Featuring
Episode 135 starts with Wagner's dash toward Moscow, moves through the Supreme Court's affirmative-action ruling and Silicon Valley's AI buying spree, then finishes with startup fraud and a hum from the universe. The real fire is Sacks versus Jason on whether Prigozhin exposed a wobbling Putin or merely earned himself the world's worst severance package. Sacks has the best debate episode: his narrow calls on Putin's grip and MosaicML's strategic fit aged better, even if his polling confidence and the $1.3 billion price both deserve a raised eyebrow.
Spice rack
Did the Wagner mutiny seriously weaken Putin, or leave his regime broadly intact?
Original point: The mutiny was an embarrassing black eye, but no Russian power center joined Prigozhin and Putin emerged with society consolidated behind him.
What everyone argued
Jason Calacanis
Jason rejected the victory-lap interpretation. He called the march on Moscow an extraordinary sign of instability, argued that fear and state propaganda made headline approval numbers unreliable, and predicted that Putin would lose power within ten years because of illness or the Ukraine war.
David Sacks
Sacks described the episode as a real mutiny with 'coup optionality,' not a staged event. He argued that Prigozhin tested for elite support, found none, accepted a deal, and left Putin embarrassed but still in command; he cited roughly 80% approval and warned that regime change could produce a more dangerous hard-liner.
Winner circle
Sacks wins the narrow question. He treated the mutiny as real and damaging while correctly distinguishing a black eye from a successful fracture of Putin's command. Jason was right to challenge propaganda and polling certainty, but hindsight did not deliver the loss of control his argument leaned toward.
Commentary
Jason Calacanis
Assumptions and fact checks
Fear makes Russian approval polling essentially unusable.
Why it mattersRepression creates real measurement risk, but Levada used random telephone sampling and reported stable participation patterns; its own methodological work found no evidence that only government supporters were answering. The numbers deserve caveats, not automatic dismissal.
An armed challenge by a former Putin ally necessarily showed that Putin had already lost substantial control.
Why it mattersThe march exposed a severe lapse and elite conflict, but a failed challenge can reveal weakness without producing durable loss of command. The absence of elite defections and the rapid end of the mutiny cut against the stronger version of Jason's claim.
David Sacks
Sacks won by keeping his core claim narrow enough to survive hindsight: serious embarrassment, no successful elite break, continued control. His detour from regime stability to broad social enthusiasm asked the polling evidence to prove more than it could.
Assumptions and fact checks
Because no major power center publicly joined Prigozhin, Putin retained effective elite control.
Why it mattersPublic loyalty can hide private dissent, but coups require coordination and defections. The failed march, lack of an elite cascade, and subsequent containment of Wagner support the practical-control claim.
High approval polling and patriotic popular culture showed that Russian society was united behind the war.
Why it mattersThose signals support continued public acquiescence, not the stronger claim of authentic unity. Coercion, preference falsification, and large differences by age and media source remain material alternative explanations.
Russia's Defense Ministry required volunteer formations to sign ministry contracts by July 1, a demand Prigozhin rejected.
CheckA UK Foreign Secretary statement to Parliament recorded the June 10 contract order and Prigozhin's immediate refusal, supporting Sacks's proposed trigger for the confrontation.
Putin's polling was around 80% and did not collapse after the mutiny.
CheckLevada's post-mutiny survey found 76% trusted Putin and reported that the mutiny did not reduce his rating. This validates the approximate number as a poll result, while not proving that stated approval perfectly measured private belief under repression.
Was Databricks buying MosaicML a strategic AI-infrastructure deal or peak-cycle froth?
Original point: The price reflected strategic demand for a scarce part of the enterprise AI stack, so the acquisition was not irrational even amid a financing mania.
What everyone argued
Chamath Palihapitiya
Chamath warned that new sectors see their frothiest acquisitions when hype is highest and facts are scarcest. He compared the moment with optical networking and argued that early strategic urgency can produce prices that later fall, benefiting sellers while burdening incumbent shareholders.
David Sacks
Sacks argued that MosaicML occupied a scarce, strategic layer: efficiently training and customizing models for enterprises that wanted private data controls. He acknowledged a financing mania but said Databricks needed an end-to-end AI toolchain and that competing investors had already bid MosaicML's valuation sharply upward.
Winner circle
Sacks wins, with low confidence. He provided the more specific mechanism and the asset became a real part of Databricks' platform, so the strategic case aged well. Chamath still lands the essential finance caveat: no public evidence proves that a good capability was worth that exact price.
Commentary
Chamath Palihapitiya
Chamath correctly made price—not product quality—the burden of proof. To win, he needed a deal-specific estimate of replacement cost, revenue potential, or dilution instead of relying mainly on the history of other bubbles.
Assumptions and fact checks
Peak strategic M&A usually arrives early in a technology cycle, after which acquisition valuations fall.
Why it mattersThe pattern occurs in speculative cycles, but it is not a rule and does not price an individual asset. Platform fit, scarce talent, time-to-market, and the buyer's own equity can make an early premium rational.
MosaicML was meaningfully analogous to optical-networking acquisitions that later vanished.
Why it mattersThe analogy highlighted cycle risk but skipped the key mechanism: MosaicML's training and model-building capabilities were integrated and expanded inside Databricks' continuing Mosaic AI platform.
David Sacks
Sacks did the better deal-specific analysis because he named the scarce capability and why this buyer needed it. The missing piece was a sober valuation bridge from strategic importance to $1.3 billion.
Assumptions and fact checks
Owning the model-training layer was strategically important enough for Databricks to pay a large early-cycle premium.
Why it mattersThe subsequent integration and continued Mosaic AI product line support the strategic-fit mechanism. They do not prove the exact price, but they show Databricks bought a capability it actually deployed.
Enterprise reluctance to send proprietary data to a third-party model provider would create durable demand for customized models and governed AI tooling.
Why it mattersDatabricks continued to build model serving, governance, vector search, and enterprise AI tools around that premise through 2026. The market also evolved toward using both private and hosted third-party models, so the strongest 'every enterprise builds its own model' version would be overstated.
Databricks acquired MosaicML for approximately $1.3 billion and integrated its generative-AI model-building capabilities.
CheckThe completion announcement identified MosaicML as a generative-AI platform, and Databricks later said the acquired model-building capabilities had been entirely integrated and substantially expanded within Mosaic AI.
Jason had the strongest objection to Sacks's evidence: a chart hit and a patriotic song cannot by themselves prove authentic public unity. He would have made the case much stronger by separating 'this was a dangerous crack' from the harder claim that the crack had already weakened Putin's operational grip.