The four All-In co-hosts work through Nvidia's blowout quarter and Salesforce's 20% jump, then spend the middle of the show on the bond market — Scott Bessent's buybacks, Stan Druckenmiller's Wall Street Journal op-ed opposing them, and the $10 trillion of federal debt that has to be refinanced in the next twelve months. The back half turns into an argument about whether Mr. Druckenmiller cheated by writing that op-ed with AI, a read on the Ukraine war, and a long explainer on how Moderna's cancer immunotherapy actually works.
Host: Jason Calacanis
Also on: Chamath Palihapitiya, of Social Capital; David Sacks, of Craft Ventures; David Friedberg, who runs a plant-breeding company that does its own DNA sequencing
Published: 28 August 2026 on the All-In feed
Watch on YouTube | Episode page | 1 hr 37 min
✅ Time saved: 1 hr 6 min
Key Takeaways
Nvidia guided to 70% growth next year against a street expecting 45%
Revenue of $96.2 billion in the quarter, up 106% year over year, against a $92 billion consensus
"Nvidia is only trading at 12 times earnings" — David Sacks, after the print
The SaaSpocalypse trade broke on the horizontal names and is still live on the vertical ones
Salesforce beat adjusted EPS by 80%, $5.90 against $3.27, and raised full-year guidance to $46 billion
Chamath Palihapitiya and Mr. Friedberg both say they are short vertical SaaS; Mr. Sacks says it is case by case
Benioff is letting Claude become the front end of Salesforce on purpose
"Benioff here is willing to risk disintermediation in order to ride this AI wave because he knows it's what the customer ultimately wants." — David Sacks
Nothing the Treasury Secretary does moves the 30-year against $10 trillion of refinancing
Mr. Friedberg puts Mr. Bessent's maximum buying authority at about $1 trillion against $10 trillion he has to sell
The federal government's average cost of debt is 3.4% and the 30-year is at 5.2%
Each 1% move in rates costs 1.25% of GDP a year in extra interest, per Mr. Friedberg
Spending cannot be cut because no single person controls it
435 House members, 100 senators, no line-item veto, and a media that Mr. Sacks says will not report the story straight
The real break is 2030 to 2032, when Social Security and the states run out
Mr. Friedberg expects Social Security restructuring and state bailouts driven by unfunded pensions in that window
Government spending is the root of inflation, and every program raises the price of what it subsidizes
Mr. Friedberg on student loans: administrative costs up 6x, tuition compounding at 8% a year for 30 years
AI is the only growth story big enough to outrun the debt
"We are kind of counting on AI as the deus ex machina to grow our way out of this thing." — Chamath Palihapitiya
Three of the four co-hosts see no problem with Druckenmiller using AI to write his op-ed
Mr. Sacks issued a standing disclosure that everything he publishes has been through AI
Mr. Calacanis calls it the lip-syncing of writing and wants it disclosed
Moderna's cancer therapy is a decades-old technique, not a new drug
Market cap went from $20 billion to $60 billion on the readouts; clinics already do the same process for $50,000
Science Punishes Heterodoxy, and That Is Why It Has Stagnated
Mr. Friedberg opens on the two interviews he had just done, with Eric Weinstein and with Kratsios, which he calls two sides of the coin of science in America. Asked who Mr. Weinstein is, he describes a scientist who worked with Peter Thiel, had a podcast for a while, called string theory bullshit, and proposed an alternative unification theory — and who was cast out of the scientific community for it.
The mechanism Mr. Friedberg names is funding, not ideas: if you do not follow the mainstream you do not get grant funding, tenure, jobs, or invitations to lecture
Mr. Sacks says the substance holds up even if the language offends: until string theory is proved, it is unproved, and Mr. Weinstein is more right than wrong on that point
Mr. Calacanis compares it to Jay Bhattacharya being told the lab-leak hypothesis was misinformation, when everyone was working from theory about where COVID came from
Mr. Palihapitiya says exclusion is what has stagnated American science: everyone thinking the same way means no heterodox thinking and nothing new being discovered. He puts it as a comparison — there is probably a lot of incrementalism in 2026 that did not exist in 1926
Mr. Friedberg's example is general relativity, which was completely outlandish when published and was still studied and assessed rather than dismissed
China Is Behind on AI in Every Category Except Optimism
The panel watches clips from a robot Olympics held in China — humanoids sprinting, one of them losing a leg mid-race — and spends several minutes on it as banter, including a running joke about the "Friedberg Sprint" and a clip of the poker player Jason Koon setting records at 42.
