
How AI Is Changing Warfare | Palantir CTO
What this covers
Patrick O'Shaughnessy interviews Shyam Sankar, Palantir's CTO, over about ninety minutes that move from personal management philosophy to national strategy. The first stretch is about people: the Gamma Ray Talent Model, in which leaders throw high-potential people into problems they are not yet qualified for and accept that some will fail. Alongside it sit Superpower and Kryptonite, the idea that a hopeless weakness should be accepted rather than fixed, and a doubtful view of structured career ladders. He then describes forward deployed engineering and the origin of Palantir's Warp Speed program, which began when engineers noticed that final assembly lines ran on Excel.
The middle of the conversation is about the military and its suppliers. Sankar tells the story of Hyman Rickover and the nuclear submarine, then the 1993 "Last Supper" that shrank the primes to five. He contrasts the WWII industrial base, where 94 percent of major weapon spending went to dual-purpose firms, with today's 86 percent from defense specialists, and he criticizes cost-plus contracting. He also argues that US deterrence eroded over the last decade and has only begun to recover. These claims are often contested.
The last major thread is China and re-industrialization. Sankar describes Chinese long-range planning and a conception of war built on deception. He points to the US reliance on China for generic drugs, WuXi's rise, Apple's spending there, and the roughly $2 billion rare earths fix, and he calls for a domestic effort on the scale of a Marshall Plan. The talk closes briefly on AI, institutional legitimacy and fatherhood.
Sankar argues that America's military and industrial decline comes from suppressing 'heretics', the founder-builders who defy process and bureaucracy, so restoring deterrence and re-industrializing requires rebuilding a culture that rewards such people rather than adding more process.
- Breakthrough military and industrial innovations (nuclear submarines, the Higgins boat, the Sidewinder, the Skunk Works U-2) came from heretics who defied the system, while process-driven channels delivered little.
- Post-Cold War consolidation and financialization of the defense industrial base removed the risk-taking founder culture and replaced it with cost-plus contracting and conformity.
- Offshoring of production has eroded innovation itself, because innovation depends on productivity and manufacturing learning, so re-industrialization is a precondition for regaining technological leadership.
This asset isn't compiled yet
You're seeing its claims, ranked. Compile it to build the argument threads, weight them, and check each claim against your library — the full view.
The military is unlike commercial companies because it separates supply from demand: the services man, train, and equip forces without responsibility for demand, while combatant commands face real-world events, so the buyer who wants a capability often cannot buy it, adding a layer of non-market complexity.
“The military is the only institution in the world that kind of profoundly divides up supply and demand.”
The 1993 Last Supper at the Pentagon, where the Secretary of Defense told about 15 of 51 major defense primes they were not going to survive and could consolidate or exit, triggered a merger wave that reduced the primes to five; the deeper harm was not lost competition but the financialization and conformity that drove out the crazy engineers.
“This is the moment of profound financialization and conformity in the industrial base. You lost the crazy people.”
The US offshored production, which undermined innovation itself because innovation is a consequence of productivity: Google's 2017 'Attention is All You Need' transformer paper grew out of a seemingly incremental 3 percent Google Translate improvement, and without manufacturing stimulus the innovation cannot be captured.
“innovation is itself a consequence of productivity... if you don't have the input stimulus on improving production, you're actually not going to be able to capture the innovation.”
Hyman Rickover, despite being a poor performer by conventional military measures, pushed through the first nuclear-powered submarine, built in about seven years from start to finish, against Oppenheimer's skepticism and Navy hostility, and his heretical persistence produced a durable asymmetric advantage that still exists today.
“he built the first nuclear submarine in 7 years, start to finish. The Navy hated him so much that his first office for this project was literally the women's restroom.”
Rickover's engineering culture for naval reactors, shaped by an 'I'm designing this so my son could be in it' standard, produced zero deaths from nuclear incidents in the US submarine force and a safety standard roughly a hundred times stricter than the believed minimum, whereas the Soviet submariners spent long periods recovering from radiation exposure.
“We've had no deaths due to nuclear incidents in our submarine force and he designed it to a standard... it's a hundred times safer than what we believe the minimum standard to be.”
In WWII, only 6 percent of spending on major weapon systems went to defense specialists; 94 percent went to dual-purpose companies like Chrysler (missiles and minivans), whose R&D on production, materials, and machine tooling subsidized national security, whereas today 86 percent of major weapon systems come from defense specialists only.
“Only 6% of spending on major weapon systems went to defense specialists... One, that 6% number today is 86%.”
