SpaceX-Cursor Deal, SaaS Debt Bomb, New Apple CEO, SPLC Indictment, Colon Cancer Spike
In a Nutshell
SpaceX is acquiring Cursor for $60B by 2026 (or paying $10B fee), leveraging xAI's compute to dominate AI coding amid 74% Polymarket odds, boosting SpaceX's $2T IPO path. SaaS faces a debt bomb (e.g., Medallia's $6.4B buyout collapsing), with AI agents driving deflation, crashing stocks like Salesforce (-32%), and shifting value to tokens/custom tools. Key updates: Tim Cook stepping down for John Ternus at Apple to revive innovation; SPLC indicted for wire fraud laundering $3M to hate groups via offshore accounts; picloram pesticide linked to 80%+ colon cancer spike in under-50s.
These notes were generated by AI and may contain inaccuracies.
Jason is at the intersection of SPLC files, described as SPLC adjacent. Reference to carrying a stiletto and P35 for protection. Open source to fans. Episode 270 of the All-In Podcast with David Friedberg (Sultan of Science), Chamath Palihapitiya (Dick Tater), Jason Calacanis, and David Sacks (Rainman).
David Sacks was in DC at the White House with POTUS, pushed the podcast back an hour. President Trump is pleasant, genial, asks questions, interested in subject matter, charming. Media portrays him as yelling or moody, but that's not the experience. Lucky to have him as president during AI revolution.
If Kamala were president, no data centers, AI used to censor, promoting DEI values per Biden executive order. Trump wants country to win, no doomer neurosis. Supports specific regulations for problems, not halting progress. Trump suggested over a year ago AI companies stand up own power generation behind the meter for data centers, better than shutting down.
Trump talks to business leaders, listens to top guys like Anthropic, synthesizes info. Gave flowers to Anthropic guys as brilliant, working on a deal. Balanced view: smart guys, great product, leftwing but workable, tried to tell Pentagon what to do which Pentagon didn't like. Trump likes high IQ people, loyal people, good on camera.
SpaceX signed huge deal with Cursor, AI coding startup that defined the category. xAI and Cursor building world's best coding and knowledge work AI. SpaceX will buy Cursor by end of 2026 for $60 billion ($10 billion more than rumored raise) or pay $10 billion collaboration fee (Bloomberg calls it breakup fee). Cursor run rate $2 billion end of February, expect $6 billion by end 2026 (tripling). SpaceX projected revenue $22-24 billion in 2026, accretive. SpaceX IPO targeting $2 trillion valuation, ~80x topline revenue; Cursor at 30x.
Cursor built off Anthropic's LLM initially, any LLM usable, but March released proprietary Composer 2, ranked between GPT-4.5 (5.4) and Opus (4.6). Elon has 550,000 GPUs in Colossus, scaling to 1 million, then to space. Infrastructure matters, peanut butter and chocolate combo. Prediction: moves SpaceX/xAI/Cursor to front of coding leaderboard in 12 months.
Chamath (SpaceX shareholder via Starlink acquisition): Deal structured to not stale S1, avoid rewriting disclosures/risks. Deal effectively done. SpaceX at $1 trillion now, $2 trillion later, stock-for-stock at $60 billion is 50% discount. Gets compelling coding model/service (immediate revenue), hard-fought RL patterns, cracked team. Grok had steep output token discount leading to high usage/experimentation, then normalized; excess Colossus capacity used to jujitsu acquire most valuable third-party wrapper service at effective $30 billion. Smart deal.
Sacks: Companies complementary. Cursor strong in coding, xAI brings compute/foundation model. Cursor competed against generalists (OpenAI, Anthropic) now vertically integrating (Cloud Code, OpenAI Codex). Alliance with xAI makes sense, brings training data, enterprise clients, coding experience; accelerates xAI.
Likely dump Kimi K2.6 (Cursor Composer 2 uses Moonshot model); no way Elon pays $60B without running on Grok. Cursor's strength: toggle choice of models, well-built IDE (best UX over Codex/Copilot). Developers want choice.
