All-In Podcast in 3 minutes

Here is The Daily FM summary of the All-In Podcast that aired on Friday October 2nd. The episode centered on what the hosts called a historic White House “Superintelligence Summit,” where President Trump convened major AI leaders—including Elon Musk, Mark Zuckerberg, Jensen Huang, Dario Amodei, Satya Nadella, and representatives from OpenAI—to discuss both AI competition and safety.

David Sacks, who appeared closely involved, described the meeting as “the Bretton Woods of superintelligence.” The headline result, he said, was a White House Accord on Superintelligence, signed by six leading frontier-model companies. Its framework would make companies responsible for their own safety controls, require internal oversight, external audits, independent board committees, and potentially create enforcement exposure through the SEC and FTC if companies fail to live up to public commitments. Sacks argued this approach has practical “teeth” without waiting for legislation or an unwieldy international treaty.

Chamath Palihapitiya praised the speed of the agreement and framed the core geopolitical message bluntly: “whoever wins superintelligence wins.” He argued that pausing AI development would simply hand an advantage to China. The hosts repeatedly dismissed what they called “doomerism,” saying alarm without concrete solutions is less useful than safety measures that allow development to continue. A notable moment was the apparent thaw between Sacks and Anthropic CEO Dario Amodei, previously a frequent target of Sacks’s criticism. Sacks said their first in-person meeting was cordial, with Amodei jokingly calling him his “nemesis.”

David Friedberg offered the episode’s most consequential prediction. He argued that open-source and open-weight AI models have become too widespread for any government to centrally control software development. Instead, he expects the next battleground to be compute: governments may eventually regulate access to data centers, chips, GPUs, and inference capacity, allocating resources to defense, financial systems, and other critical sectors. The group agreed that AI has turned data centers and domestic electricity production into national-security infrastructure. Sacks warned that America’s power grid has barely grown in decades while China continues expanding rapidly, and called opposition to data centers dangerously shortsighted.

Still, Jason Calacanis pushed back on the summit’s optics. He said ordinary Americans may see wealthy technology executives securing favorable policy while worrying that AI will eliminate their jobs. He urged companies and government to make the benefits tangible through cheaper education, better healthcare, housing, training, and local community investment. Friedberg responded that open AI systems could be more broadly empowering than closed platforms, much as the open internet prevented telecom gatekeepers from controlling access.

The conversation then shifted to the economy and midterms. Sacks argued that revised GDP growth, stronger-than-expected jobs numbers, falling poverty, and cooling inflation point to a healthier economy than public sentiment suggests. Friedberg cautioned that elevated interest rates could impair smaller banks and consumer affordability. The hosts split on elections: Calacanis predicted Democrats could sweep Congress, while Sacks and Chamath expected Republicans to retain the Senate and Democrats to narrowly take the House.

Finally, the panel condemned media coverage of an alleged attempted hijacking of a FlyDubai flight carrying Israeli passengers. They praised the injured captain, off-duty pilots, and passengers who reportedly subdued the co-pilot and landed safely, while criticizing outlets for initially calling it an “altercation” rather than an apparent terrorist attack.

Thank you for listening to All-In Podcast in 3 minutes from The Daily FM. See you next time!

Here is The Daily FM summary of the All-In Podcast that aired on Wednesday September 30th. This episode featured two conversations about turning attention and creative success into durable businesses: first with boxer, creator, and investor Jake Paul; then with Chainsmokers members Alex Pall and Drew Taggart on music, venture capital, and the danger of hype-driven investing. [1]

Jake Paul described growing up online, beginning on Vine before moving to YouTube, Snapchat, and Facebook. His core lesson was that authenticity matters more than simply chasing virality. He says he created content because he enjoyed storytelling, entertaining people, and encouraging fans—not because he had identified a trend. Still, he acknowledged the ugly incentives of the attention economy: creators, journalists, and platforms can be pushed toward increasingly outrageous hot takes and clickbait. He called this the “YouTuber disease.”

Paul told a striking story about Vine’s demise. He said the platform’s top 20 creators asked Vine for one million dollars each annually, arguing they supplied the content and audience. Vine offered one million dollars total to divide among them, so the creators left for competing platforms. Within months, Paul says, Vine was effectively dead—a reminder that platforms depend on the talent they may underestimate.

