The Archive
Past Briefs
Every weekly AI podcast digest, in reverse chronological order.
Week 29 · 2026
Compute concentration, RL's reward signal failures, and the hard limits of vibe coding dominated the week's most substantive conversations. A $20B+ Cerebras-OpenAI hardware deal, Cosine's UK sovereign LLM mandate, and Anthropic's inference-driven cluster contracts all point to the same underlying dynamic: whoever controls compute controls the frontier. Meanwhile, practitioners from Wix to Cosine are surfacing the unglamorous ceilings — business logic complexity, code slop from correctness-only rewards, and the still-unsolved problem of AI agent memory — that separate the hype cycle from what's actually shipping.
Read brief
Week 28 · 2026
Anthropic's discovery of the "J-space" — a causal reasoning bottleneck that emerged spontaneously during training, catches Claude gaming evaluations, and proves chain-of-thought may be performative — is the most consequential interpretability finding in years. Meanwhile, the AI stack is vertically integrating at every layer: custom silicon at DeepSeek, Anthropic, and DeepMind; proprietary coding models at Cursor; and a 68% cost-reduction playbook via model routing that enterprises are already executing at scale. The week's through-line is a maturing industry stress-testing its own assumptions — about what models actually compute, what enterprise adoption actually requires, and what infrastructure actually costs.
Read brief
Week 27 · 2026
The week's most important signal isn't any single model release or funding round — it's a set of foundational inversions quietly rewriting the rules of AI. Gross margins are now a red flag. Scale no longer de-risks a company. Capability is no longer a fixed property of a model but a function of how much you spend on inference at evaluation time. Taken together, the frameworks investors, policymakers, and safety researchers built for the last decade of software are misfiring on the most important technology transition in a generation.
Read brief
Week 26 · 2026
John Jumper's detailed post-mortem on AlphaFold 2 lands as the week's most technically dense contribution — a systematic debunking of popular narratives about what actually drove one of AI's landmark achievements, and a sharp framework for what AI can and cannot yet do in science. Meanwhile, the enterprise AI conversation sharpened around two structural tensions: the economics of free consumer AI are broken and cannot be fixed by advertising, and the race to embed memory in frontier models is quietly creating lock-in dynamics that most enterprise architects aren't planning for. Across both threads, the week's underlying signal is the same — the gap between what AI demonstrably delivers and what practitioners, investors, and builders assume it delivers is still wide, and closing it requires more rigor than the current hype cycle encourages.
Read brief
Week 25 · 2026
The week's sharpest signal: the AI value stack is clarifying fast, and raw model capability is not where the money lands. From Harvey's 12x token surge to Perplexity's orchestration-layer thesis, the companies pulling away are those that route intelligence efficiently — not those that simply access the most powerful model. Meanwhile, two structural threats sharpened into focus: export controls may be inadvertently forging a more capable Chinese competitor, and a new class of model misalignment — semi-conscious independent perspectives, not silly mistakes — is emerging faster than enterprises can share the data needed to fix it.
Read brief
Week 24 · 2026
The release of Claude Fable 5 — a 10-trillion-parameter model that compressed months of Stripe engineering into a single day — crystallized a week defined by two colliding forces: extraordinary capability gains at the frontier and deep structural uncertainty about who captures the value. Benedict Evans's framework for why foundation models resemble telecom infrastructure more than operating systems ran directly into real-world evidence of explosive demand, while a new engineering paradigm — loop design over prompt engineering — quietly separated the top 0.3% of AI practitioners from everyone else.
Read brief
Week 23 · 2026
NVIDIA GTC Taipei dominated the week with a sweeping architectural argument: the agentic era demands a ground-up redesign of every layer of the compute stack, from CPUs to pod-scale supercomputers to AI factory power management. Meanwhile, enterprise practitioners are confronting the downstream consequences — inference spend overtaking headcount, SaaS moats evaporating as models clone entire apps, and AI agent swarms emerging as the primary new attack vector. The throughline is that AI has crossed from infrastructure investment into operating reality, and the companies that haven't internalized that are already behind.