Value in AI realized by writing software; agents are quickly spun-up apps but inefficient, need engineering. Enterprises spinning million agents (Amazon example: wasteful resources, redundant data/stores/API calls). Need strong dev environment/IDE like Cursor to centralize, engineer harnesses.
Software engineers win arms race. Optimize token bills (enterprises spending more MoM); use open source/less expensive models for mundane tasks, frontier for frontier. Need middleware to route models efficiently.
White hot center: cyber. Mythos (Anthropic, 10 trillion param model, compute constrained, maybe not commercial) wakes up potential for AI in cyber offense/defense. Race to train dedicated cyber models (Mythos-comparable, lower token cost) for IT/CISOs fearing AI-powered hackers; hottest market in 3-6 months.
Polymarket: SpaceX acquiring Cursor 74% chance, SpaceX IPO by end August 80%. Deal structure motivates Cursor team (wants $60B over $10B fee, but compute not cheap). Some xAI co-founders left post-acquisition, now on SpaceX stock.
Elon tweet (5 weeks ago, 50.8 million views): "xAI was not built right first time around. So is being rebuilt from the foundations up." Like Tesla. Magnet for talent; aligning to SpaceX philosophy. Easier with all in same building.
Thoma Bravo (Orlando Bravo at All-In Summit last year) handing Medallia (SaaS customer experience) to creditors. TB acquired 2021 for $6.4 billion all-cash at market top, incurred $3 billion debt. 2021 revenue $470 million, 20% growth. Debt servicing costs tripling from $100 million to $300 million/year. Blackstone et al. refused lifeline; wiped out $5.1 billion equity.
Chamath: Thoma Bravo well-run, bonkers returns, Orlando excellent. Likely got equity out via dividend recaps, positive return. PE playbook: transform broken assets with operating partners. Suspect core rot unfixed, downside protection achieved. Issue for bondholders, maybe borrowing costs.
Medallia: customer support/experience management (post-cruise surveys for insights). Alternative: AI agent cheaper/easier/custom than SaaS.
Sales team 18% of target. Hard time sourcing new customers, higher attrition. Predictable NRR (e.g., 118%) allowed leveraging; now agents crush vertical SaaS sales.
Link to Kevin Warsh Fed chair hearing: AI deflationary effect. SaaS apocalypse as business phenomenon or lens on deflation in business operating costs.
AI delivers on its deflationary promise by drastically reducing costs to run a business, eliminating the need to pay premium prices for SaaS products. Warby Parker highlighted AI's deflationary evolution, expecting unprecedented productivity growth, but uncertainty remains on labor market response and potential dislocations. Economic deflation means some businesses see revenue decline; if levered with debt expecting perpetual growth like SaaS top lines, debt impairs, creating adverse ripple effects. However, this deflation from AI innovation—not cost-cutting or contraction—leads to overall economic expansion as savings reinvest into growth. Cutting 50% of SaaS budget allows reinvestment in business expansion.
The way that we've been measuring inflation is wrong... you can do a survey of any household and they'll tell you, my god, everything's so expensive. So all of the indices... are completely misrepresenting what the average American is actually feeling.
Warby Parker criticizes Fed's inflation measurement and wants rethinking of interest rate policy amid deflationary pressure and AI productivity gains.
SaaS losses benefit token dealers; startups, as tip of the spear, build custom tools and dashboards. SaaS product index shows Salesforce down 32% in past six months, ServiceNow down 54%, Snowflake down 43%, Adobe down 33%, Figma down 67% post-IPO pop. Tokens now eat SaaS and software business.
Historically, software exits were IPO or M&A; private equity introduced debt-financed buyouts (1/3 equity, 2/3 debt), attractive for stable cash flows. SaaS mature businesses offered predictable cash flows for debt financing.
Pricing now attractive for private equity: public SaaS at $1B ARR, 20% growth, 80% gross margins trade at 3x ARR—buying dollar for 50 cents (used to be 13x). Salesforce down 9% today post-Medallia announcement. Category leaders now at 3x ARR.