He framed boxing, business, media, and investing as one mutually reinforcing flywheel. Boxing began as a challenge match with British creators, but after winning by knockout, he became committed to the sport and recognized the advantage of bringing a huge existing audience to it. He argued that his Most Valuable Promotions organization can challenge the UFC by putting fighters first: he claimed UFC athletes receive roughly 15 percent of revenue, compared with around half in major traditional sports. He also pointed to his work elevating women’s boxing and funding free youth boxing gyms as the work that gives him the most satisfaction.

In a surprising closing prediction, Paul said politics may be his path at age 40. He believes future elected leaders will need direct, social-media-native relationships with audiences, and he sees public office as a way to make broader change.

The Chainsmokers then explained their move from DJs into serious venture investors through their firm, Mantis. Their early musical breakthrough came from scrappy internet marketing: identifying bloggers behind the influential Hype Machine chart and sending unusually personal emails promoting their remixes. Today, they see music as far more crowded, with an estimated 300,000 songs uploaded to Spotify every day. They said live DJ performances offer better economics than streaming, while artists increasingly face a choice between label support and retaining ownership.

On investing, the duo rejected the idea that celebrity alone creates venture success. They argued venture is illiquid, long-term, and brutally competitive, with a small share of firms generating most returns. Fame can open doors, they said, but it can also make institutional investors skeptical. Their response is to do the work, focus on cybersecurity, AI infrastructure, deep tech, and health tech, and contribute through branding, distribution, and relationships.

The final discussion turned skeptical about today’s venture market. The guests warned against buying into companies simply because their valuations keep rising, especially when later investors pay multiples more without underlying performance improving. Their bottom line: attention can be valuable, but only if it is translated into real execution, lasting businesses, and actual cash returns.

Thank you for listening to All-In Podcast in 3 minutes from The Daily FM. See you next time!

Here is The Daily FM summary of the All-In Podcast that aired on Monday September 28th. Daniel Ek, Spotify’s co-founder and now executive chairman, joined Jason Calacanis and David Friedberg to discuss Spotify’s origins, his new preventive-health company Neko Health, and his cautiously optimistic view of AI. [1]

Ek began with Spotify’s unlikely creation during music piracy’s peak. In 2006, Sweden had fast broadband, rampant illegal downloads, and virtually no legal digital-music options. The music business was collapsing, but Ek and co-founder Martin Lorentzon saw that consumers would pay for a vastly better experience: all the world’s music, instantly accessible. The catch was licensing. Ek said he and his partner invested their own money and persuaded record labels with a clever guarantee: if Spotify failed after a year, the labels still met their financial targets; if it worked, everyone won. It took years of negotiation before Spotify launched in Sweden in 2008, then the UK, and finally the U.S. in 2011.

The larger focus was Neko Health, Ek’s eight-year-old healthcare startup. His argument is that American healthcare remains designed to react to acute illness, even though chronic diseases are the biggest cost and suffering burden. Many serious illnesses can be treated far more cheaply and effectively when found early, he said, but the system lacks enough high-quality, longitudinal patient data.

Neko’s answer is an annual, roughly one-hour health scan priced at $499. It combines 53 blood markers, cardiovascular measurements, strength testing, and more than 6,000 high-resolution skin images, followed by time with a clinician. The company has performed over 100,000 scans and says about 1 percent of members have had a serious undiagnosed condition identified. Ek stressed that most customers are healthy; the purpose is both to catch hidden risks and make lifestyle improvements more concrete.

The most striking example involved skin screening. Ek said the average Neko patient has around 950 moles, an astonishing number that drew disbelief from the hosts. Rather than relying on a doctor to notice one suspicious spot, Neko uses AI to catalog and flag lesions, then has clinicians and dermatologists review findings. On repeat visits, it can detect subtle changes over time that even an excellent physician could not remember from one year to the next.

Ek acknowledged that $499 is still meaningful money, but argued preventive care should become as routine as an annual dental visit. He said Neko’s vertically integrated model—its own clinics, equipment, software, and staff—lets it offer the service with positive unit economics. He also identified a central structural problem in U.S. healthcare: insurance is frequently tied to jobs, so insurers have little reason to fund prevention whose payoff may arrive a decade after a patient switches employers.

On AI, Ek resisted sweeping pronouncements about whether development should slow down. Instead, he argued the industry does a poor job explaining AI’s positive uses, such as early disease detection and better music personalization. He endorsed a future with both closed frontier models and open-source systems, while suggesting that access to large-scale compute may matter more for safety than trying to regulate software alone.

