Travis Kalanick & Michael Dell Live from Austin, Texas
In a Nutshell
Travis Kalanick unveils CloudKitchens after 7 years in stealth, now rebranded with a mission to automate physical industries like food production, mining (via Pronto acquisition), and transport, treating atoms like bits in a "physical computer" framework (manufacturing as CPU, real estate as storage, logistics as network). Michael Dell shares Dell's AI data center surge to $50B revenue, Texas's pro-business edge over California, and champions the Invest America Act—providing $1,000 government-seeded investment accounts for all U.S. kids under 18 (expanding to newborns from 2027), already claimed by 4.5M, as Social Security 2.0 to foster ownership and compound wealth directly into families. Both emphasize physical AI's imminent breakthroughs, capital as a competitive weapon, U.S.-China collaboration, and bold relocation to Austin for innovation amid California's decline.
These notes were generated by AI and may contain inaccuracies.
Michael Dell reveals he was an angel investor in Travis Kalanick's company. Travis has been building in stealth for seven years. Employees couldn't list the company on LinkedIn, even salespeople and recruiters; parents thought they worked for the CIA. Dell invested secretly, happy to be on the cap table. Company came out of stealth today. Operates in 30 countries with generic names: CloudKitchens in the US, Kitchen Valley in Korea, Namah in the Middle East, Casino Cocinas in parts of Latin America, four in China.
Thousands of employees lived in stealth mode on LinkedIn, operating on hard mode but found it fun. Massive density of stealth startup people in Los Angeles.
Company has been acquisitive. Announcing new branding today. Not just renting kitchen space. Mission: physical automation to transform industries and move the world. Expanded beyond food computers to mining (physical mining for atoms, more productive mines to power Earth's industries) and transport (Wheelbase for robots, including mining equipment).
Acquired Pronto (San Francisco-based, automating mining equipment), deal inches from closing.
Travis's philosophy from Uber days: digitizing the physical world, treating atoms like bits in an atoms-based computer.
- CPU (manipulates bits) → manufacturing (manipulates atoms)
- Storage (stores bits) → real estate (stores atoms)
- Network (moves bits) → transportation/logistics (moves atoms)
Company name was obtuse and boring: digitized real estate. First computer: food computer (manufacturing, real estate, logistics for food). Mission: infrastructure for better food. Goal: meal prepared and delivered as efficient as grocery store cost. Do to kitchens what Uber did to cars.
Uber had existing roads and unused cars; food lacks high-capacity infrastructure. Restaurants have 20% capacity utilization via Uber Eats/DoorDash. Needs industrial production and logistics like Amazon warehouses for e-commerce.
Mining challenges: surveying locations, going deep (energy cost scales with distance squared). Automation unlocks capacity, makes existing mines more productive, justifies inhospitable/remote locations (less labor, safety issues). Robots monitored remotely. References Boring Company for tunneling.
Rare earths are not rare but uncommon; issue is land access and regulation. Automation enables more places.
Physical AI stack includes land development, chemistry, manufacturing. Tesla is the Google of this era (like Google dominated 2000s startups). Many opportunities beyond Tesla/Waymo (Austin shoutout).
Rumors Travis not done with self-driving. Players: Waymo ahead with existence proof but lacks manufacturing, scale, urgency. Tesla on hard mode (vision without other sensors), timeline uncertain (ChatGPT moment for vision?). Uber had autonomy project, now different strategy. More bark than bite overall; no other standouts.
Vision-language-action models not ready for prime time. Energy in research/implementation. Hope for OS like Android for these models (voice commands → identify objects → act).
ChatGPT 3.5 → 4.0 world-changing moment; similar for physical AI imminent? ML now conversational (e.g., driving agents: "yo look out over there" → response). Language as compression mechanism (100W human brain vs. AI power plants). Waymo uses 100x more energy than human driver.
Humans still GOAT at language. Early physical AI processes all data (e.g., irrelevant clouds); language compresses efficiently for agents/safety systems.
Travis moved to Austin in December (Lake Austin place since 2021, waterskier, 15 weekends/year). Loves it: bigger homes for half the cost, dope people/food, future-building vibe. Lived in NY/LA/SF; Austin feels like home, diverse industries.