Mr. Calacanis reads it as propaganda: the Chinese Communist Party filled a stadium with people cheering for AI and robotics, while in the United States the message is that all jobs are over
Mr. Friedberg pushes back that in a country of a billion people, finding 4,000 spectators is not evidence of a grand unifying theory
Mr. Sacks says the polling backs the sentiment gap: "It is true that China is much more optimistic about AI than we are. AI optimism in China is over 80%" on the question of whether AI will be more beneficial than harmful, against roughly 30% in the United States
"the U.S. is ahead of China in every category except one, which is optimism" — David Sacks, who calls American pessimism the biggest risk to winning the AI race
Mr. Calacanis calls the whole spectacle a psyop designed to make viewers feel bad for the robots
Optimus Would Win Half of Those Events, and Agent Swarms Beat One Agent
Mr. Palihapitiya says he has seen unreleased Optimus footage from Elon Musk and that it would already win half the events at the robot Olympics. He says he asked whether it was CGI and was told it was real, and that he does not want to say more
Mr. Sacks lays out where humanoids still fail: you can train them for a specific action like a race or a jump, but the hard part is a condition they were not programmed to expect, and physical interaction — folding laundry, picking something up without breaking it
On Grokbot, which Mr. Sacks says he has been using and has already hit the usage limit on at $200 a month: "What's great about Grokbot is that it's always on. It's in the cloud." Earlier agent harnesses such as OpenClaw or Hermes sat on the desktop and stopped when the computer slept, which is why people were running Mac minis
Multiple agents beat one agent because they specialize: Mr. Sacks says agents in a swarm develop more context and expertise and get better at different things
Mr. Calacanis says he has pitched a multiplayer mode — two agents plus two people from your team in the same room, humans in the loop — as the thing that would make it go viral
Nvidia Guided to 70% Growth When the Street Wanted 45%
Mr. Calacanis read the quarter off the docket, so the figures below are the show's own research rather than any co-host's claim.
Revenue of $96.2 billion in the quarter, up 106% year over year, against a Wall Street expectation of $92 billion
Guidance of 70% growth next year, against the roughly 45% the street was carrying — "they guided to 70% growth next year, whereas the street was only expecting around 45%", Mr. Sacks said, calling that rather than the quarter itself the reason for the move
Mr. Calacanis added that the 70% would have been higher but for supply constraints
$60 billion in net profit, which Mr. Calacanis called the most profitable core business quarter of any public company ever, with Google and Amazon having had higher one-off quarters
Mr. Sacks put it as almost $100 billion of revenue and $60 billion of profit at 75% gross margins, the most profit ever generated by a company in history excluding one-off items
Mr. Calacanis: a billion dollars a day in revenue
Nvidia is the world's most valuable company at $5.5 trillion; Polymarket had a 79% chance it ends the year as number one on $6.3 million of volume
Heading into earnings the stock was only up 12% year to date, on par with the S&P 500 and lagging the broader chip index, which Mr. Calacanis attributed to the belief that competition is coming
The takeaway Mr. Sacks drew is that the capex-bubble narrative got shredded: Nvidia's numbers and forecast say the AI build-out continues well into the future, and "Nvidia is only trading at 12 times earnings"
The SaaSpocalypse Was a Vertical Problem, Not a Horizontal One
Mr. Palihapitiya had called Salesforce oversold on 15 May, and the show played the clip back. The stock was up over 20% on Thursday after revenue of $11.3 billion, up 11% year over year and in line, an adjusted EPS beat of 80% at $5.90 against $3.27, and full-year guidance raised to $46 billion. Mr. Calacanis noted the company is a major investor in Anthropic.