China is a formidable adversary because its conception of war is deception and systems warfare, aiming to win without firing a shot, while the American conception is kinetic and heroic; much of the conflict happens below the threshold of traditional war.
“Their conception of war is deception... You want to win without firing a shot.”
The US should incentivize heretics by recognizing them publicly and by making defense primes more valuable, because the defense industry's 9 percent operating margins and under two times revenue valuations deter talent, and the worst incentive is cost-plus contracting, which removes risk and rewards conventional work.
“the worst incentive is cost-plus contracting. Your upside is capitated, you're not supposed to take any risk”
Apple has spent the inflation-adjusted equivalent of two and a half Marshall Plans over five years building talent and manufacturing capacity in China, so the US should attempt the equivalent of at least one Marshall Plan domestically to rebuild machine-tool and manufacturing talent.
“Apple has spent the equivalent on an inflation-adjusted basis in the last 5 years of two and a half Marshall Plans building talent and capacity in China. How about we try to spend one Marshall Plan here?”
About 80 percent of US generic drugs come from China, either as active pharmaceutical ingredients or finished products, creating a national security vulnerability in which a child's ear infection could become life-threatening during great-power competition.
“80% of our generic drugs come from China in one way or another, whether it's the APIs or the finished product themselves.”
Deterrence against a peer adversary requires the ability to generate a strike every single day until the enemy submits, not a single dramatic operation; a single strike like Midnight Hammer is massively deterring but is only one event.
“the deterrence capability against a peer adversary is can you generate that strike every single day until the enemy submits to your will.”
The US lost deterrence because it prayed at the altar of process, trusting requirements processes and bureaucracy to yield good results, and the only thing that process produces is decline.
“The reason we're screwed right now is we we prayed at the altar of process... And the only thing that leads to is decline.”
US naivety, the belief that prosperity would make China liberalize, was a mistake: after admission to the WTO China cheated, and its pre-Deng Xiaoping rhetoric showed it called the US the big enemy; the US assumed economic development would produce positive-sum alignment, which did not happen.
“part of it is our own naivety, our own belief that capitalism and human flourishing is a natural consequence of prosperity.”
The rare earths problem was largely solved over about 15 years for roughly 2 billion dollars, a rounding error relative to US spending, and a 10 percent marginal improvement in domestic production capacity would dramatically reduce an adversary's leverage.
“we more or less have solved it for like 2 billion bucks. It's roughly a rounding error... making a marginal improvement of even 10% dramatically reduces the adversary's leverage.”
The US military today is the best in the world and its service members are better than deserved, but it encumbers them with process and deprives them of agency; the structural problem is that the military has failed at deterrence over the last 10 years, with Crimea's annexation in 2014, Chinese militarization of the Spratly Islands, the invasion of Ukraine, and gray-zone operations in the South China Sea as examples.
“If you look back over the last 10 years, we've not been able to deter a lot of conflict.”
Europe has created zero companies worth more than 100 billion euros from scratch in the last 50 years, while the US created all its trillion-dollar companies from scratch in that period, showing that cultures which protect human flourishing and allow founders to succeed despite resistance produce outsized innovation.
“Europe has created zero zero companies worth more than a hundred billion euro from scratch in the last 50 years.”
China's WuXi moved from low-cost pipetting for pharma contract research to producing about 50 percent of all clinical trial drugs, showing how seeding production can ultimately transfer innovation, and the US seeded innovation by holding the incorrect precept that it could keep innovating while offshoring production.
“now 50% of all clinical trials are drugs that are created in China.”
China's principal asymmetric advantage is long-term planning: after Gulf War I in 1991, China mapped the critical dependencies of US military technology and has been investing against them since then, broadcasting its intent throughout, while US strength lies in unpredictability that comes from its heretics.
“their principal asymmetric advantage is long-term planning... they have been doing that since '91.”
Bruce Banner's gamma-ray exposure is a model for developing talent: leaders take people with high potential and expose them to high-stakes problems they are not yet qualified for, accepting that some will not survive, while building enough information flow that the leader can intervene when someone goes off the rails.
“there's absolutely an element where we're taking people and just irradiating them.”
The US has lost deterrence over the last decade, with the 2014 annexation of Crimea, Chinese militarization of the Spratly Islands, the invasion of Ukraine, and gray zone operations in the South China Sea, but deterrence has begun to be restored in the last year through operations such as Midnight Hammer and the Maduro raid.
“If you look back over the last 10 years, we've not been able to deter a lot of conflict... arguably in the last year we've started to restore deterrence.”