Thank you for listening to All-In Podcast in 3 minutes from The Daily FM. See you next time!

Here is The Daily FM summary of the All-In Podcast that aired on Friday September 25th. The hosts used the episode to argue that the AI debate has shifted from theoretical fears of “superintelligence” to a more practical question: should companies be held accountable for products they choose to release?

David Sacks and Chamath Palihapitiya criticized leading AI developers, especially Anthropic and OpenAI, for calling themselves “labs” while operating as enormous, profit-seeking corporations. Their central argument was that companies should not seek broad immunity or global governance schemes to avoid responsibility. Instead, they should test products, slow down releases when necessary, and face normal product-liability rules—much as Tesla, Meta, and other scrutinized technology companies do.

Sacks argued that international AI regulation would arrive too slowly to matter and could dilute accountability. Chamath’s sharper formulation was that these are not neutral research labs: they have shareholders, giant valuations, major capital needs, and should be judged accordingly. The group also pushed back on the idea that competitive markets inherently make AI unsafe. Their view was that customers do not want unreliable products, and businesses already have strong legal and financial incentives not to release dangerous software.

David Friedberg then laid out the episode’s biggest technology takeaway: open-weight AI is progressing at a startling speed. He cited a flood of recent releases from Chinese and other developers, including models capable of advanced coding, image generation, robotics, and technical reasoning that can be downloaded and run locally. In his telling, AI capability is no longer confined to a few frontier companies or giant data centers. The hosts argued that this makes calls to “ban superintelligence” increasingly unrealistic: software has already spread widely, and restrictions could simply move talent and innovation abroad.

The panel debated what that means for Anthropic and OpenAI. Chamath said model capabilities are clustering, while the real differentiation increasingly lies in the “harness” around a model—the agents, interfaces, and workflows that let it take useful action. He warned that customers will pressure companies to move from expensive premium models to cheaper open alternatives. Friedberg noted that Anthropic may retain an edge in difficult scientific and life-sciences applications, where customers could pay heavily for frontier performance. Sacks was more optimistic about both Anthropic and OpenAI, calling them a durable frontier-intelligence duopoly—but warned they must keep improving because their lead may last only six to twelve months.

Anthropic’s possible IPO became a focal point. Sacks said the company’s own public warnings about catastrophic AI risk, combined with its push for regulation and its plan to open a biological research lab, could create unusually severe investor risk disclosures. Chamath countered that CEO Dario Amodei has built a formidable company and talent magnet, but said the IPO may need a lower valuation to compensate investors for those risks.

The episode ended on a more optimistic note around Meta’s Muse and xAI’s Grokbot. The hosts described these as the first broadly accessible AI agents that could book travel, manage inboxes, identify savings, process returns, and act like free personal assistants. Their prediction: once ordinary users experience that utility, fear of AI may give way to demand for it. Thank you for listening to All-In Podcast in 3 minutes from The Daily FM. See you next time!

Here is The Daily FM summary of the All-In Podcast that aired on Wednesday September 23rd. The episode featured Luca Ferrari, CEO of Milan-based Bending Spoons, explaining how his company turned a failed AI startup, a leftover $40,000, and a $10,000 app acquisition into one of technology’s most aggressive—and unusual—serial acquirers. [1]

Ferrari said he and his co-founders began with an AI company in 2010, but were simply too early and it failed after three years. Their investors effectively let them keep the remaining $40,000 rather than spend more on liquidation. Instead of taking the suggested vacation, the founders used it as seed capital for Bending Spoons in 2013. Their key insight was unusually candid: they were not especially good at discovering product-market fit from scratch, but had become strong at engineering, design, marketing, and monetization. So they decided to “buy product-market fit” from others, then improve it.

Their first deal was a roughly $10,000 acquisition of a lightly monetized iPhone keyboard-customization app. It had users and favorable App Store positioning, but Bending Spoons rebuilt it entirely. That basic formula has since expanded to much larger brands, including Vimeo, AOL, Eventbrite, Airtable, and Miro. Ferrari described Bending Spoons not as a holding company but as an operating platform: when it buys a company, it deeply integrates it into a shared system of more than 50 internal technologies, covering areas such as AI-model orchestration, recruitment, A/B testing, payments, and operations. [2]

The most controversial part of the model is its lean staffing. The hosts compared Bending Spoons’ post-acquisition cuts to Elon Musk’s approach at X, but Ferrari said the company learned its staffing discipline organically. Early acquisitions often came without the original team, forcing Bending Spoons to operate comparable businesses with much smaller internal teams. Its theory is that small, highly talented, highly accountable teams are more likely to produce “10 out of 10” performance than large organizations. Of about 800 core employees, roughly three quarters work in engineering, AI, product design, or product management. Ferrari said the company received 800,000 applications last year and hired fewer than 300 people.