Mr. Palihapitiya frames it as AI moving into a third phase. If the first phase was models, the second is harnesses and agents — "if the first phase of AI were models and think of like a model as a brain, the second phase of AI is harnesses", giving that brain eyes, hands, a notebook for memory and a keyboard
The third phase is context: training that autonomous agent to actually be a lawyer, a customer service rep or a sales agent, which takes a large amount of contextual information
The companies holding that context are the large systems of record, which is why he says Salesforce's net dollar retention, revenue and guidance are all strong. He put the move at about 43% since he said it had bottomed
Mr. Friedberg's own story is the evidence. About a year ago his company built an internal CRM over a weekend with Claude Code and Cursor, then kept discovering it needed features, accounts, security, access control and a protected data repository
Meanwhile the company had competing demands for engineering time on plant-breeding software, the thing unique to it, and tools for people in the lab and the greenhouse
The conclusion was that the best return on time and dollars is in the workflows unique to your organization, not in rebuilding Slack, Gmail or a CRM. Nobody is going to rebuild Excel
Mr. Benioff kept after him — a best salesman texting daily, "Are we up yet?" — until Mr. Friedberg let him send someone to the office. He recounted that story on the Salesforce earnings call
"It's more of a vertical SaaSpocalypse, which is software that's designed for just one vertical." — David Friedberg, who says that is where things get blown up
Mr. Sacks says the whole narrative was overdone: "I do not believe that core systems of record like CRM are going to get ripped and replaced with something vibe coded." Enterprises want certainty, compliance and professionally managed software, and twenty years of filed bug reports against Salesforce are the hard part to get right
Mr. Palihapitiya's view is that vertical SaaS has no system of record at all — it is workflows and process. The horizontal monoliths he named as hard to replicate were Salesforce, Workday, Oracle for the general ledger, SAP and IFS. Asked whether he is long or short vertical SaaS, Mr. Sacks answered that it is specific to the company and its moats
Mr. Palihapitiya's reading of the bearish trade: one of Leopold's major positions was shorting SaaS including Salesforce, and he suspects it was a leverage trap rather than a thesis — a prime broker demanding liquid shorts against a big net long, producing Adobe and Salesforce, with the narrative worked backwards from there
Mr. Sacks' one takeaway from the day: "simple extrapolations of the future based on these trends don't work out". The simple extrapolation was that AI agents can code, therefore all software goes to zero. He put the job-loss narrative in the same category — AI can do knowledge work, therefore knowledge workers all lose their jobs — and said the answer in both cases is adoption: if you adopt the technology, you survive
Benioff Gave Claude the Front End to Stay the System of Record
The Anthropic deal is not mainly about offering Anthropic's models inside Salesforce. Mr. Sacks says the critical part is the reverse: Salesforce is being integrated into Claude, so Claude gets the primary user relationship and access to the data, the workflows and every action a user can take
"Benioff here is willing to risk disintermediation in order to ride this AI wave because he knows it's what the customer ultimately wants." — David Sacks
The reason it is not suicidal, in his telling, is that agents still need a canonical source of truth to read from and write back to, with no probabilistic variability around the data
The upside Mr. Benioff described to him is unlocked functionality. Mr. Sacks says he spoke to him that morning and the underestimated point is "there's a lot of trap value in Salesforce" — a broad, rich platform whose average user never finds most of it, but whose agent will, and will surface those actions unprompted
Mr. Palihapitiya's read: that codifies Salesforce as the system of record
An agent that knows Salesforce perfectly also replaces the team of highly priced consultants hired to figure it out, and once a user trusts it they click always-allow
The instruction to SaaS founders is to build for agents, not users: great APIs and a great command line interface, because that is how agents will relate to the tools. "You have to build the best agent interface." Founders also have to relax the idea of fully controlling the customer relationship, because most of the agent activity will happen outside their product
Mr. Calacanis' own version of this: he told Mr. Benioff at Davos that his team wanted to leave Slack because he could not get his data out, and after being put on with the sales team he now has full access. Salesforce has been rolling out a product called Slack Today, which he rates a 7 out of 10 and says is improving about 10% every couple of weeks
His view of OpenClaw is that innovation stalled after Sam Altman hired its founder, Peter, into OpenAI — he was explicit that he is not alleging anything nefarious, only that it lost its edge
Nvidia Is Buying Its Way Into Open Source, and Every Company Will Do Everything
Mr. Calacanis: Nvidia bought Hugging Face for $12 billion and did an acqui-hire of Poolside for $6 billion — about $20 billion of acquisitions, far less than 1% of its market cap — and now owns the distribution layer for open source. Poolside makes the coding model Laguna
Mr. Palihapitiya was not sure what was actually acquired: whether the roughly 100 engineers going to Nvidia work on Poolside's model or on its on-premise coding harness, and said the announcement was not clear to him
Mr. Sacks noted the deals are reported rather than confirmed
Mr. Calacanis' theory is that Sam Altman poked the tiger. After the $100 billion Nvidia deal in 2025, Mr. Altman announced the AMD deal and his own inference chip — Jalapeño, as Mr. Palihapitiya named it — and Jensen Huang responded that Nvidia does not have to make those OpenAI investments
Mr. Palihapitiya's generalization is convergence. The clean separation of customer and supplier is melting away: hyperscalers spin their own silicon, so it is entirely appropriate for Nvidia to spin its own models, host them, sell APIs and inference, and become a cloud provider
Nvidia still relies meaningfully on the hyperscalers — he put it at $24 billion in the quarter — but has built a neocloud business he says is now equivalent in size
"Every company is going to do everything." In five years, he says, the big companies will each have their own cloud, models, silicon and data centers, and the competition will be about which is better — which he argues is better for everyone
Nothing Bessent Does to the 30-Year Survives $10 Trillion of Refinancing
The setup from Mr. Calacanis: the 30-year Treasury hit a 19-year high at 5.3% last week; Mr. Bessent doubled the Treasury's long-dated buybacks from $2 billion to $4 billion on 19 August; CNBC reported he is considering ramping up bond buying and Fox Business that he wants to put the fear of God into traders shorting long-dated bonds. Mr. Druckenmiller, his former mentor, published a Wall Street Journal op-ed opposing the plan, arguing that the problem is spending rather than prices.