Bearish side: private equity needs predictable cash flows; SaaS net dollar retention dropping from 120% to 80% due to customers attriting to tokens or bespoke software disrupts model.
Startups disrupt by pricing at 10% of value delivered ($10 value, charge $1). Stacking venture capital and growth equity raises return hurdles (clear $300M-$1B pre-money plus 15-20% IRR), forcing price increases that invite competition and targets. Private equity layers billions in equity/debt, must raise prices to service—can't cut to gain share.
SaaS unit costs out of control (from 10% to 30% of value); customers plan to cut renewals by half, 2/3, or 75%. Headless products via APIs/agents eliminate per-seat pricing: need 2 seats not 50 when agents interact. Multiplied across companies, creates falling knife.
Salesforce at $140B enterprise value on $15B free cash flow trades <10x FCF—potential bargain. Marc Benioff buying back stock; his headless product announcement smart, going fully headless unlike Workday's AI tolls.
Historical multiples suggest Salesforce bargain at 10x FCF, but cash flows cut off at year 5-7 compress to 3-5x regardless of quality—discount years 7+ to zero. Free cash flow war chests provide massive optionality (Salesforce, Apple, Meta, Google, Uber weather storms).
Founder-led companies like Benioff's Salesforce maneuver best—bet it all, burn boats. Managers maintain old models (charge fees). Track per-seat license unit price trends/inflation in 10-Ks to predict deaths.
Benioff mastered prior waves (social, mobile, big data); high odds he succeeds with AI.
Venture debt fine for private equity experts, but founders treat like equity, forget repayment, face covenants reducing maneuverability—bank oversight hinders pivots. Debt creates fixed payments, increasing vulnerability to disruptions.
Equity preserves optionality, aligns interests; hot startups always attract more VC. Venture debt as last runway leads to rug-pull—banks can't afford losses at 8% returns like VCs with 10-1000x upside.
Debt holders extract: double interest, demand warrants. Chim's experience: $420M credit line collapsed in market disruption, Credit Suisse implosion risk—worst professional moment; now anti-debt rule.
Buffett has this line... this is how smart guys go bankrupt is they take on debt. Debt equals prison.
Governments socialize debt. 1950s corporate pensions shifted to 401ks (defined contribution) to avoid over/underfunding bankruptcy risk; governments retained pensions via unions, now bankrupting states. California public employee pensions: $450-500K salaries, overtime spikes, 2/3-1/2 pension.
Solution: forced superannuation fund or 401k-style accounts. California government grifts: homeless industrial complex ($12B/year, homelessness rises), racism industrial complex.
Nick Shirley uncovers California union salaries/pensions; could whistleblow for billions but does public good via citizen journalism, shaming old media (Bari Weiss/CBS follows). Old media protects government waste/fraud/abuse unlike past (Pentagon $900 hammer, Tyco $6,000 umbrella stand). Elon Musk vilified for waste-finding.
Productize waste/fraud/abuse elimination (20-30% of system) to cut taxes, win 2028 elections.
Liquidity sold out for All-In Summit (5th edition, LA, Sept 13-15); apply at allin.com/events.
Political panel: Senators Dave McCormick and John Fetterman (PA) on stage, left/right perspectives; dress code joke on Fetterman's hobo style vs. McCormick's Brioni suit.
John Ternus, 25-year Apple vet (iPad, AirPods), Poly Market favorite, bold decision-maker. Tim Cook squeezed nickels from Steve Jobs products (iPhone, Apple TV, Watch) for decade. Product person needed as tools stale: Siri degraded, AirPods degraded, system not innovation-built.
Freeberg
Freeberg agrees Apple should ring more profits. Hopes for new CEO: build ubiquitous AI layer like advanced Siri across all devices—personalized, knows user via email, calendar, music tastes, home connections; engage via natural language. Suggests buying Whisper Flow to replace Siri team, as Siri can't spell "Polyhapatia" or "Calacanis" after 20 years and billions in iPhone sales.