Grew up in LA (parents born/bred there). California heartbreaking, getting weirder (truth/justice deteriorating). Exodus: Elon first, then tech, now Joe Rogan, comedians, music, NYers, restaurateurs. Dozens of Lake Austin startup CEOs want to leave but tied by employees/offices/facilities.
Advice: painful moves often not as bad; just do it. Setting up team/office on Lake Austin (jet ski to work). SF nostalgia but busted: empty $400M/mile bike/bus lanes, Market Street ruined (empty, painted red for virtue signaling).
Group raising $500M for tech/business coalition to fight in Sacramento. Tech was libertarian (stay out of politics); no longer. Ballot initiatives powerful. Target DAs not enforcing crime (e.g., Chesa Boudin recall led by All-In Pod/Sacks; LA's Gascon). Truth and justice as society's immune system; suppress it, social ills flare. Homeless industrial complex like Batman villain (running city for criminals). Police officers quitting from PTSD, unable to serve.
Squirrel jumps trees on 0.1W; biology perfectly efficient. Robots: 1200W to walk 4 feet due to compounded inefficiencies (metal, motors, actuators). Need evolution in hardware (actuators, materials) beyond software. Questions going down humanoid lane.
Evolving actuators, materials, and other components will make physical AI work and scale. Questions focus on humanoid robots, with obsession over hands like in Terminator 2, a critical part. The Achilles tendon is the real vulnerability in these machines—where the action is. Mechanical engineers have long dealt with actuators and electromechanical interactions. In food machine space, can open a paper bag and put a bowl in without tearing it. Less expertise in proprioception—senses for awareness and touch. In mining, no need for delicate tasks like playing tennis; some equivalents exist. Humanoid demos online show rapid improvement in dancing and martial arts. Focus is on specialized, gainfully employed robots, not humanoids.
Big opportunity for venture money and research in material science for physical AI stack, manipulation, and related areas—massive space. Software advances require hardware to catch up. Significant investments already happening.
Capital as a weapon pioneered at Uber: removing capital from the table gives advantage over competitors. Others now follow this playbook at larger scale, post-stealth. Capital enables execution; as strategic weapon, it's essential. In early Uber, without capital, even superior app loses—e.g., Masa investing $1B in competitor drops 20% market share overnight. Raising capital better than competitors is a world-class competency.
Theories suggest Middle East capital (Dubai, Qatar, Saudis) closing to US due to Iran war. Middle East business planned IPO in January; Saudi market dropped 20% over two months, partly from falling oil prices. In KSA, everyone wants oil prices up. Not raising money now amid war; sees news like everyone else. Optimist view: conflicts temporary, like tariffs—not end of world. Progress, abundance, golden age coming via AI, physical AI, massive productivity gains. Shock and awe leads to steady state; hopes for deposing evil dictators in Iran for stability.
Hopes for big trade deal with China. China tour shows advanced manufacturing base: Shenzhen now one-upping Singapore (from feeling like humid Kansas City 50 years ago). Xiaomi and scrappy, hungry players everywhere. 2008 Beijing Olympics opening ceremony signaled takeover ambition—it's happening. China fiercely pursues truth and progress; best ideas win. US should step up game in friendly competition, not like 90s Detroit Pistons—play like adults. Employee in China (largest built kitchen builders there) has American wife; hard to bring him to US despite decade-long collaboration. Wants good relations, normalcy for mixed-citizen couples. Immigration effed up badly, but good migration brings innovators for American dream—e.g., from Freiburg. Keep eye on ball, give China run for money.
Michael Dell, Texas native born in Houston, started Dell Computer in Austin dorm room at UT (Adobe? [likely typo for Dobie]) with $1000, 42 years ago, 10 days before finishing freshman semester. Now ~$140B revenue this year. Compounds over time: start small, keep adding.
Texas has low-tax, pro-growth, progressive business climate. Texas economy outperforms US without it. Austin nearing top 10 US cities; Texas has 4 of top 10 largest cities. 1/10 US children born in Texas. More NYSE companies in Texas than New York. University of Texas in Austin as wellspring for companies, innovative pioneering spirit. Other cities (New York, LA, Bay Area) decelerating, un-American; disparity widening. Migration stats show Texas attracting enormous numbers; life better here.