Mr. Friedberg starts with the relationships: Mr. Bessent and Mr. Druckenmiller worked closely together, and Kevin Warsh was in Mr. Druckenmiller's office every day for the last several years. He calls it a triumvirate — one running market activity, one the Treasury, one the Fed — and says that degree of closeness in those three positions is unprecedented. Mr. Sacks asked who is Caesar, who is Pompey and who is Crassus
The yield-curve explainer he gives the audience: around 2020 the whole curve was sub-half a percent out to the three- and five-year, and the 30-year could be had at 1.7%. The government should have rolled all its debt into 30-year paper then, but the debt is spread across the curve
Today the one-month costs the government about 3.8% and the 30-year 5.2%
He attributes the level to persistent inflation from excess government spending, and says the trap is that cutting aggressively now would hit unemployment and cause a recession because the federal government is such an intricate part of the economy
The deficit this year will be roughly $2 trillion. Average cost of debt across the $40 trillion outstanding is 3.4%. Every 1% change in the interest rate costs the government 1.25% of GDP a year in extra interest
"over the next 12 months, they have to refinance $10 trillion of debt" — David Friedberg. Those bonds come due, the principal has to be repaid, and the money has to be raised again at today's rates
"there is no action that Bessent can take that's actually going to have a meaningful effect on the long end of the curve" — David Friedberg. Maximum buying authority is about $1 trillion against $10 trillion he has to sell
Mr. Friedberg reads the op-ed as cover rather than an attack. Whether AI wrote it or Mr. Druckenmiller did, he says the point of the note is correct and its function is to tell the market that the yield curve is not the Treasury Secretary's fault or responsibility — it is Congress's and the president's
Asked by Mr. Calacanis whether the exchange is kayfabe, he said he does not think the three are allowed to talk to each other, and that this is Mr. Druckenmiller signaling to the market. "He can't be Atlas. He can't lift this earth."
Yield Goes Up, Trust Goes Down — and the AI Build-Out Has No Government Balance Sheet Behind It
Mr. Palihapitiya explains the tactic: buying bonds bids up prices and pushes implied yields down, so new issuance is marked against a 5.1% print rather than a 5.3% one. He calls it yield curve suppression, aimed at the refinancing tsunami Mr. Friedberg described
The structural problem is arithmetic: debt growing at 7% against GDP growth between 2% and 4%. He calls it a congressional problem rather than a partisan one — both sides consistently spend more than they should
Mr. Druckenmiller's prescription is cutting entitlements, and Mr. Palihapitiya says he is probably right; the question is when, and who forces Congress to confront it. His answer is the bond market, because that single price reflects faith in the United States
"Yield goes up, trust goes down." — Chamath Palihapitiya
"if you see the 30 year at 6%, it is the beginning of a death spiral" — the same speaker, who added that it would not be immediate and that the pain would last years
The timing is what makes it awkward. A century ago the US government was the balance sheet behind an industrial build-out; today it cannot be. "these guys are putting the entire U.S. economy on their back" — Mr. Palihapitiya on Nvidia, Google, Microsoft, Meta and Amazon, against what he called the chirping on the internet about Nvidia's balance sheet
He credits the president with finding $2 trillion on the tail end of the trade deals to pump into the build-out, but calls it a short-term band-aid, because that money comes from other nation states whose processes are long and meandering
Spending Cannot Be Cut Because No One Person Controls It
Mr. Sacks calls it a tragedy of the commons: 435 House members, 100 senators and a president, with the power of the purse originating in the House. If a couple of people could sort it out in a room they probably would; when everyone has an incentive to protect their own programs it becomes very hard
The president has no line-item veto and, "despite what the resistance liberals say, he's not a king"
On DOGE: Mr. Sacks says the president is the only one ever to back something like it, and that Elon Musk got as far as cutting one sub-agency before the hysteria — accusations of killing millions of children, Tesla dealerships firebombed
His second cause is the press. Reform needs massive public pressure, and he says the media does nothing to report the story accurately and no one is trying to create that pressure. Mr. Friedberg later named Nick Shirley and 60 Minutes as the kind of thing needed to call the spending and fraud out
He does not think a change of control fixes it: the energy in the Democratic Party is behind the large new spending programs the DSA wants, so voting for change that way means more spending
Mr. Palihapitiya disagrees that public pressure is the route at all — he says it has to be an acute moment. Socialism, he argues, is not actually available as a choice, because American debt is interconnected with domestic and international holders and there is no button to push that does not bankrupt the country overnight