Sachs: Apple needs AI answer, empower Siri with user-chosen models (ChatGPT, Grok, Claude) for customization and storage control. Praises Tim Cook's 15-year run: market cap up over 10x, revenue from $100B to over $400B, shifted to higher-valuation services, released many products (none iPhone-scale), no scandals, beloved brand due to privacy focus. Nurtured good presidential relationship.
Chimath: Tim had impossible job succeeding Steve Jobs, top steward CEO. Allocated capital/people well: appropriate R&D, divested Intel, spun silicon line (useful for AI/open cloth), light acquisitions, capital efficient. Returned money to shareholders (shrunk share count ~44-50%), unlike Jobs (zero returns, kept cash). Charts show this contrast.
Missed opportunities under Cook: lightweight glasses (not 17-pound Vision Pro) pairing with phone for videos; killer Siri; Perplexity-like search; self-driving car; TV set. Apple stores should've sold glasses, car, TV. Apple TV produced great programming. Lost innovative soul after Jobs and Jony Ive; shifted to profits/iteration. No notable acquisitions. Vision Pro not mainstream.
Future for John Turnis (likely new CEO): AI rips open device canvas to heterogeneous world (pens, orbs, glasses); hard to escape iPhone addiction with high margins. Expand wearables (AirPods, Watch); AI synchronicity across eyes/ears/watch/phone/desktop. Add robotics (Roomba, Optimus) and consumer car—Steve Jobs would've executed.
Sachs: Succession mirrors Disney—Steve warned Tim not to cling to his vision like post-Walt Disney. Roy Disney (COO-type) stewarded 5 years post-Walt; then uninspired CEOs until Eisner (1984) revitalized. Tim like Roy: 15 years, 10x market cap.
Question for John Turnis: 1970s Disney funk or 1980s revival? Eisner used "vault strategy" (re-release IP every 7 years: Bambi, Snow White). Iger bought Pixar, Marvel, Star Wars using balance sheet/distribution/brand. Bold acquisitions accretive vs. buybacks/cash distribution.
SPLC indicted (not guilty yet) on 11 counts wire fraud/money laundering. Allegation: 2014-2023, used hidden offshore accounts ($822M total) to funnel $3M donor money to paid informants infiltrating hate groups (KKK, Aryan Nation, neo-Nazis, Proud Boys, Oath Keepers). Mission: catalyst for racial justice, dismantle white supremacy.
SPLC labeled Sam Harris "hate adjacent." Post-Charlottesville (2017 hoax on Trump "very fine people" quote, Biden's run inspiration), revenue spiked from $58M (2016) to $136M, stayed elevated.
F37: SPLC informant in Unite the Right online leadership chat, attended event, made racist posts under supervision, coordinated transport; paid >$270K (2015-2023).
Additional: funneled >$3M to violent racist groups (KKK, American Nazi Party, Aryan Nation, United Klans of America) to organize protests, then fundraise on "racism problem." $270K investment yielded $81M post-Charlottesville return. Used fictitious entities to hide from donors (e.g., Hollywood celebs).
SPLC claim: monitoring via informants, not planning; hid accounts like FBI would. But paid leaders (not moles) to foment activity, not just inform; concealed from donors.
Chimath: Nonprofits cosplaying overlords; dismantle them. Donors should sue, reclaim $822M offshore. Future donations need full audits—may support opposite causes (racism, anti-gay discrimination, anti-trans) to create narrative, amplify via media, raise money/power.
Freeberg: IRS 501c3 for charitable/religious/educational/etc. activities; most nonprofits don't qualify, hide money commercially. Not blue/red—stop tax-exempt status, gov't funding, deductions for nefarious acts. Libertarian ok with it, but not subsidized. Theme: audit everything (gov waste, NGOs, chemicals like Dao linked to cancer 30 years later). Disrupts society not knowing race truth; fueled cancel culture (e.g., host's 2014 "most offensive tweet" by Vice on check journalism, modest impact).
