Intelligence for Heavy Machines: Qasar Younis on Why Physical AI Will Change Farms, Mines and Roads First

Asked what the next few years look like if things go well, Younis starts with the industrial revolution. It brought child labor and monopolies, he says, but also benefits that are hard to imagine living without: broader access to health care, material goods, and basics like heating and cooling a home. He points to letters from German prisoners of war in 1940s America who were stunned that towns were lit up, when most German towns lacked electricity.

In his optimistic version, things only wealthy or well-connected people enjoy today become available to everyone. A personal coach tailored to you rather than a generic chatbot is one example, and he thinks solving problems like cancer will be tied to this AI boom. He expects net suffering in humanity to fall significantly, and calls himself a fundamental optimist about technology.

For physical AI in particular, he focuses on mobility. For someone who is disabled or cannot afford a car, near-free self-driving would be transformative, and in a place like Rwanda where the nearest hospital is two hours away it matters in a very real way. He adds that much of the negativity about AI comes from people already living very good lives, and recalls that his family in Pakistan once could only write letters because long distance calls were so expensive.

Younis separates two things: public anxiety about AI and investors reacting to specific stocks. On anxiety, he says the core root of fear is misunderstanding. He cites videos of chatbots failing to grasp which way up a cup is, and notes that the nunchuck-wielding humanoid videos show pre-programmed motors, reportedly at a cost of $15 million for the video. People fill the gap with visions of sentient machines, whereas a welding robot in a car factory, which is far more impressive, causes no worry because the gap is understood.

His advice is to drop the emotionally loaded word AI and just say technology, learn how it works, and then actively push it toward good uses, whether as a founder, employee or citizen. Every technological shift, such as WhatsApp, damages some companies and people, so leaders should steer the funnel deliberately.

On the sell-off, he offers a guess rather than research. Drawing on his Harvard MBA, he says hedge fund managers are smart and hardworking but lack the deep edge outsiders assume. When a consultant builds a Figma-like app in a few weeks, investors price in the risk that vibe coding replaces billion-dollar companies, even if the real product is far deeper. He says that is a separate matter from society's anxiety, and that at least one calibrated investor told him it was time to buy.

Younis notes that self-driving cars are robots too, and that whichever way you slice the available statistics from self-driving companies, they are far safer than human drivers. He predicts that in 20 or 30 years we will look back on human driving the way we look at child labor today: a normal thing that now seems unthinkable. People drive tired, stressed or under the influence and then jump into a car.

In the United States, he says, over 30,000 people will die in the next year from car accidents, and it is easy to let that number wash over you as a statistic. Talking to a family that has lost someone makes the human cost real and the fear of AI robots fade. The host adds that he now uses Tesla's self-driving all the time and finds it far better than a few months ago.

The argument extends to trucking and mining. When people hear about self-driving trucks, they immediately ask about driver jobs, but Younis says there are not enough people who want those jobs and that people die in truck accidents. He asks listeners to notice that their monkey brain is wired to read a rustle in the bush as a snake, and to look at what the best evidence says: nobody is clamoring to work in a remote mine.

Asked how close we are to robots in daily life, Younis frames it as a spectrum that already includes robot vacuums and coffee machines. He asks the host to imagine being in 2006, just before the iPhone: a year or five might have been the answer to when a Star Trek phone would arrive, but nobody could have predicted Instagram or Uber. The pace can be fast, he says, but the form factor is hard to pick, and what matters is hardware being generally available.

Early use cases will be those with the most bang for the buck, such as a car that drives itself or a mining vehicle. The machine itself was engineered over 50 or 60 years, so adding intelligence leverages all that work. He says our brains love the humanoid idea because we are monkeys, but the pragmatic path is putting intelligence into things that already exist, with new applications emerging in five to seven years.

Every car company, he says, is working on something like Tesla's FSD, which he describes as an L2++ product with few sensors, no high-fidelity maps and cheaper compute. The Waymo approach, with many sensors, heavy compute and maps, works better in constrained geographies. Younis compares the moment to navigation systems that once cost thousands of dollars before CarPlay and Android Auto made them free, and expects autonomy to become close to free, as pricing pressure in China already hints.

He says the same logic holds in construction, mining and defense: teaming with a big physical machine that acts like an agent unlocks large productivity gains. He likes Moltbook and the OpenClaw revolution but notes that it is still a very small slice of society. His test is the Detroit airport gate, where hardly anyone would know what OpenClaw is, so the broader impact of AI will be in farming, mining, construction and self-driving trucks.

The host raises a theme from recent episodes: AI is coming just in time to help with aging and shrinking workforces. Younis agrees, saying these industries need autonomy and it cannot come soon enough. The average farmer is in their late 50s, so many will retire within ten years, and he says the same holds in every vertical.

His hypothesis is that the people are still there but the trade-off no longer seems worth it. In the 1980s and 90s, a long-haul trucking job meant a father away for days or weeks, whereas today the same working-class family can drive for Uber or DoorDash and turn the app off to pick up a child. He does not believe in a future where an entire industry is suddenly automated, because replacing it with robots is still too complex.

To someone anxious about AI, he recommends reading history books about the industrial revolution, since lots of people wrote about it from both sides. He warns that pumping the brakes on technology hurts the lowest end of the labor market most. In his view American growth over the last decade has come from Mountain View and the Bay Area rather than Detroit, so a future that ignores average workers is a big problem, but slowing frontier technology is not the solution.

Younis says his company recently read *House of Huawei*, and the lesson is that Huawei is not a profit-driven company like Apple. About a quarter of its couple hundred thousand employees are Communist Party members, and he describes it as an extension of the state, noting that the name means China's ambition. Americans, he says, impart their own understanding of markets onto China.

So he suggests thinking of OpenAI as competing against the Chinese government rather than against DeepSeek, and Apple against the government rather than Huawei. Second, a goal that is not profit allows spectacular research but is not sustainable. He contrasts celebrated Chinese EVs with Rivian, which makes great products but loses money and is punished by public investors; if America simply built great EVs without caring about profit, he thinks the results would be impressive too.

On whether China can succeed only if America fails, he says open and free markets let everyone succeed, but China's markets are not open and free. That does not require an antagonistic relationship, does not make China incompetent, and does not mean it deserves less attention. He adds that the nuance will never reach the average person in a Detroit airport who only hears that China is bad.

The host notes that Younis's first tweet drew roughly 2 million views, with replies from Elon Musk and a quote tweet from Marc Andreessen calling him the best AI CEO nobody knows. Younis says staying under the radar was intentional, inspired more by Berkshire Hathaway than by Silicon Valley darlings. His motto is radical pragmatism, so before copying him founders should check that the advice fits their situation.

He could stay quiet because he was already known in the ecosystem, he says; someone without a network can use a following to recruit people, investors and customers, as investor Naval has said of fame. For Younis and his co-founder Peter, there was also little emotional satisfaction in public activity. Every minute spent on a podcast or writing for the public, he argues, is a minute not spent on customers and product.

He changed course because even a company like his wants to get its message out. Growing up as an immigrant from Pakistan in Warren, Michigan, he felt like an outsider and skeptical of the mainstream, a psychology he traces in many founders. Marc Andreessen pushed him to leave that baggage behind and be pragmatic, and he now wants to share ideas about physical AI and about what these technical changes mean for society.

Drawing on his time as COO of Y Combinator, Younis says good companies tend to show traction fairly early and sustain it for a decade or more. For founders two years in who are struggling, his heuristic is whether market feedback is pointing toward an increasingly specific path. If not, he would consider a reset, because often the foundation of the house is wrong: the co-founders, the market, or how much effort the founder is willing to give.

He tells Applied Intuition alumni who start their own companies to treat the first startup's first three years as a zero. He compares it to woodworking, where a wobbly first table is just the first table, and says being a founder is a muscle. Many founders put so much pressure on themselves to succeed first time that they miss the learning, and he notes that his third company is his most successful.

He and the host agree that great ideas often come from low-expectation tinkering, and that founders must hold two conflicting views: the work matters enormously, and it also does not matter that much. Younis says pivoting is easy before raising money or hiring employees, but afterward a founder who changes the mission loses credibility. He told Sam Altman, then YC's president, not to announce his COO appointment for a year, in case the partners did not want him.

Younis shares Applied Intuition's meta values: radical pragmatism and "our best work is done alone and quietly". He gives tactical advice for choosing values when a company gets a little traction: do not philosophize, but write down the five to ten reasons you are succeeding. Theirs became speed above everything, never disappoint the customer, technical mastery, high output matters, and laugh a lot.

Laughter, he says, keeps people grounded and gives a gentler way to say something is not the best. Another value is that half the work is follow-up: taking notes and following up is the business. The speed value is worded "move fast, move safe", and managers are assessed, compensated and promoted against these values.

Coming from automotive engineering, he sees quality and maintenance as the heart of operations: a car is built every 30 seconds, cheaply and globally competitively, and a factory is about safety and maintenance. He urges founders to see their company as a machine to be kept well tuned, noting a man who cannot command himself is unfit to command others. The company's employees clean the office themselves each week in a Japanese-inspired "cleaning zen" and take off their shoes.

Younis says the company's claim to fame is that in almost ten years, with over a thousand engineers, it has never spent the money it raised. He half-jokes that it is connected to cleaning the office, and sees a direct line from being quiet, keeping your desk clean and writing good software. Joe Montana, one of the investors, inspired the line in their Series D post that the valuation takes care of itself.

Younis urges founders to take the best of Japan, Germany, China, Detroit and Silicon Valley, and to be humble and learn from everything around them. He quotes Charlie Munger saying he has never met a very successful person who does not read all the time. His own ethos is to read old books, since time has filtered out the noise, and he notes that most people will read perhaps 50 to 100 books in a lifetime.

He says reading something like Malcolm X's autobiography will make you a better founder, even though the link is not a direct one-to-one if-then. Being well rounded, understanding society and history, somehow leads to better products. The host compares this to LLMs, where feeding in diverse data somehow produces something almost conscious.

He is currently reading *The Emperor of All Maladies*, which he says changes how he thinks, and the vibe coding book his company is reading, which goes against his own heuristic. Among classics he praises Sam Walton's *Made in America*, Colin Powell's memoir, *Guns, Germs, and Steel* and *Collapse*. His method is to pick an area he knows little about, such as Roman history, and find the best book in that space.

The host contrasts this with Marc Andreessen's barbell strategy of only today's news on X and books from ten years ago. Younis says Andreessen influenced him to join X, and that as the company grows more influential it is his responsibility as CEO to spread his ideas to the AI founder community, then technology leaders and the world. He notes that venture capitalists were once anonymous, until a few like a16z and John Doerr became individual voices that create gravity in Silicon Valley.

Asked about listening to naysayers, Younis describes the ideal company as one that can absorb many competing ideas, where the best one wins without emotion. Founders often resist this because popular media suggests a strong leader holds a hard view. He says they take inputs from customers, employees, competitors, investors and society, and he tells everyone to speak up, even the most junior person, because a single person's experience at Tesla, Waymo or a Chinese company may be the key idea.

He illustrates with Google, where he worked in the era Facebook was emerging. Google was then the apex predator of Silicon Valley, yet it could not beat a company of perhaps a thousand people at social media, because, as he puts it, a gorilla cannot learn to fly by not being a gorilla. Momentum creates a wall of sound that drowns out the signal that the market is changing, and small companies can be led astray by a founder's view that is just ten degrees off.

Being open has to be balanced with decisiveness: under the speed value, the behavior assessed is decisiveness, and a founder must know when to hold and when to fold. There comes a point, sometimes sooner than expected, when no more information will arrive and a decision must be made. The host adds a technique of imagining nobody's feelings would be hurt to find the right decision, then working out how to help people through it.

Younis says emotions are generally not helpful for rational decisions, since they are reactions shaped by life experience rather than optimized for a product review. A good heuristic is that several people in the company reach the same decision, so emotion acts as a filter you remove. He thinks it underestimates human thought to say AI is just associations of words, but stresses that a leader must ultimately be right and that the company must be a sustainable business.

Asked about his view that many Silicon Valley CEOs lack taste, Younis is careful not to claim that he has it himself. He says many are simply not exposed to many interesting, good things, a theme of the whole conversation. Someone who grows up in Cupertino, goes to Berkeley, starts a company straight out of school, and has never been an employee misses an important experience.

He spent over a decade at huge organizations such as General Motors and Bosch, where he learned in the bowels of the bureaucracy how bad it is to be an employee. That memory shapes how he makes policy and builds culture as a leader. He thinks taste means understanding humans and life and being able to discern what is good, and he suspects that someone who has backpacked around the world for a few years is likely to be a better founder, though he has no peer-reviewed research to cite.