The structural changes that have mattered in their lifetimes came from 9/11, the global financial crisis and COVID, and he expects the same shape here: a fiscal wall, something badly broken, then a change in expectations
Mr. Sacks' rejoinder is that each of those crises ended in a power grab by the government
The Real Break Is 2030 to 2032, When Social Security and the States Run Out
Mr. Calacanis asked for a timeline and a breaking point, noting the last person to balance the budget was Clinton and asking whether the trigger is a lender demanding 12%.
Mr. Friedberg's answer is that a democracy breaks toward a vote, not a default. Between 2026 and the 2028 cycle, persistent inflation and worsening affordability push people toward whoever promises a solution, and he sees no winning message that says the government will spend less
He expects that to favor candidates and principles he describes as socialist, which he considers damaging to individual liberties and to what makes a free market work
His central claim on inflation: deficit spending correlates closely with the cost of housing, health care, education and everything else. "If we went back to 2019, we would actually have a budget surplus right now as a country." The economy has grown enough since then that 2019 spending levels would leave the federal government paying down debt
Emergency COVID measures were normalized into permanent spending cycles, the same thing that happened with the 2008 stimulus
The second and larger break is 2030 to 2032, when Social Security has no money left to pay and, in his view, a number of states cannot meet their obligations because of unfunded pensions. He expects both a Social Security restructuring and state bailouts in that window
Mr. Calacanis said bailing out California would be cataclysmic; Mr. Palihapitiya added New York
Every Government Program Raises the Price of the Thing It Subsidizes
The tangent that follows starts with California's high-speed rail and becomes Mr. Friedberg's general case against program spending.
Mr. Calacanis calls the California railway a project with no track that has consumed tens of billions, and says a quarter-trillion-dollar train from San Francisco to Fresno that costs more than an airline ticket should be shamed out of existence. Nobody, he says, is even saying to turn it off
Mr. Friedberg's counter-example is Brightline in Florida, built by a private company at a fraction of the price: if this were a good idea, private industry would have built it
The pattern he draws across programs: when the government underwrote student loans, administrative costs went up 6x and tuition compounded at 8% a year for 30 years. The same in housing, where young people now cannot afford a home, and in health care, where paying people to stay home and care for relatives without checking whether care was delivered drove costs up
"the core root of inflation in this country, the core root of it is government spending" — David Friedberg, who says the reason the country ends up socialist is that nobody reins the spending in, and that the markets are now saying they will not underwrite it
Mr. Palihapitiya's caveat: fraud is layer one of about eighteen layers
The Mortgage Rate Traces Back to Federal Spending, and Nobody Connects It
Mr. Sacks walks the chain for the audience: the 30-year mortgage averages 6.73%, a little under 7%, about 150 basis points above the 30-year Treasury. Higher spending means higher Treasury rates, which means higher mortgage rates, which means a smaller house for the same monthly payment. His question was whether the average person connects those dots
Mr. Friedberg's answer is no, because the first thing a human does is look for someone to blame — including, he noted, the four of them on a podcast. You do not ask what the solution is, you ask who is at fault, and then you re-elect someone
The candidate who wins that moment is the one offering free housing, free buses, free mortgages and taking land from the landlord, which he calls putting more fuel on the fire when the fire is government spending
AI Is the Only Growth Story Big Enough to Outrun the Debt
Mr. Sacks: "only AI can create the exponential growth necessary to make our economy grow big enough that we can grow out of the problem". His warning is that a new regulatory apparatus taking years to approve model releases would end that hope
"We are kind of counting on AI as the deus ex machina to grow our way out of this thing." — Chamath Palihapitiya, who says the economy needs quarters like the ones Nvidia and Salesforce just posted, repeatedly, for something like ten years
The exchange had started as a joke, with Mr. Calacanis pretending to blame Dario Amodei, Canada and Putin for the deficit before making clear he was kidding and inviting Mr. Amodei to the summit
Texas Let People Build, and Austin Prices Fell From Their 2022 Peak
Mr. Calacanis delivers an unbroken monologue on housing — Mr. Palihapitiya's reaction was "I love a good J-Cal polemical" — that the others treat as the set piece of the episode.