Elon had great experience building Tesla Gigafactory in Texas—they let you build. In California, couldn't; Fremont factory was ancient retrofit. Texas enables data centers: massive AI infrastructure buildout needs power, land, permission to build. Advantage in West Texas towns with low population, land, power—middle of nowhere ideal.
Enormous token demand drives AI data centers globally. Dell introduced first H100 server weeks before ChatGPT announcement. AI business: $2B → $10B → $25B → $50B this year. Phase change in computing: 60 years of calculating/computing to machines that think and help us think. Models improving, always demand > supply. Not just hyperscalers/cloud; 4000 enterprises get Dell AI factories. Sovereign AI (e.g., Palantir): protect data, bring AI to it.
Fortune 100 companies bought AI servers early, built own models or used open source, algorithmic trading, machine learning derivatives. Needed help; Dell launched Dell AI factory product ~2 years ago.
Administration implemented accelerated depreciation rule allowing 100% write-off in the year of purchase for data centers and infrastructure. For $100 billion spent this year, deduct 100% this year, paying far fewer taxes. This 10-year deal accelerates investment.
Question on hyperscalers vs. end-user ROI: 20% ROIC starting year one? Dell sees use cases with 20%+ ROI immediately in productivity/efficiency, but requires process work.
Company processes reflect tools available when created. To adopt AI:
- Step back: assess tool improvement trajectory and desired outcomes.
- Simplify/standardize processes.
- Consolidate tools and data.
- Apply technology tops-down.
Requires wholesale re-architecture, reimagining company operations. Dell told team 3 years ago: new competitor in every business—faster, more innovative, lower cost—will put us out of business unless we become that company.
Biggest benefit: speed in applying innovations. Dell infrastructure business grew 73% last quarter; guiding 100%+ this quarter.
Lived through PC revolution, client-server, network, internet, cloud, mobile—each caused massive disruption over decade or two (e.g., typing pools, mail rooms abstracted by networked PCs). AI shift 5x faster (quarter = year).
Only 10-15% of large companies figured it out; rest fumbling. Tendency: rush to "do AI" to impress boss/board without strategy.
Barrier to adoption: not technology, but culture, leadership, courage, willingness to change. Heart change hard, uncomfortable, especially if bonus tied to not messing up.
Internet analogy: Macy's.com transitioned better than Sears; internet natives (Amazon, CSN stores, Wayfair) blew them out.
AI natives will disrupt every industry. Stripe's Collison brothers: 2025 cohort companies grow 4x faster than 2018 cohort due to new tools.
Incumbents have expiring assets (brands, balance sheets, customers) unless they change quickly. Must be bold to survive/thrive, prepare for 2030s.
AI enables doing way more: solve more problems, accelerate scientific discovery, invent new things. Not just same things with fewer people.
Original view: all intelligence in hosted proprietary cloud data centers. Reality: all of the above.
- Tons of inference at the edge in industrial, natural resources, manufacturing, retail, logistics—growing fast. Dell makes embedded equipment; telcos too.
- Inference close to data creation.
Public cloud love until bill arrives—costs more. Lowest cost token generated where data is: on phone, PC, embedded equipment.
Dell perspective: 10,000 customers embed Dell products (medical devices, security, hospitals, industrial plants).
Desktop full circle: 2-3x computer size. Hobbyists ordering powerful desktops (dell.com) for local models, like early PC days ($4,000 in 1980s down to $500-800 laptops).
Desktop revolution back: protect data/skills from cloud (Sam Altman, Google). Not everyone $10k desktop, but Dell/Hugging Face portal qualifies open models (Google Gemma, OpenAI, NVIDIA Nemo/Tron) on all machines. Enormous open source progress.
20% audience has set up OpenClaw (autonomous agents). Inside companies: more agents with security (authenticate/validate, controls).
No personal AGI/superintelligence definition. Latest models (Gemini 3.1, Opus 4.6, OpenAI 5.4) hit threshold of tremendous quality.