Financially, Bending Spoons has relied mostly on reinvesting free cash flow and using debt, rather than repeatedly raising equity. Ferrari said its blended debt cost is around 9 percent, fully hedged, with maturities extending to 2031. His surprising argument was that higher rates might not be disastrous: while new borrowing would cost more, the valuations of acquisition targets could fall, creating better opportunities for a disciplined buyer with historically strong unlevered returns.

Ferrari argued Bending Spoons differs fundamentally from traditional private equity. Private-equity owners usually keep companies separate so they can resell them; Bending Spoons transforms them onto a common technical foundation and shares elite talent across the portfolio. That makes its model harder to copy, but also means it cannot simply flip assets.

The hosts also challenged the assumption that Europe lacks world-class technology talent. Ferrari acknowledged Europe’s structural problems but said its half-billion-person market, solid education, and ambitious people eager to prove themselves provide an underrated advantage. Milan remains Bending Spoons’ largest talent hub, but London, Madrid, and eventually the United States are growing. His closing message was that the company stays focused on becoming exceptional at one thing: acquiring mature digital businesses and making them materially better. Thank you for listening to All-In Podcast in 3 minutes from The Daily FM. See you next time! [3]

Here is The Daily FM summary of the All-In Podcast that aired on Tuesday September 22nd. The episode focused on California politics, featuring Republican gubernatorial candidate Steve Hilton and media personality turned Los Angeles political candidate Spencer Pratt. Their shared message was that California’s government has become an entrenched, one-party “machine” that rewards bureaucracy, special interests, and disorder while making ordinary life unaffordable.

Hilton, a British-born former tech entrepreneur, presented California as a uniquely consequential test case. In his words, California means to America what America means to the world: a place historically defined by innovation, ambition, agriculture, entertainment, and technology. Yet he argued that 16 years of Democratic control have produced the country’s highest poverty rate, unemployment, housing costs, and regulatory burden, along with a continuing exodus of residents and businesses. He blamed one-party rule for letting unions, trial lawyers, donors, and politically connected nonprofits dominate Sacramento. [1]

His economic pitch centered on lowering the cost of living. Hilton proposed cutting California’s regulatory count from more than 420,000 to below 200,000, reducing government headcount, canceling high-speed rail, ending some homelessness-related nonprofit funding, and using the savings to eliminate state income tax on the first $150,000 earned. He also pledged to reduce gasoline prices by rolling back state climate-related fuel charges and to cap annual vehicle registration at $73 by abolishing the DMV. The hosts pressed him on his difficult polling position, but Hilton said his campaign had narrowed the gap against Democrat Javier Becerra, especially among independents and financially stressed working-class voters.

Hilton’s sharpest attack was on Becerra, whom he portrayed as a career politician and a likely continuation of Governor Gavin Newsom’s policies. The conversation was more persuasive when it focused on concrete examples than personal insults: Hilton described small businesses facing illegal street vendors, excessive fees, and rising energy costs, and argued that California has inverted incentives, punishing rule-following businesses while tolerating disorder.

Spencer Pratt joined to recount his own Los Angeles campaign after losing his home in the Pacific Palisades fire. Pratt’s anger over the fire, empty reservoirs, brush-management rules, and what he sees as city incompetence drove much of the discussion. He said he entered politics to force leaders to take public safety, infrastructure, and rebuilding seriously. He also argued that viral AI-assisted campaign videos helped him break through traditional political media, though he emphasized that the message—not the technology—was what resonated.

The most contentious section concerned mail voting and ballot harvesting. Pratt alleged that California’s rules enable highly organized political groups to collect ballots and influence voters, while acknowledging he lacked evidence of specific fraud in his own race. Host Jason Calacanis repeatedly challenged him to distinguish legal ballot collection, campaign persistence, and actual criminal fraud. The exchange exposed a real disagreement: Pratt and some hosts viewed the system as structurally vulnerable, while Calacanis insisted extraordinary claims require evidence and legal action.

The episode ended on Hilton’s more optimistic note. Beyond cutting waste, he wants California to build “ten new cities,” revive housing and business creation, and reclaim its role as a place where people come to build. Thank you for listening to All-In Podcast in 3 minutes from The Daily FM. See you next time!

Here is The Daily FM summary of the All-In Podcast that aired on Tuesday September 22nd. The episode featured Boom Supersonic founder and CEO Blake Scholl, who made an energetic case that commercial aviation has stagnated for decades—and that his startup can bring back faster-than-sound passenger travel while also becoming a major supplier of power for AI data centers. [1]

Scholl’s opening argument was deliberately provocative. In 1969, humanity landed on the Moon and flew an airliner through the sound barrier. Yet more than 50 years later, he said, neither lunar exploration nor supersonic civilian travel has maintained that momentum. Concorde’s retirement in 2003 left commercial passengers flying at essentially the same speeds as previous generations. He also criticized the aerospace incumbents, noting that Boeing’s last all-new airliner program began in 2004 and arguing that the giant has become complacent. [2]

Boom began in far less glamorous fashion: its early prototype was essentially a cardboard-and-plywood mockup with office-store seats. Scholl, formerly a software engineer and advertising executive, recalled pitching Jeff Bezos for seed funding in 2015. Bezos passed, while warning that startups could not build a composite airliner in a garage. Scholl said he took that personally. Boom later built XB-1, which it describes as the first privately developed jet to break the sound barrier, with a team of only about 50 people. [3]

The technical centerpiece was Boom’s approach to the sonic-boom problem. Rather than eliminate the boom at its source, its “boomless cruise,” or Mach cutoff, concept uses atmospheric conditions to bend the shockwave upward before it reaches the ground. Scholl compared it to a tree falling where nobody hears it. He said this enables flight around Mach 1.3 over land—about 50 percent faster than today’s commercial speeds—with no boom heard below. Over oceans, Boom’s planned Overture aircraft would fly at Mach 1.7, cutting a transatlantic trip to roughly three and a half hours.

Scholl said the longstanding U.S. ban on civil supersonic flights over land ended by executive order after Boom’s testing, and he emphasized that permanent legislation has advanced unanimously through the House and Senate Commerce Committee. The political surprise, he argued, is that supersonic travel has become a rare bipartisan cause. He praised the current administration for moving quickly, saying an invitation to the White House arrived within a day of his public campaign. [4]

The larger business surprise was engines. After a public split with Rolls-Royce, Boom decided to develop its own engine rather than outsource the task. Scholl now believes that setback may have saved the company: the same high-power turbine core needed for supersonic aircraft can be adapted into generator systems for data centers. Boom says it has demand in the tens of gigawatts and plans to test its first engine core soon. Its trailer-mounted systems could provide 42 megawatts of on-site generation without water cooling.

Asked when passengers can fly, Scholl estimated four years as the goal, though he stressed that speed matters more than predictable deadlines. He put Overture’s transatlantic break-even round-trip fare at about $3,500, with airlines likely charging more. His long-term definition of success was broader than premium travel: a future where children find a six-hour transcontinental flight as strange as they now find rockets that fail to land themselves. Thank you for listening to All-In Podcast in 3 minutes from The Daily FM. See you next time!

Here is The Daily FM summary of the All-In Podcast that aired on Monday September 21st. The episode featured AI entrepreneur Naveen Rao, co-founder and CEO of Unconventional AI, making an audacious case that artificial intelligence’s biggest near-term constraint is not model capability but energy—and that conventional computer architecture may be fundamentally too wasteful for the intelligence systems people want to build. [1]

Rao is an AI-chip veteran who founded Nervana Systems in 2014, sold it to Intel, then led Intel’s AI group. Later, he helped build infrastructure for training large models before joining forces with Databricks. His newest company is attempting what he described as a first-principles redesign of computing, with a target of delivering 1,000 times better power efficiency within three and a half years.

His core warning was stark. Google alone reportedly processes 3.2 quadrillion AI tokens per month. Using a rough estimate of 10 joules per token, Rao said that could imply 12 gigawatts of power—against approximately 40 gigawatts currently going into all U.S. data centers. Whether the exact estimate holds, his point was that AI demand and model sizes are growing far faster than electricity supply. Data-center planning, he said, has already shifted from finding floor space, racks, or GPUs to first securing power contracts. Energy represents about half the cost of serving an AI token.

Rao argued that today’s computers waste energy mostly by constantly moving information between separate memory and compute components. GPUs remain descendants of the old von Neumann architecture: extremely fast, but designed around digital abstractions and data movement rather than efficiency. Biology provides an alternative benchmark. The human brain uses roughly 20 watts; a squirrel’s highly capable brain uses only about 8 milliwatts. Rao’s provocative ambition is not merely to imitate biology, but ultimately to “beat biology.”

Unconventional AI’s proposed answer is a “dynamical computer,” where memory and computation are integrated into physical, time-varying circuits. Rao illustrated the concept with metronomes placed on a movable board: even when started out of sync, physical interaction causes them to synchronize. His company seeks to harness similar emergent behavior in semiconductor circuits, rather than forcing every task through conventional matrix multiplication and repeated memory access.

The most notable reveal was Rao’s claim that his team has built the first physical dynamical computer. The startup began in earnest in January, sent its chip design to fabrication in June, and now has a working prototype in the lab. It generated images using approximately 500 nanojoules per image, which Rao contrasted with millijoules for conventional GPU-based computing—suggesting orders-of-magnitude lower energy use.

He also said a key breakthrough involves “sparsity”: deliberately removing many connections in a system rather than connecting every element to every other element. Counterintuitively, Rao said this can improve scalability, efficiency, and trainability all at once.

In the discussion afterward, Rao said the company expects to have a full product in about two years, beginning with data-center racks that provide tokens through a network connection while hiding radically different hardware underneath. Existing models could be adapted at the model level, although migration will require work. The team spans theorists, neuroscientists, physicists, and chip engineers—a difficult mix to coordinate—using Python-based libraries to bridge disciplines.

Rao’s closing thesis was optimistic: cheaper intelligence may increase total AI consumption through Jevons paradox, but it could also decentralize computing, enable local data centers, and make vast robotic systems economically viable. Thank you for listening to All-In Podcast in 3 minutes from The Daily FM. See you next time!

Here is The Daily FM summary of the All-In Podcast that aired on Sunday September 20th. The hosts interviewed AppLovin founder and CEO Adam Foroughi, a low-profile executive behind one of the market’s most remarkable recent turnarounds. AppLovin began as an advertising platform for mobile-game developers, helping them monetize the enormous amount of time adults spend playing casual games. Foroughi estimated that more than a billion people play mobile games each day and that the ecosystem now generates roughly $50 billion a year in advertising spend. [1]

The company’s larger ambition is to move beyond getting players from one game to another and use its recommendation technology to introduce shoppers to products they did not know they wanted. Foroughi drew a useful distinction between search advertising and discovery advertising. Search, including advertising that may eventually appear in AI chatbots, serves someone who already knows they want dress shoes or another product and is ready to research it. Discovery advertising, like Facebook and Instagram ads or AppLovin’s newer commerce efforts, attempts to create demand by presenting a relevant product unexpectedly. He argued that this produces real economic expansion, not merely a better route to an already-planned purchase.

Foroughi described advertising as “machine learning 1.0.” Recommendation systems and large language models differ technically, but he said both grew from related research and both improve by predicting outcomes. Advertising has an unusual advantage: it gives instant feedback on whether a prediction produced a transaction, making it an extremely profitable proving ground for deep-learning systems.

One of the most notable exchanges concerned the persistent suspicion that phones are secretly listening to conversations. Foroughi rejected that idea as impractical, given the scale of data needed to record and interpret audio for ads. He suggested the eerie coincidence usually comes from more ordinary signals—searches, browsing behavior, or connections between people on social networks. The hosts raised the possibility that platforms may infer interests among nearby friends or social groups, but Foroughi’s broader point was that targeted advertising is less mysterious than it appears, and that people often prefer relevant ads to irrelevant spam.

The most dramatic part of the episode was AppLovin’s market collapse and recovery. The company went public in 2021 at about a $28 billion valuation, rose to $40 billion, then fell to roughly $3.8 billion in 2022—a 92 percent drawdown—even as its EBITDA grew from $600 million to around $1 billion. Foroughi blamed a flood of insider selling and insufficient interest from major investors during the Covid IPO boom. Rather than spend time persuading Wall Street, he told his team to become their own best investor. AppLovin spent roughly $6 billion buying back stock, retiring 20 to 25 percent of its shares; at its peak, those repurchases were worth more than $50 billion.