He says he lived in New York, Los Angeles and San Francisco without appreciating what he now sees, having moved to Texas: that people in expensive cities are miserable because they cannot afford a home and pay exorbitant rent, while existing owners refuse to allow building
Austin housing prices are down 27% from the 2022 peak, which he offers as proof it can be done where leadership makes housing a priority and lets people build
He points at Florida for the same reason: Brightline's trains run at 125 miles an hour because the state got out of the way and let private industry build them
"These things can be solved. It just takes leadership." — Jason Calacanis, whose complaint is that no leader will say enough already and let people build more houses
Asked whether he had practiced it, he said it was off the top of his head, then joked that he had hired Mr. Sacks as a writer to punch it up
Druckenmiller Wrote His Op-Ed With AI, and the Panel Split Three to One
Mr. Calacanis raised the side quest: Mr. Druckenmiller was caught using AI to write the op-ed, and the Wall Street Journal said it was fine. The argument that follows runs for a substantial stretch of the episode.
Mr. Palihapitiya's position is that the objection is a class of people virtue signaling. "Stan Druckenmiller is the most incredible investor of our generation." Maybe top three or four, Mr. Sacks agreed. The complaint about an em dash and an AI flourish, he says, misses that the man is telling you what is about to happen
The questions he says people should have asked instead: how does this affect my company, my portfolio, my 401k, my parents' retirements, how should I hedge it
Mr. Sacks: "A more precise way of putting it is the take is his. And then he uses the AI to help him write it." He calls the pearl-clutching unwarranted, notes the piece cites Mr. Druckenmiller's own past positions many times, and adds that if he is plagiarizing anyone he is plagiarizing himself
Mr. Calacanis is the dissent. "I do think it's kind of the lip-syncing of writing as a writer." Research, proofreading and grammar checking are fine; letting the system supply the opinion is not, and the failure is non-disclosure
His argument is the reader's expectation: people care about Mr. Druckenmiller's opinion and want to be sure it is not AI slop, which is why they felt betrayed — and he thinks the Journal should be checking submissions the way professors should check student papers
He said an AI detector put the piece at 90% written by AI, which Mr. Palihapitiya disputed
Mr. Friedberg's counter is the tool analogy. "Do I have to disclose that I used Excel or that I used a calculator or that I used a synthesizer or that I used a Photoshop as opposed to doing it in an analog way?" Digital tools magnify human creativity, and "using AI" is as nebulous a category as any of those — spelling, grammar, novel sentences, whole paragraphs, or editing an output built from your own inputs
Mr. Sacks then made a disclosure of his own: "This is a standing disclosure for all future publications that I put out, including tweets, is that I try not to publish anything without running it through AI first to fact-check it, to line edit it, and to help me make the best argument I can."