Deterrence against a peer adversary requires the ability to generate a decisive strike every single day until the enemy submits, and the US has only started restoring deterrence in the last year, as shown by the Midnight Hammer operation and the Maduro raid, which also showed that Chinese and Russian weapons in Venezuela did not perform.
“can you generate that strike every single day until the enemy is submits to your will.”
Palantir's Warp Speed program came from a secret: before COVID, nearly everyone assumed manufacturing ERP systems worked, but forward deployed engineers saw that final assembly lines ran on Excel, which signals the software is so ill-fitting that users eject their data and hand-roll solutions.
“everyone's using Excel. So, if you're thing is like there's a contradiction here that no one is somehow forced to reconcile with”
Talented people are highly uneven, and high achievers often misidentify their superpower: they treat effortful activities as their strength and get dopamine hits from them, when their real superpower is effortless and therefore less rewarding to exercise.
“particularly for high high-achieving people when they're young they misattribute what their superpower is... Superpowers are effortless”
A person's kryptonite, a weakness that is far below average and cannot be improved, should be accepted rather than fixed, because accepting it makes the person more valuable and frees them to work on their superpower.
“The only strategy for Superman around kryptonite was to avoid it.”
Structured career ladders with defined progressions give employees a false sense of growth; the fastest learning occurs when people are thrown in over their heads, so the rate of learning is coincident with one's maximum tolerance for pain.
“the maximal rate of learning will be coincident with your maximum ability to tolerate pain.”
Success on hard problems involves a lot of pain, and the instinct to reduce the pain next time is the wrong instinct, because it diverts from the learning that difficult work produces.
“success looks like a lot of pain. And that that instinct to figure out how to reduce the pain, like how do I do this the next time with less pain is exactly the wrong instinct.”
When hiring unproven talent, the best test is whether the person can turn an inch into a mile; the worst bet is giving someone a mile and seeing them turn it into an inch, and early indicators are first-derivative progress, creating something from zero to one, and expressed agency.
“If I give this person an inch, can they turn it into a mile? And the worst sort of bet to make is I gave this person a mile and somehow they turned it into an inch.”
Palantir is 'wildly inductive': it does not show up at a customer with a preconceived ontology but instead maps the specific context to find what makes that customer different, aiming to make software that makes each customer more different rather than more similar, which Sankar calls alpha not beta.
“we want to make software that makes you more different, not more similar. We think about it as alpha not beta.”
The US now resembles Germany of WWII: it makes a small number of exquisite things while its adversary, China, is the world's best at mass production, which should cause discomfort because the Germans were better engineers yet lost the mass-production race.
“Now, if you come to the present moment, we sell like the Germans. We make a very small number of truly exquisite, eye-wateringly exquisite things, and our adversary is the world's best at mass production.”
Andrew Higgins built the Higgins boat on his own without a Navy specification, and the Navy initially rejected it, yet about 92 percent of boats in World War II were Higgins boats, showing that a heresy can win the war; companies like Anduril can replicate this by building products to their own specs and absorbing the risk.
“Andrew Higgins built that boat on his own. There was no spec... then 92% of all boats in World War II are the Higgins boat.”
Momentum, not persuasion, aligns a large organization during change: each change is a mini insurgency, so leaders should identify who already believes the change, add incremental weight behind them, and let the rest catch up.
“Every change is itself a mini insurgency inside the company... my experience has been momentum is the thing that aligns the whole company.”
In change management, a worst-case state is having no signal: the pain of world-class performance and the pain of failure look similar, so leaders must determine whether the pain is producing results, because managed decline feels like steady progress without feedback.
“if you're doing something world-class, it's really painful. Also, if you're totally failing, it's really painful... the worst world to kind of be in is where you have no signal.”
Institutional legitimacy is declining because basic services and products fail (such as doors falling off planes), and this breeds nihilism; leaders in government and industry steer a wheel disconnected from the machinery, creating a managerial middle that separates decision makers from the people doing the work, which Palantir's operating system is meant to fix.
“The core problem that we're set up to solve is the legitimacy of our institutions... it breeds nihilism.”
America's asymmetric advantage is its unpredictability, which comes from its heretics; its bureaucracy is predictable and steady managed decline, while the AI pivot showed the economy can change direction on a dime when the unexpected arrives.
“Our asymmetric advantage is that we're crazy... our bureaucracy is predictable, right? It's not the steady Eddie managed decline, it's it's the crazies.”