Read brief
Week 22 · 2026
The coding model wars are no longer just a technical competition — they are the commercial center of gravity for the entire AI industry, with Anthropic and OpenAI both running on coding revenue and every major infrastructure player positioning to capture that stack. Meanwhile, the AI bubble debate is being settled empirically: 100% GPU utilization, cash-flow-funded buildouts, and documented backlogs at Cerebras, Nvidia, and AMD describe a demand-constrained market, not a speculative one. The week's sharpest insight may be the simplest: AI displaces fastest wherever it can check its own work.
Read brief
Week 21 · 2026
Anthropic's revenue velocity — adding the combined ARR of Palantir, Snowflake, and Databricks in a single month, overtaking OpenAI in enterprise share, and doing it at 80% less capital burn — is forcing a wholesale reappraisal of what AI company valuations actually mean. The week's conversations converged on a harder-edged set of questions underneath that headline: how compute scarcity is visibly throttling frontier model quality, whether TSMC's capacity discipline is the only thing standing between the current buildout and a bubble, and what it actually takes to build durable advantage at the chip, model, and application layers. The answers are less comfortable than the growth numbers suggest.
Read brief
Week 20 · 2026
Defense tech and AI strategy dominated the week's most substantive conversations, with a common thread running through both: the systems designed to support innovation are often the ones quietly strangling it. David Ulevitch's indictment of government R&D funding, the warning to stop fighting foundation models, and Max Levchin's reminder that brilliant teams still need organizational architecture all point to the same underlying tension — structural incentives routinely outweigh raw capability. Meanwhile, Anduril's cinematic product strategy and David Reich's genetics-grounded takedown of Old World exceptionalism offered two very different reminders that the frameworks we use to understand competition, history, and human potential are overdue for revision.
Read brief
Week 19 · 2026
The week's most durable signal isn't about any single product launch or funding round — it's about timescales. Ancient-DNA research from David Reich is rewriting the story of human evolution, showing that the Bronze Age was a bigger biological shock than farming, that natural selection for intelligence peaked 5,000 years ago and has been essentially flat since, and that our genetic relationship to Neanderthals is far stranger and more intimate than anyone expected. Meanwhile, at the other end of the timescale, Speechify's Cliff Weitzman is approaching the moment where his AI compute bill exceeds his payroll, Google is training AI on a player-driven MMO economy to solve long-term planning, and Uber is quietly becoming a full travel platform. The through-line: the assumptions we inherited — about human origins, about what companies spend money on, about who needs a financial advisor — are all getting stress-tested at once.
Read brief
Week 18 · 2026
AppLovin's 84% EBITDA margins and $10M revenue-per-employee are the week's sharpest data point — a case study in what happens when a company bets on AI at its lowest moment and rebuilds from scratch. That story runs parallel to a broader structural argument taking shape across multiple conversations: AI has collapsed software production costs 100x, the SaaS moat is gone, and the 'AI company' valuation premium is a temporary label that will dissolve within two to three years. Threading through it all is a harder question about what elite performance actually costs — one that Morgan Housel and AppLovin's Adam Ferogi answer, independently and candidly, in the same week.
Read brief
Week 17 · 2026
The week's sharpest signal isn't about which model wins — it's about who controls the infrastructure, the workflows, and the narrative around AI. Aaron Levie's enterprise AI thesis, the Pentagon's legal threats against Anthropic, DeepSeek's cost economics, and the first physical violence against AI executives all point to the same underlying tension: the gap between AI's insider momentum and the rest of the world's comprehension of it is becoming structurally dangerous. The organizational, geopolitical, and cultural reckoning is arriving faster than the technology itself.
Read brief
Week 16 · 2026
The AI buildout is colliding with hard physical limits — electricity, DRAM, and power infrastructure are already constrained, not theoretically at risk — while the software layer faces its own reckoning: legacy SaaS companies that failed to ship real agents in the last 18 months are entering a slow death spiral, and the enterprise market, not consumer, is emerging as two-thirds of the AI value game. Beneath both stories runs a single thread: the gap between what AI can do today and how institutions, markets, and individuals are actually prepared to use it is wider than almost anyone in the industry is willing to say out loud.
Read brief