Mr. Palihapitiya cross-examined the position — is Mr. Druckenmiller allowed a computer, a word processor, Grammarly? Mr. Calacanis said yes to all three and disclosed that he is a Grammarly shareholder. Mr. Palihapitiya's guess at the actual process is that Mr. Druckenmiller had the take, gave it to the AI in writing or by voice, got an output and edited it, against Mr. Calacanis' picture of it: "He opens ChatGPT and he writes, yields up bad, gets the whole thing, printed out"
Mr. Sacks' objection to the lip-syncing analogy is that lip-syncing is a song performed a thousand times, while this was a novel criticism of a specific policy his former colleague had just announced. He also noted that Mr. Druckenmiller says he is not the greatest writer, and that what makes him interesting is that he is a thinker with a long-held take he was willing to use against a friend
Mr. Calacanis' closing position: he will not think as highly of Mr. Druckenmiller and will wonder how much of the work was his — it is only 800 words, write it yourself and work with a proofreader. Mr. Palihapitiya pressed him on whether he would now discount the substance; he said he would wonder what the actual opinion was. They agreed to disagree
Ratcliffe Went to Moscow, and Both Governments Deny What It Was About
The facts as read out: CIA director John Ratcliffe made an unannounced visit to Moscow on Tuesday and flew back the same day, the first visit by a sitting CIA director since November 2021, when Bill Burns went to warn Putin not to invade Ukraine — he invaded three months later. The Wall Street Journal reported Mr. Ratcliffe was there to warn Russia off attacking a NATO country, and new US intelligence reportedly shows Putin may plan to test NATO's resolve with a limited assault. Polymarket put a Russia-Ukraine ceasefire by year end at 21%
Mr. Friedberg called it speculation and said so himself, relaying that the reported message was don't invade NATO and don't use nukes on Ukraine, while noting there is no evidence
Mr. Sacks said both governments denied it — the president called into a show to say the story was untrue and the contact routine, and the Russian readout said roughly the same. He believes the denials because invading NATO countries would make no sense for Russia, and calls the story the same threat inflation heard from neocons for years: "This is the same kind of threat inflation that we've heard for years from neocons about Russia's supposedly imperial ambitions to conquer all of Europe."
His read on the war is that Ukraine is losing. Russia has air superiority, Ukraine is basically out of air defense, Kyiv is bombed routinely and Russian forces can destroy what they want at will, on the eve of an expected cold winter with Ukrainian power being taken out
Odessa and the Black Sea ports are shut, grain cannot be exported, and the economy is being destroyed. Zelensky is asking for another $35 billion; the United States is not funding it and he is trying the Europeans
Casualties are huge on both sides though the numbers are unknown, and there are stories about drafting women. Russia has four or five times Ukraine's population and can sustain far larger armies
"Zelensky should have listened when President Trump told him, you don't have the cards in this war." — David Sacks, who does not think Ukraine comes out on top
Mr. Calacanis' contribution was a joke that Mr. Ratcliffe flew to Moscow to warn Putin not to use AI to write his next essay
Meta's Instagram Limits Let Parents Stop Policing Devices
Mr. Palihapitiya thanked Meta for imposing limits on Instagram, saying it solves his biggest problem: he is tired of playing whack-a-mole with his kids, their fake accounts and secret accounts
"they got dragged kicking and screaming to do the right thing" — the same speaker, who says it could have happened much sooner but applauds it, hopes TikTok and YouTube follow, notes that similar limits are arriving in Australia, Canada and the UK, and says such limits are physiologically and psychologically helpful for children
He does not allow his own children on social media until they are 16
Mr. Calacanis held the line to 16 as well and said he lost at 17. His complaint is that parental control apps do a decent rather than a great job, which leaves a parent permanently on the back foot, and that he would rather spend time with his children than scream at them to get off their phones
Both said the point is that platform-level limits let parents parent instead of police
Moderna's Cancer Shot Is a Technique, Not a Drug
Mr. Calacanis asked about the cancer vaccine and the stock that tripled in two weeks, and Mr. Friedberg gave the longest explainer of the episode.
Moderna went from roughly a $20 billion market cap to $60 billion on positive readouts from what is being called a cancer vaccine
He objects to the word vaccine: "A vaccine means you prevent yourself from getting something, whereas in this case, you're actually getting rid of something you already have, so I don't like the word vaccine."
How the technique works, in his account: a neo-antigen is a small peptide or protein unique to a cancer. Put it in the body, the immune system reads it as an invader, makes cells that destroy that protein, and those cells then hunt the cancer carrying it
The idea dates to the late 1990s. DNA sequencing made it personal: sequence a tumor sample, find the sequence unique to that cancer, and build an antigen for that patient, because no two cancers are identical
Melanoma is a prime target because UV damage gives it very distinctive mutations
Moderna's variant delivers mRNA rather than the protein, so the patient's own cells make the protein in vivo — the same mechanism as the COVID vaccine
The alternative is making the protein outside the body: put the DNA sequence into E. coli in a bioreactor, let billions of cells make billions of copies, isolate it and inject it. "So you don't necessarily need mRNA, is my point."