Teams accomplish in days/2 weeks what took months/9 months. Reinforcement learning + private dark data untapped.
Auto research: retrain standard model on private data, keep private, build org advantage.
Starting AI-first company: reimagine business knowing tool trajectory to 2027-29.
AI most unfavorable term (worse than ISIS, Democrats; ICE better). Issue: presents human-like; call it "linear algebra, matrix, statistics"?
AI helps healthcare, longevity, education, housing/food costs.
Optimistic: technology cycles have disruption/inequality, but do more with tools—accelerate education, science, healthcare, energy, unsolved problems. Amplifies human potential, extends frontier.
Mostly software: can't stop open source models on home computers for medical/legal advice (NY proposed law banning AI medical advice—also anti-books).
Michael Dell/Susan giving $250 to 25 million children (ages 2-10) in zip codes median income ≤$150k = $6.25 billion. One of greatest philanthropic gifts; kids sign up to claim (100k+/day now).
Idea from Brad Gerstner (2021); discussed prior admin (no action). Passed as miracle; thousands companies matching government contribution.
Brad: Worked 4 years; All-In pod momentum. April 2025 tariff strife, reconciliation bill window—5 days to draft Senate bipartisan, get to Oval (text to DJT). Trump respects business leaders.
Not red/blue—red-white-blue. Reconnects 70% feeling left behind to American dream, defends ownership society/capitalism (under-40s <50% favorable view).
Dell foundation focused on children in urban poverty.
Central focus of the foundation is on this initiative, though it does local community work in Central Texas. Idea seen as a platform for directly giving to targeted people. Initially considered doing it just in Texas first, but company success led to going bigger.
What if 10 more Michael Dells show up? Dozens of people could make equal or greater gifts. Many hands make light work; 1,000 people can make significant gifts. This is becoming a movement, not a moment. Much queued up.
Michael Dell and speaker chair the Invested America Giving Committee; ambitious, knocking on doors, having conversations, texting people.
Every child under 18 is eligible to claim their account. No restrictions like kids born between '25 and '28; legislation creates accounts forevermore. Every child born in America starting January 1st, 2027, automatically gets a Trump account at birth, stapled to their Social Security card. $1,000 has to be reauthorized every four years, but accounts persist.
This is Social Security 2.0, biggest change to the social contract in 50 years. 3.7 million kids a year get an account that can compound—a 401k from birth.
Not just billionaires: companies donating stock on IPOs into accounts; wealthy people; states; moms and dads; corporations. Over 15 years, estimate to move $5 trillion into pockets of families that would have had zero.
Leadership Michael showed in helping get the meeting that passed this into law; moments like that either happen or don't—no law without it, doesn't change kids' lives.
Two key points: $5 trillion through government programs would get incinerated due to crony structures. Giving directly into accounts avoids that.
Can replace Social Security—fake Ponzi scheme—with defined contribution accounts where everyone has one, owns a piece of their future. Payroll tax deduction goes into account to buy company pieces, with some choice possible.
4.5 million kids have claimed accounts; on trajectory to 10 million by July 4th (250th anniversary). Parents and kids see app like Robinhood on phone: "You've received your $1,000 or $250"; shows bits of Nvidia, Walmart, Dell—decomposes S&P 500 into parts so they see ownership in America's upside.
Use money to buy via app. Moms and dads double-click Apple Pay $5-10 into account. Send QR codes to friends for birthdays, Christmas, bar mitzvahs—friends add. Companies add. See it grow, unlocks human potential: "I'm in the game, I have a shot."
Biggest crime of Social Security: sacred promise, but government requires 10% of earnings into it—a black hole nobody sees. Social contract evolution in Industrial Revolution kept country together.
Imagine if that money was in an account with your name: see it grow, know exactly what's there, get excited to add more, have choice. That's the possibility; will end up there.
Brad adopted Indiana; Ray and Barbara Dalio adopted Connecticut; many more. Super easy: add 100 kids, adopt a zip code, school district, town.
Give it up for one of the great entrepreneurs; incredible pledge—doing all in.
Keep All-In Podcast in your library
Save the videos and channels worth coming back to, and find them again in one place.