Rickover's engineering culture at Naval Reactors is still alive today because he was both the inventor of the technology and therefore qualified to regulate it, which made regulation credible.
“he was in this unique role of both being a human who would invent the technology and therefore was qualified to think about how to regulate it.”
Every contribution to the world should come from one's superpower, and everything else is basically a waste of time; this requires releasing ego and accepting one's kryptonite, which cannot be worked on and can only be avoided.
“all of your contributions to the world are going to come from your superpower. Like everything else is basically a waste of time.”
Forward deployed engineering was developed because conventional software companies use sales as the feedback loop (can we sell it?), while the true test is whether the software delivers the outcome in the field with the end user; so Palantir sends engineers who solve backward from the operator's problem and build software through a form of back propagation.
“This software valuable? Did it deliver the outcome? The only way you can assess that is actually in the field with the end user, on the factory floor, in the foxhole.”
Palantir's core thesis is that decisions, not data, are what make enterprise data valuable; its ontology is a model of both data and actions (kinetics) that makes the business programmable, creating an API-like layer that lets organizations compound improvements in decision-making on top of it.
“our core thesis is what makes this stuff really valuable is decisions, not data.”
Lockheed's Kelly Johnson, founder of the Skunk Works, built 41 airplanes in his lifetime, many still flying today, and designed the U-2 in 13 months, illustrating the artist-like engineering that process has suppressed; modern process-heavy development is worse because it made messy, frictionful work less messy and therefore less functional.
“Kelly Johnson was the founder of Skunk Works at Lockheed. In his lifetime, he built 41 airplanes... he built the U-2 in 13 months.”
Being given a high-stakes opportunity is what allows people to discover their own potential; Sankar's formative experience was leading a deployment at a co-op where success was uncertain until the last second, and he learned that success looks like a lot of pain.
“it was never obvious until basically the last second whether it was going to work or not... oh, success looks like a lot of pain. And that instinct to figure out how to reduce the pain... is exactly the wrong instinct.”
Culture is not self-sustaining: entropy works against it, so organizations and countries must invest in their culture every day, and the best cultures are cult-like, highly differentiated, and hard to join inorganically, requiring recruits who see the present job as a calling rather than as a stepping stone to the next one.
“the entropy of the universe is against you. Like, if you don't invest in your culture every single day, you will lose it.”
John Boyd's central advice was that people can either be somebody, playing the promotion and accolade game, or do something, which feels worse and goes unappreciated but yields intrinsic reward; Sankar and Palantir are trying to do something rather than be someone.
“you can either be somebody or you can do something. But you can't really have both.”
Palantir's organizational structure should be a 'quantum org structure' that crystallizes around the current problem and reforms as the problem changes; flat structures do not work, and big-company reorganizations every five years are a sign of ossification because entropy makes any fixed structure more wrong each day.
“Every day the entropy is against them and every day is slightly more wrong than it was before.”
Engineers fall into two types: those who know how to build the right thing (MacGyver-like problem-solvers who get a high from solving the problem) and those who know how to build it the right way (artists who value architectural elegance); forward deployed engineers skew toward the first, product engineers toward the second, and the tension between them is an unstable equilibrium that needs a dictator-like decision maker.
“there really are two types of engineers. There are people who know how to build the right thing, and there are people who know how to build it the right way.”
Customer needs follow a power law rather than a Gaussian distribution, so surveying customers and averaging their requests produces mediocre products; instead, a few customers 'living in the future' reveal valuable problems no one has solved, and solving them early can create a monopoly position about five years ahead of competitors.
“if it's a Gaussian, you should just like survey all your customers and average it out... some percentage of my customers are living in the future.”
The whole business is a series of decisions, and every decision should run an OODA loop, with the more important goal being a good first derivative: learning how to make a better decision tomorrow than today.
“I want the first derivative to be good. I want to learn tomorrow how to make a better decision than the one I had today.”
Palantir's early government deployments required disagreeableness: because operators who use the software are not the people who buy it, and IT program owners see the company as a structural threat, going along with customers would have produced software that did not work, so the team had to deliver what worked for operators even at the cost of angering nearly everyone, including suing the Army to compete.
“The problem is if I did what they want it wouldn't work and so that's where the disagreeableness comes in... I'm going to piss off literally everyone along the way.”
AI can be America's 'David's slingshot' for re-industrialization: by making the American worker 50 times more productive than a worker elsewhere, the US could change the efficient frontier of what can be made domestically, which justifies being 're-industrialization maximalist' rather than relying on friend-shoring.