Hundreds of trials have run on neo-antigen therapy, and what is being worked out is delivery, technique, cancer type, dosing, adjuvants and combination therapies rather than the underlying idea
The cost argument is where Mr. Friedberg gets angry. Sequencing and the upfront work are cheap — he says his company does DNA sequencing all day long — and making the mRNA or the protein should also be cheap. What costs money is human trials and safety work
"You can pay someone $50,000 to do this for you today." People are going to clinics in Montana, where right-to-try laws apply, having peptides printed for their cancer sequence, and getting cured
"So why is Moderna saying that they're going to charge $500,000 for this?"
He calls it a form of regulatory capture: a technique developed over decades largely on NIH and other public research funding, now patented, FDA-approved and priced at half a million dollars. He argues it should be open-sourced and every hospital trained to do it
On patents, his view is split. Moderna does have strong coverage on its own mRNA tools — the methods that keep the mRNA out of the human genome and stop it replicating in the body — but the general therapeutic process is well understood and, he expects, will proliferate cheaply overseas over the next couple of years
Mr. Calacanis asked whether the patent survives globally given medical tourism, and made the point that all of this depends on early detection, naming the Galleri blood test; Mr. Friedberg confirmed that is Grail's test
The wider picture is new modalities, not one product. CAR-T therapy takes T cells out, reprograms them with CRISPR or a similar system to hunt a specific protein, and puts them back. It costs a million dollars and works best on blood cancers such as multiple myeloma, where the cells float individually rather than clumping in a tumor. He says the efficacy there is now incredible, and that in aggregate the direction of cancer treatment is a reason for optimism
Mr. Friedberg's bottom line on the debt segment is that no Treasury action clears a $10 trillion refinancing and the break arrives between 2030 and 2032 when Social Security and the states run out of money; Mr. Palihapitiya's is that the bond market, not the electorate, is what will eventually force Congress to act. On the day's earnings, Mr. Sacks' conclusion is that the SaaS-is-dead and AI-capex-bubble narratives both got shredded by a single afternoon of results. And on the news that closed the show, Mr. Friedberg's position is that Moderna's cancer therapy is a public-funded technique being sold as a private drug.
Products, Companies & Tools Mentioned
Nvidia (Guided to 70% growth against a street at 45%, on $96.2 billion of quarterly revenue and $60 billion of profit; Sacks says it still trades at 12 times earnings)
Salesforce (The stock the SaaSpocalypse trade was built on; up over 20% after beating adjusted EPS by 80% and guiding the full year to $46 billion, and now integrating into Claude)
Anthropic and Claude (Salesforce is both an investor and, per Sacks, the party giving Claude the primary user relationship over its own front end)
Hugging Face and Poolside (Reported Nvidia acquisitions at $12 billion and $6 billion; Calacanis says they give Nvidia the distribution layer for open source. Poolside makes the coding model Laguna)
Grokbot (The always-on cloud agent harness Sacks is using at $200 a month, contrasted with desktop harnesses like OpenClaw and Hermes)
Optimus (Tesla's humanoid; Palihapitiya says unreleased footage he was shown would already win half the robot Olympics events)
Moderna (Went from a $20 billion to a $60 billion market cap on cancer immunotherapy readouts; Friedberg calls the $500,000 price regulatory capture)
Polymarket (Priced Nvidia at 79% to end the year as the most valuable company, and a Russia-Ukraine ceasefire by year end at 21%)
Workday, Oracle, SAP and IFS (The other horizontal systems of record Palihapitiya says are very difficult to replicate)
Slack and Slack Today (Calacanis wanted off Slack over data access; the new Salesforce product he rates a 7 out of 10 and improving about 10% every couple of weeks)
Brightline (The privately built Florida railway, offered twice as the counter-example to California high-speed rail, running at 125 miles an hour)
AMD (The OpenAI deal Calacanis says provoked Jensen Huang, alongside OpenAI's own Jalapeño inference chip)
Grail's Galleri test (The blood test Calacanis raised as the early-detection piece that cancer immunotherapy depends on)
Grammarly (Allowed under Calacanis' own standard for AI writing help; he disclosed that he is a shareholder)
Meta, TikTok and YouTube (Meta imposed teen limits on Instagram; both Palihapitiya and Calacanis want the other two to follow)
Books & Resources Mentioned
Stan Druckenmiller's Wall Street Journal op-ed on the Bessent buyback plan (The piece the panel argued over both for its substance — spending, not price manipulation, is the problem — and for having been written with AI)
Apple Podcasts (The episode on Apple)
Episode page (The show's own page for this episode)
Get the latest market chatter as it happens: