Featuring
The four besties opened their 2026 prediction ledger with California's proposed wealth tax, then sparred over Trump's use of force, AI's effect on junior careers, and the deal structures shaping the AI race. The sharpest exchange belonged to Jason and Sacks over Venezuela; Jason had the stronger labor-market receipt, while Chamath's licensing thesis has aged best so far. A prediction show with actual scorekeeping—dangerous, but refreshing.
Spice rack
Did Trump's Venezuela operation amount to neoconservative regime change?
Original point: Trump was behaving like a neocon by using force in Venezuela and Iran and threatening action elsewhere, even if the Venezuela raid itself went flawlessly.
What everyone argued
Jason Calacanis
Jason argued that the label should follow the willingness to use military force and accept escalation risk, not only the presence of a long occupation. He praised the operation but warned that captured or killed U.S. troops would have produced a radically different judgment.
David Sacks
Sacks defined neocon regime change by three linked mechanisms: invasion, occupation, and nation-building. He argued that the Maduro capture had none of those durable commitments and therefore belonged to a different, limited paradigm.
Winner circle
Sacks wins narrowly on the question actually posed. A limited military strike and capture can be dangerous and legally contested without yet becoming the invasion-occupation-nation-building model that gives neocon regime change its distinctive meaning. Jason was right about ex ante risk and wrong to let the political label do more work than his mechanism.
Commentary
Jason Calacanis
Assumptions and fact checks
A short military intervention aimed at removing a foreign leader is enough to call the policy neoconservative even without occupation or nation-building.
Why it mattersThat is a defensible broader use of the label, but it makes the term cover nearly any coercive intervention and does not answer Sacks's narrower mechanism-based definition.
The operation's non-zero risk of a disastrous outcome should weigh heavily in judging the policy despite its actual success.
Why it mattersDecision quality should be judged from ex ante risks as well as outcomes. Jason was right to resist pure outcome bias, although he offered no evidence that his worst-case scenarios were equally likely.
U.S. forces used military force inside Venezuela to capture Maduro.
CheckThe administration's own statement says U.S. Armed Forces conducted targeted and limited military strikes inside Venezuela to support Maduro's apprehension; later Justice Department material names the mission Operation Absolute Resolve.
David Sacks
Sacks won the classification question by supplying a clear mechanism and matching it to the observed policy. He would have made the case sturdier by conceding that limited strikes are still hostilities and then explaining why their scope and exit conditions matter more than the mere use of force.
Assumptions and fact checks
Neoconservative regime change requires invasion, occupation, and nation-building rather than merely force used to remove a ruler.
Why it mattersThose mechanisms capture the costly, open-ended feature that made Iraq and Afghanistan the relevant comparison. The definition may be narrower than common political usage, but it is analytically useful.
Because the raid succeeded quickly, Jason's escalation concerns no longer bear on whether the choice was sound.
Why it mattersA successful outcome does not erase the probability or cost of failure at decision time. Sacks needed to compare those risks with the risks of inaction instead of waving them away.
The Venezuela operation lasted three hours.
CheckThe closest official timing says U.S. forces were on the ground for about two hours. The broader operation may have lasted longer, so Sacks was directionally close but too precise without defining his clock.
U.S. forces did not remain on the ground in Venezuela after capturing Maduro.
CheckSecretary Rubio said there were no U.S. forces left on the ground and described their presence during the capture as about two hours.
Is AI or a cultural shift doing more damage to entry-level hiring?
Original point: Companies find it easier to automate junior tasks with AI than to train new graduates, removing the bottom rungs of white-collar career ladders.
What everyone argued
Jason Calacanis
Jason said companies are automating the bottom tier of tasks typically assigned to new graduates, making entry-level white-collar work harder to obtain. He conceded that motivation, COVID-era disruption, and family wealth could also matter, then argued that AI fluency will divide young applicants.
David Friedberg
Friedberg relayed CEO anecdotes that firms still hire juniors but see weaker preparation, motivation, and executive function among recent graduates. He argued the employment problem may reflect COVID-era and cultural changes as much as automation, while stopping short of denying an AI effect.
Winner circle
Jason wins, but only after accepting Friedberg's multicausal correction. The best available evidence now shows a distinct early-career contraction in AI-exposed work, which makes automation more than a convenient excuse. Friedberg was right to resist monocausality; his cultural alternative simply arrived with weaker receipts.
Commentary
Jason Calacanis
Jason improved his case by conceding multiple causes instead of defending an AI monocause. The strong version is about task substitution hitting juniors first; the weaker motivational sermon risks blaming workers for a labor-market structure he had just described.
Assumptions and fact checks
AI automation is a material cause of the missing bottom rungs in white-collar career ladders.
Why it mattersThe age-by-exposure pattern is consistent with this mechanism and has strengthened since the recording. It is not a complete causal decomposition, so 'material cause' is more defensible than 'primary cause.'
Young people who use AI tools will generally be able to find work.
Why it mattersAI-relevant education is associated with better early outcomes, but tool fluency cannot guarantee employment when the number and design of junior roles are also changing.
Young workers in AI-exposed occupations have experienced a distinct employment decline.
