wait, using the interval to build repeatable self-serve or founder-led motion first. The payoff is a GTM that skips legacy SaaS playbooks entirely — consultative, technically-led, and able to compress enterprise deals. GC AI sold to legal teams with no prior buying behavior for AI, closing fast through “naivete and first principles.”
POC velocity and AI automation rate replace pipeline metrics as primary leading indicators. 7AI (agentic security for CISOs) went from first conversation to full production deployment at DXC — 120,000 employees — in 8 weeks. Intercom’s Fin charges $0.99 per resolved outcome and tracks resolution rate paired with CSAT to prevent hallucination-gaming; Fin’s “total automation rate” measures how much of an account’s total support volume AI is actually handling, functioning as an AI-native share-of-wallet metric.
Forward-deployed engineers (FDEs) have replaced solution engineers as the GTM critical path, shifting from a 1:3 FDE-to-AE ratio in SaaS toward 1:1 in AI. 39% of the top 200 AI companies are actively hiring FDEs — including product managers, deployment strategists, and “solutions attorneys.” HappyRobot fields 25 FDEs out of 110 total employees. 7AI’s AI security engineers are assigned to every account, carry no quota, and can commit code directly to customer environments. Clay’s “GTM engineers” — only 20% from traditional sales backgrounds — run the entire sales cycle.
GTM engineering teams decouple growth from headcount by shipping growth plays the way engineering ships products. Clay’s 7-person GTM engineering team builds pre-meeting notes, bespoke demo tables, and personalized follow-up automation. Fyxer’s growth engineers ran 90 experiments each last year with full outcome ownership. incident.io spends nearly $1M/year on Claude and offers an unlimited AI budget, with the philosophy of “adding multiple digit percentage points of spend on AI to replace work.” The efficiency target: lifetime burn multiple below 1x (incident.io targets 0.5x) — a sharper signal than ARR-per-employee, which can mask poor gross margins and seven-figure token bills.
Quotable:
“We didn’t know you’re supposed to be slow or that enterprise deals should take six months. And we were selling to legal teams who are not used to buying AI yet. But we just moved fast. The naivete and first principles approach have been huge accelerants.” — Cecilia Ziniti, co-founder of GC AI
‘Lenny’s Newsletter’ via PubsforSubs · email · 6 mins
Hire “barrels,” not just headcount — the barrel-to-ammunition ratio determines throughput. A “barrel” is someone who can take an idea from zero to shipped outcome without hand-holding; most companies have only a handful. Adding headcount without adding barrels doesn’t increase output — it adds coordination tax and drag. The ratio of barrels to ammunition is what sets the ceiling on how many important things a company can pursue simultaneously.
For consumer products, customer research isn’t just useless — it actively corrupts decision-making. Consumer choices are subconscious; when you ask someone to explain a subconscious decision, they confabulate, even in good faith. Ask a Porsche owner why they bought the car and 99% will cite every reason except the real one. Once misleading feedback enters the organization it locks into people’s mental models and distorts every subsequent decision downstream.
The PM role is structurally incompatible with weekly AI capability shifts. Building a sequential year-long roadmap and coordinating between teams made sense when capability curves were flat. The skill that now matters across PM, design, and engineering is business acumen — understanding the company’s core equation and knowing what to build next. At Rabois’s top portfolio companies, the heaviest AI power user is now the CMO, not the engineers: running analytics, shipping campaigns, and generating insights that previously required entire deputy teams.
Quotable:
“Ask any Porsche owner why they bought the car, and 99% will cite every reason except the real one. Once misleading customer feedback enters the organization, it locks into people’s brains and distorts every subsequent decision.” — Keith Rabois on why consumer research is actively harmful
Mark Gurman at Bloomberg · email · 8 mins
Apple’s Vision Products Group originally planned three head-worn devices roughly a decade ago: an iPhone-tethered AR headset (target: early 2020), a mixed-reality headset (late 2021), and standalone AR glasses (mid-2022). Only the mixed-reality headset shipped, eventually launching as the Vision Pro in 2024. The tethered device was scrapped, and true AR glasses remain years away — roughly eight years behind the original plan.
Apple added display-free smart glasses to its roadmap only around 2022, after Meta’s camera-equipped Ray-Bans demonstrated real consumer demand for the category. The project is internally code-named N50, with an announcement planned for late 2026 or early 2027 and actual retail launch in 2027. Like Meta’s glasses, N50 will handle photos, video, phone calls, notifications, music, and hands-free Siri — which will be “significantly upgraded” in iOS 27.
Apple is going fully in-house on industrial design rather than partnering with EssilorLuxottica (Meta’s path) or Warby Parker (Google/Samsung’s path). At least four frame styles are in testing: a large rectangular Wayfarer-like shape, a slimmer rectangular style resembling Tim Cook’s own glasses, larger oval/circular frames, and a smaller refined oval option. Units use acetate rather than standard plastic, in colors including black, ocean blue, and light brown. The camera uses vertically oriented oval lenses with surrounding lights — a deliberate departure from Meta’s circular design — as Apple’s bid for an instantly recognizable “icon.”
The N50 glasses are part of a three-pronged AI wearables push that also includes new AirPods and a camera-equipped pendant. All three devices use computer vision to interpret the user’s surroundings and feed contextual awareness into Siri and Apple Intelligence, enabling features like improved turn-by-turn navigation and visual reminders. Apple’s bet is that tight iPhone integration, in-house chips, and its retail presence will replicate the Apple Watch playbook — not first to market, but ultimately dominant — even against Meta’s head start and Google’s Android ecosystem scale.
Quotable:
“If executed properly with a functional Siri, these glasses could follow a trajectory similar to the Apple Watch: not first to market, but ultimately dominant.” — on Apple’s competitive positioning in display-free smart glasses
The New York Times · email · 9 mins
An estimated 50,000 Americans are in a chronic vegetative state — technically awake but presumed to have no awareness — while another 200,000 to 400,000 are in a “minimally conscious state” with fleeting periods of awareness. Research suggests roughly a quarter of all these patients can follow commands like “imagine opening and closing your hand,” yet no one is systematically counting or testing them outside of research trials.
The discovery traces to 2006, when neuroscientist Adrian Owen at Cambridge University placed a 23-year-old woman diagnosed as vegetative into an fMRI scanner and asked her to “imagine playing tennis.” The motor-planning region of her brain lit up identically to a healthy person’s brain — proof of hidden consciousness. The finding was widely replicated and has held up for nearly two decades.
Despite 20 years of replication, covert-consciousness testing remains confined to a handful of research ICUs — only in Boston and London, Ontario are patients routinely screened. Journalist Katie Engelhart interviewed families of vegetative patients whose doctors never once raised the possibility that their loved one might be covertly conscious, leaving families to find the science themselves.
American ICUs operate under a documented “culture of pessimism”: neurologists often reach early conclusions that a brain-injured patient won’t recover meaningfully. The counterpoint is contested — most patients who emerge from the vegetative state retain severely impaired consciousness and significant disabilities — and what counts as a “good enough” life is not a medical judgment to make alone.
A 2011 study of locked-in syndrome patients — those who can only communicate through eye movement — found that only 7% wished for euthanasia, and patients who had been locked in longer were more likely to report being happy. The implication: clinicians and families may systematically underestimate quality of life for people with extreme neurological impairment.
Quotable:
“Unless I were in Boston or London, Ontario, where they test patients at a few I.C.U.s, I wouldn’t know if he were covertly conscious. So I would speak to him as if he were aware.” — Katie Engelhart, journalist, on what she would do if her own loved one were unresponsive