“AI is David's slingshot here... If you can make the American worker 50 times more productive than a worker anywhere else, you're going to change the efficient frontier”
AI rewards the beginner's mind: Palantir's youngest hires were the first adopters, and overwhelming productivity from new employees pressured cynics into adopting AI, so leaders should empower insurgents rather than gatekeepers.
“one of the things I think is unique about AI is how much it plays to the beginner's mind. Obviously, the first adopters of AI for us were the youngest hires.”
Value in the AI stack will accrue at the chips layer and the ontology layer, because model providers are commoditized and must move up the stack, while AI-native applications are moving down the stack into infrastructure like Palantir's AIP, which Palantir believes it has a head start on due to 20 years of platform work.
“We think that the value is going to accrue at the chips layer and at the ontology layer.”
Alex Karp modeled Palantir as an artist colony, making it closer to a Hollywood talent agency than a typical software company, and he resists cargo cult processes that corrode value, a principle taken to the extreme producing something like the Soviet Union.
“from the very early days he modeled our entire company as an artist colony... more in common with like a Hollywood talent agency”
Heretics, meaning founder-type builders who obsess over delivering something that bureaucracies resist, are the people who actually drive change; the only things that ever helped the US win wars were things heretics did, and nothing that went through the bureaucratic machine delivered anything.
“the only that ever worked, the things that helped us win all the wars, were the things that the heretics actually did. Nothing that went through the machine delivered anything.”
Rickover's obsession was a founder's view of the world rather than a manager's view: a founder designs to a standard that expresses personal stakes, while a manager optimizes for compliance and process.
“It is a founder's view of the world.”
Forward deployed engineering is only viable for expensive products whose vendors can capture a share of the value they create, because the company invests heavily in customers upfront and its economics depend on downstream customer economics.
“you're not going to get paid a lot up front. You're actually going to invest a huge amount into the customer... Our economics will be downstream of our customers' economics.”
The person is the program: big outcomes such as the Apollo program or the Minuteman ICBM depended on specific people like Gene Kranz, Edward Hall and General Shriver, and getting the right people into roles is determinative of success, just as a company's future is determined by its founders rather than a recipe.
“The person is the program... it was actually Gene Kranz's program and the primacy of those people and getting the right people in those roles is determinative to the success.”
Military innovation has repeatedly come from rogue projects rather than the process: Churchill built the tank as a 'land ship' when the British Army could not see that horses would not win the next war, and the Sidewinder missile was a rogue project that humiliated the Air Force's missile in Vietnam. Process fixes lose to those who put everything on the line.
“every major military innovation including the Sidewinder missile which was like a rogue project that some guy kind of built with lighters”
America's greatness depends on a belief in exceptionalism and on plasticity of thought: no army that lost its morale has won a war, and cultures that cannot update their priors, as Europe still arguing over events 300 years ago, cannot make progress.
“No army that lost its morale ever won the war. And so believing in yourself, believing that greatness is possible, is a precondition”
Anduril shows the right model for hardware defense procurement: a company builds a product to its own specs and opinions, absorbs the risk, and lets the government buy it if it wants, which reverses the cost-plus dynamic.
“I'm building a product. These are my specs. These are my opinions. I'm absorbing the risk. You should buy it if you like it or not.”
Organizations change structure only after breaking, because even a correct structure at a point in time degrades every day as entropy works against it, and waiting for total breakage is a fundamental weakness.
“Every day the entropy is against them and every day is slightly more wrong than it was before.”
The US gained a key World War II production advantage from Lend-Lease: by producing goods for allies before Pearl Harbor, it built factories, retooled existing ones, and reached full-rate production by the time the US entered the war.
“We had one key advantage in World War in terms of timing, which is that we had Lend-Lease.”
Palantir reprioritizes its backlog dynamically against reality, tearing up the roadmap whenever new data arrives, which is painful and generates chaos but is the right thing to do, and the company's culture must enable this.
“my backlog is something I re-prioritize dynamically relative to reality... The second some new data comes in, I'm going to rip that up and re-prioritize it.”
When Palantir began working with Airbus during the A350 ramp in 2015, it spent time on the Toulouse final assembly line and separated non-conformities that are training issues from recurring design defects, since how quickly a company solves that distinction determines the shape and cost of the production ramp.
“one of the most important things to figure out as quickly as possible is what is an issue that is just a non-conformity... And what is a recurring issue that is actually a design defect”
The 18 Theses were Sankar's first major public statement, arguing that the US is in an emergency, has lost deterrence, and must act because its diagnosis is that it lacks the industrial base of WWII and must recognize the person as the program; the first derivative and even the second derivative of change in the Department of War are now positive, which he says is more change in the last year than in the prior 19 years.