CheckStanford's June 2026 update found employment for ages 22–25 in AI-exposed occupations contracting 3.8% annually since ChatGPT, versus 2.0% growth in the least-exposed occupations, in its balanced ADP sample.
David Friedberg
Friedberg supplied the necessary causal caution, but his cultural diagnosis leaned heavily on elite-network anecdotes and generational generalization. He would have been stronger separating measurable preparation gaps from labels such as entitlement or weak temperament.
Assumptions and fact checks
A broad decline in graduate motivation and preparedness is a major cause of weaker junior hiring.
Why it mattersThe anecdotes are plausible but do not measure the trend or distinguish changed workers from changed employer standards. Friedberg named a hypothesis, not evidence strong enough to rank causes.
Because the weakness began before ChatGPT, AI cannot be the primary current driver.
Why it mattersA pre-existing decline and a later AI acceleration can both be true. The timing evidence weakens monocausal claims but does not settle the size of AI's incremental effect.
Employment deterioration in AI-exposed jobs began before ChatGPT's late-2022 release.
CheckA 2026 study using unemployment-insurance records and LinkedIn profiles found risk and entry gaps opening earlier in 2022. That supports pre-existing forces, though it does not rule out additional damage after generative AI adoption.
AI has not yet changed employment at most firms.
CheckA Stanford evidence review reports that only 5% of firms in Census data saw any AI-related employment effect, split evenly between gains and losses; the aggregate result can coexist with concentrated harm in junior, exposed roles.
Will the next giant AI deal be an acquisition or a licensing workaround?
Original point: One of the largest technology companies will attempt a $50-billion-plus acquisition of a frontier AI company such as Anthropic, Perplexity, or xAI.
What everyone argued
Chamath Palihapitiya
Chamath agreed that enormous transactions were coming but argued their legal form would be IP licenses, minority investments, and talent transfers. A conventional frontier-model acquisition would face years of review across several regulators, while a licensing structure could put technology and key people to work immediately.
Jason Calacanis
Jason predicted that cash-rich Mag 7 companies would try to buy a frontier AI company outright and argued that a deal-friendly Trump administration could accelerate approval. He treated the strategic need to keep pace in models as stronger than the regulatory obstacle.
Winner circle
Chamath leads on mechanism and current evidence. The market keeps finding ways to transfer talent, compute, and technology without buying the whole company, which is exactly the workaround he described. Because 2026 is not over and a blockbuster bid could still arrive, this is a provisional win rather than a victory lap.
Commentary
Chamath Palihapitiya
Chamath had the cleaner causal model: regulators care about control, firms care about speed, and lawyers search for a structure that transfers enough value without a full merger. His overreach was declaring all traditional M&A dead when his evidence was specific to strategically sensitive AI assets.
Assumptions and fact checks
Global antitrust review makes a conventional acquisition of a frontier AI lab commercially impractical even when the buyer can afford it.
Why it mattersThe Activision timeline and scrutiny of quasi-acquisitions support the mechanism. 'Impractical' is more credible than 'impossible,' because deal-specific remedies and political priorities can change the result.
Licensing and talent-transfer agreements can reliably reproduce the strategic value of ownership.
Why it mattersThey can deliver talent and technology quickly, but non-exclusive rights, governance gaps, partner dependence, and later regulatory reclassification leave meaningful value outside the buyer's control.
Microsoft's Activision Blizzard acquisition took about two years to complete.
CheckMicrosoft announced the transaction on January 18, 2022 and completed it on October 13, 2023—about 21 months—after regulatory review and restructuring.
Nvidia used a non-exclusive technology license with Groq while Groq remained independent and key Groq leaders joined Nvidia.
CheckGroq's own announcement describes exactly that structure: a non-exclusive inference-technology license, continued independent operation, and the transfer of its founder, president, and other team members to Nvidia.
Jason Calacanis
Jason saw the appetite but skipped the plumbing. His strongest version is that strategic desperation eventually overwhelms process; to win it, he needed a plausible buyer-target pair and a path through every regulator, not just confidence that Trump likes deals.
Assumptions and fact checks
A deal-friendly U.S. president can make a global frontier-AI acquisition close quickly.
Why it mattersU.S. enforcement posture matters, but it cannot by itself dispose of review in Europe, the United Kingdom, China, or other relevant markets. Chamath directly identified this coordination problem.
Competitive pressure will make at least one frontier lab accept a conventional acquisition rather than remain independent or sign partnerships.
Why it mattersThe pressure is real, but high valuations, existing strategic investors, governance structures, and regulatory exposure all cut against a clean sale. The 2026 prediction window remains open.
Large technology companies have enough capital and strategic incentive to pursue major AI transactions.
CheckThe scale of the Nvidia-Groq licensing arrangement, Meta's major Scale investment, and Amazon's expanded long-term Anthropic compute partnership demonstrate both capacity and strategic appetite, though not Jason's predicted acquisition form.
Jason's best move was separating a lucky or skillful outcome from the risk accepted beforehand. His weakest was treating 'they used force' as nearly sufficient proof of neoconservatism, which left Sacks room to win on definitions rather than defend every part of the policy.