He said maintaining morale required an “us against the world” culture and expanded performance stock awards for key employees who had watched their expected wealth evaporate. The turnaround accelerated after AppLovin upgraded from simpler statistical models to a deep-learning system, improving advertiser performance. When Foroughi finally returned to investor meetings in 2023, he said the stock surged from about $80 to $150 in a week. It eventually ran from $9 to $750, reaching around $250 billion in market value—an equally extreme example of investor herd behavior on the upside.

On privacy, Foroughi accepted the need for clear rules, including Apple’s restrictions on precise targeting. But he argued that privacy changes can lead users to receive worse ads, even though they still spend 30 seconds watching them in exchange for an in-game reward. His view was that regulation has now stabilized and powerful models can still provide relevance with less granular data.

Finally, he argued that AI agents will automate routine purchases, such as supplement subscriptions, but will not eliminate shopping as entertainment. Most people, he said, still enjoy browsing, comparing, tracking packages, and discovering something new. AppLovin’s edge, in his telling, is intense focus: a lean company, specialized talent across Palo Alto, Beijing, and Singapore, differentiated data, and the constant assumption that giant competitors could overtake it at any time. Thank you for listening to All-In Podcast in 3 minutes from The Daily FM. See you next time!

Here is The Daily FM summary of the All-In Podcast that aired on Friday September 18th. Legendary venture investor Bill Gurley delivered an unusually personal keynote arguing that society’s response to major disasters should follow a simple discipline: find the root cause, correct it, and prevent a recurrence. His concern was that the world has failed to apply that standard to Covid-19, despite its enormous human and economic costs. [1]

Gurley began with a tour of catastrophes where independent investigation eventually forced reform. He cited the Surfside condominium collapse, the Boeing 737 Max crashes, Hurricane Katrina, Fukushima, and the Challenger explosion. In each case, he said, a technical failure was compounded by institutional failure: bad designs, ignored warnings, regulators too close to the institutions they oversee, and organizations instinctively circling the wagons. But there were also “searchers”—reporters, whistleblowers, victims’ families, scientists, legislators, and independent investigators—who uncovered what happened and pushed governments and companies toward changes. [2]

His heroes included Richard Feynman, who cut through NASA’s evasions after Challenger by demonstrating on live television how cold could compromise an O-ring, and Japanese investigator Kiyoshi Kurokawa, whose Fukushima commission concluded that the nuclear disaster was not simply natural but “profoundly man-made” and preventable. Gurley’s broader lesson was that public scrutiny and genuine root-cause analysis matter because prevention depends on understanding the actual failure.

He then turned to Covid, which he described as vastly larger than the disasters he had just outlined—citing 15 million deaths and a $45 trillion global cost. Gurley’s central complaint was not that the answer is already certain, but that an independent, adequately resourced inquiry has never been conducted. He argued there has been no sustained major-newsroom investigation, bipartisan congressional effort, federally funded third-party inquiry, or commission comparable to those that followed much smaller tragedies.

Gurley strongly criticized prominent Covid-era scientific figures including Anthony Fauci, Francis Collins, Peter Daszak, Jeremy Farrar, and Ralph Baric. He portrayed them as conflicted “blockers,” rather than impartial investigators, particularly because of their past support for gain-of-function research. He argued that the early scientific consensus favoring animal spillover unfairly stigmatized consideration of a lab accident as conspiracy thinking, and that this chilled journalists and scientists from asking questions.

The most notable evidence he highlighted was the “DEFUSE” proposal, uncovered by the DRASTIC research group. Gurley said the document described EcoHealth Alliance collaborators proposing research that would modify bat coronaviruses, including inserting a furin-cleavage site. He argued this should be treated as a major investigative lead, not proof of any conclusion. He also pointed to China’s reported testing of tens of thousands of animals without identifying a clear intermediary host, while acknowledging that animal spillover should remain open to research.

When asked directly what he thinks caused Covid, Gurley declined to choose a definitive origin. That was his key closing point: the goal should be truth and prevention, not prosecution or partisan victory. He called for support for independent researchers, greater document disclosure from institutions such as UNC, renewed investigative journalism, and a genuinely independent commission—someone with the persistence of Feynman—to establish the root cause before another pandemic tests an already weakened public trust. Thank you for listening to All-In Podcast in 3 minutes from The Daily FM. See you next time!