SPLC put Sam Harris on list; Vice tried canceling host for saying race doesn't play role in hiring.
Panelists discuss resilience against cancellation attempts in journalism and public discourse. One speaker notes experience in vertical journalism and being targeted for cancellation. Jason claims to be 'uncancellable,' with agreement that all have proven resilient. Debate shifts to whether human rights organizations like Human Rights Watch create abuses they claim to fight.
Speaker recounts work at Amnesty International, focused on Universal Declaration of Human Rights issues like torture, imprisonment, and censorship. Notes these groups drifted from core mandates to secure funding, taking on transgender rights and censorship campaigns, e.g., targeting Sam Harris over Charles Murray and The Bell Curve. Estimates 95% chance many organizations engage in nefarious activities to fabricate abuses. Another suggests 50/50 odds for Amnesty International due to co-optation.
SPLC might have had a great origin story, but now...
SPLC (Southern Poverty Law Center) faces grand jury indictment, previewing evidence. Clarification: not a full investigation, but grand juries readily indict ('ham sandwich'). Accusations include funding KKK and American Nazi groups via confidential informants. Panel urges letting case proceed in court.
Contrast between businesses (market feedback via revenue/profits) and NGOs/nonprofits (fundraise for activities, prioritize perpetuation over results). Over time, activities matter less than fundraising justifications to retain jobs/donors. Government-funded NGOs lobby for more funds, spinning programs as successful regardless of efficacy.
No one in an NGO or nonprofit ever declares victory.
Questions why SPLC doesn't focus on southern poverty or racism origins instead of shifting goals.
Civil rights noble post-segregation/Jim Crow; groups succeeded but never declare victory. Obama's 2008 election symbolized non-racist country (highest office open to all). Instead, goalposts moved to anti-racism: equality of outcomes, not opportunity. Anti-racism demands distributions match population demographics, akin to identity socialism/communism, repackaged to avoid rejection.
It's not good enough not to be racist. You actually had to be anti-racist.
SPLC partnered with FBI to infiltrate KKK/Nazis historically, but unnecessary in 2025/2026.
Colorectal cancer now third leading cancer; 80%+ rise in under-50s over 20 years (historically age-related, spikes over 70). Barcelona team analyzed epigenome/gene expression in tumors from Cancer Genome Atlas (federally funded). Compared under-50 vs. over-70 patients, linking environmental triggers via RNA (genes on/off for protein production).
Adjusted for lifestyle (eating index, overweight, alcohol, birth weight, gender). Picloram (pesticide, orange-highlighted row) strongly associated with under-50 cases. Developed 1963 by Dow Chemical, synthetic auxin (plant hormone) causing overgrowth/death. Used on rangeland/pastures (cattle grazing), roads/railroads, industrial sites, highways/utility corridors. Persistent (>1 year), enters groundwater.
EPA last safety study: 1995. Epigenomic changes from long-term exposure trigger cancer (not immediate toxicity).
Should prompt EPA review/ban and new assessment via epigenomics for food/environmental chemicals. Picloram off-patent, generically made (likely China), ubiquitous under brands.
County-level data (CA, CT, GA, IA, NM, UT, WA) from Pesticide National Synthesis Project: higher picloram use correlates with 3x odds ratio of colon cancer (strong signal).
Combines big data, increased testing, AI for re-evaluating compounds. PCAST role in fundamental science: National Cancer Institute's $100M genome atlas (now ~few million/year maintenance) enables such analyses. RNA sequencing reveals epigenome changes. Advocates government systematically scan/eliminate cancer-causing chemicals from food/industrial supply.
Humor on planetary focus, but emphasizes urgency lacking media attention. Paper from days ago warrants awareness.
Episode 270 wraps with thanks to Chamath Palihapitiya, David Sacks, David Friedberg. Casual sign-offs: 'Love you boys,' 'Let your winners ride.' Fragmented post-credits banter on open-sourcing, fans, dogs, orgy joke, merch.
Keep All-In Podcast in your library
Save the videos and channels worth coming back to, and find them again in one place.
