Every time frontier AI models get cheaper, consumption rises faster than anyone predicts. The pattern has a name, it has historical precedent, and it is now the central fact of AI economics.
Jesse Zhang of Decagon is making a specific, checkable bet: the deployment infrastructure that enterprise teams spend months configuring today will be something agents assemble on their own within a few years. The implications for the services businesses built around that complexity are significant.
The framing around decentralized finance's revival deserves scrutiny. Several of the sector's headline trends look weaker on close inspection, and the risks drawing the least attention may be the ones that matter most.
The dominant architecture in AI does not dominate everywhere. At small data budgets and in constrained domains like Go, older network designs hold their ground. The more unsettling finding is what that implies about scale as the real variable.
The venture capitalist behind 20VC has made a specific, falsifiable call: at least one American company, valued in the hundreds of billions or possibly a trillion dollars, will be built around American-first open-source AI. The reasoning behind that bet, and what would have to be true for it to land, deserves a hard look.
The framing around decentralized finance's revival deserves scrutiny. Several of the sector's headline trends look weaker on close inspection, and the risks drawing the least attention may be the ones that matter most.
Payments, lending, contracts, and ownership transitions are all hitting inflection points at roughly the same moment, in sectors long considered immune to this kind of pace. The evidence is coming from inside those sectors, not from the outside looking in.
The framing around decentralized finance's revival deserves scrutiny. Several of the sector's headline trends look weaker on close inspection, and the risks drawing the least attention may be the ones that matter most.
Ray Dalio's case against central bank Bitcoin adoption does not rest on price, volatility, or regulation. It rests on what Bitcoin fundamentally is: a public ledger that cannot be made private. That argument is either right permanently or wrong the moment a major central bank buys in.
Payments, lending, contracts, and ownership transitions are all hitting inflection points at roughly the same moment, in sectors long considered immune to this kind of pace. The evidence is coming from inside those sectors, not from the outside looking in.
Whether the target is a customer, a recruit, or an influencer audience, the instinct to reach more people faster tends to produce worse results. Three practitioners are arriving at the same corrective, from very different angles.
The standard scaling playbook says growth requires management layers. Tom Verrilli of Whatnot says that assumption is wrong, and he has a pointed cost argument to back it up. The case for fewer, better-paid senior individual contributors is worth taking seriously.
The venture capitalist behind 20VC has made a specific, falsifiable call: at least one American company, valued in the hundreds of billions or possibly a trillion dollars, will be built around American-first open-source AI. The reasoning behind that bet, and what would have to be true for it to land, deserves a hard look.
Stanislas Polu, the founder of Dust, walked away from OpenAI holding a striking illustration of how fast AI lab valuations have moved. The stock options he gave up were worth more than the entire valuation of his own company at its Series B, just six months earlier.
The venture capitalist behind 20VC has made a specific, falsifiable call: at least one American company, valued in the hundreds of billions or possibly a trillion dollars, will be built around American-first open-source AI. The reasoning behind that bet, and what would have to be true for it to land, deserves a hard look.
President Trump described US support for the yen as an act of friendship. Economist Ed Yardeni offered a colder read: Japan holds a trillion dollars of US government debt, and Washington could not afford to let Tokyo become a forced seller.
Isaiah Taylor flagged a hard date buried in an executive order: three advanced reactors had to go critical on American soil by July 4th. The U.S. Department of Energy has now confirmed it happened, marking the first time modern advanced reactor technology has cleared that threshold on U.S. soil under a presidentially mandated deadline.
Cancer research has long centered on genetic mutations. A growing body of work from researchers across oncology, mitochondrial biology, and aging science points toward metabolic collapse as the more fundamental driver, with implications that extend from cancer treatment to psychiatry.
Civil War historian Gary Gallagher makes a pointed case: Lee almost certainly wins Gettysburg if he simply stops advancing after day one and forces the Union to come to him. The argument cuts against a century of received wisdom about why the battle turned out the way it did.
Isaiah Taylor flagged a hard date buried in an executive order: three advanced reactors had to go critical on American soil by July 4th. The U.S. Department of Energy has now confirmed it happened, marking the first time modern advanced reactor technology has cleared that threshold on U.S. soil under a presidentially mandated deadline.
Two apps, similar view counts, wildly different revenue. The gap between Wrestle AI's $17,000 first month and Green's $35 is not an anomaly. It is the operating logic of the current creator economy, and it has consequences for how anyone building an audience should think about scale.
The encryption of reasoning traces provides no actual security because the model itself will readily reveal the hidden thoughts.Ilia Shumailov
22 Aug
The incentive structure in crypto makes buying memecoins more rational than supporting real projects, because serious projects must downplay their tokens.Erik Voorhees
22 Aug
Third-party AI red teams self-censor findings to preserve future access to AI developers, prioritizing being invited back over honest reporting.Nathan Labenz
The defining line between attitudinal and behavioral data is whether the decision has real stakes.Joon Sung Park
21 Aug
Bearish analysts are more disingenuous than bulls: they never acknowledge upside risk and continually shift their arguments while maintaining a permanent bearish stance.Michael Santoli
21 Aug
In B2B advertising sales, pitching ROI does not close seven-figure deals; buyers are driven by trust and budget-safety logic ('spend it with me, it's safe') rather than ROI calculations.Sam Parr
21 Aug
The Treasury's yield-curve operation is 'soft QE', directionally money printing even if not pure QE.Ryan Sean Adams