“I've seen more change in the Department of War in the last year than the prior 19 years”
The US should reform pharmaceutical production by extending patent length for branded drug makers that manufacture their worldwide branded supply domestically, giving them enough extra exclusivity to recover the capex, since generics alone cannot support domestic production economically.
“I will extend your patent length such that you recover two extra capex investment to go do that.”
Admiral Zumwalt said 'The Navy has three enemies: the Soviet Union, the Air Force, and Hyman Rickover,' illustrating how difficult heretics are even when they are senior.
“The Navy has three enemies, the Soviet Union, the Air Force, and Hyman Rickover.”
At the start of World War II, Bill Knudsen, a Danish emigre who had been number two at GM and Ford, was given a three-star general's rank and authority over war mobilization and empowered to find other world-class engineers, showing that the US mobilized through people rather than process.
“we took the number two at GM, who was previously number two at Ford, Bill Knudsen, who was a Danish emigre”
Heretics only become heroes after their innovations prove out; Billy Mitchell was court-martialed and died penniless and depressed, and the US only created the Air Force based on his original heresy after his death.
“Billy Mitchell who invented the Air Force, you know, he was court-martialed and died penniless and depressed, and only after his death did we invent the Air Force”
The people who give others a chance, such as Kevin Hart taking a bet on a Stanford grad student, create the 'conspiracy' that enables others to take the deep-end leap, and people should assess whether an environment will bet on them or try to squash their eccentricities.
“is this an environment that I think is going to bet on me or is this an environment that wants me to be a veal cow?”
The American industrial base was founder-led: figures like Jack Northrop, Leroy Grumman, William Lear, Henry Ford and Henry Kaiser were heretics, and the talent was in industrial companies rather than institutions, so the defense system was betting on people, not just institutions.
“It was Jack Northrop. It was Leroy Grumman. It was William Lear. It was Henry Ford. It was Henry Kaiser. You had founder figures.”
The US is undermining its own strategy by reacting only after long periods of restraint: an admiral's line captures the pattern as the American OODA loop of observe, overreact, destroy, apologize, rather than Boyd's observe, orient, decide, act.
“it's the American OODA OODA Loop, observe, um overreact, destroy, apologize.”
Rickover has been used as a template for founder culture: the best engineering culture is one led by someone who both invents the technology and is qualified to regulate it, which contrasts with a manager's view of the world.
“it's not a manager's view of the world. It is a founder's view of the world.”
Palantir's Airbus engagement started on the Toulouse final assembly line by watching wrench turners work through Excel and SAP reports; the quality data asset then naturally extended to production planning and in-service uptime, showing that customers' problems exist within a connected decision chain.
“quality to production planning to then in service. Now you have the greatest data asset possible for this tail.”
The kindest thing anyone did for Sankar was when Ethel Danhoff, head of admissions at a Orlando private school, admitted him after the application window had closed and despite having no standardized data, because the family needed financial aid and she gave him an admission test anyway.
“Miss Ethel Danhoff was the head of admissions for this private school in Orlando... Give this kid the admission test. Just do it.”
Sankar's father, who fled violence in Nigeria in his early-to-mid 30s and whose business attempts went bankrupt, instilled gratitude and perspective, showing that a father's example of selfless grinding shapes a child's character more than explicit lessons do.
“a lot of fatherhood is being a man that inspires your family to be better.”
The Pentagon became the last bastion of central planning after other nations abandoned communism, which made defense a bad business while appearing to pose no threat, so defense contractors could behave like single-purpose prosperity companies; this is not a stable or healthy arrangement because civilization requires obligations across both freedom and prosperity.
“the Pentagon, the old Pentagon in particular, being the last bastion of communism except for Cuba.”
Palantir's cultural practices of celebrating heretics, holding twice-yearly weeks of revolt where people build whatever they want, and ending each new employee's first AMA by asking them to tell Sankar off to his face, are designed to institutionalize challenge while acknowledging that over-institutionalizing is a risk.
“at minimally two times a year we have these these weeks of revolt where people are allowed to go build whatever they want.”
Palantir is not really a defense company: about half its business is government and two-thirds of that is defense, which is closer to the historical pattern of dual-purpose industry, and its work is misunderstood as data collection or surveillance when its core mission is restoring the legitimacy and state capacity of institutions.
“half of our business is in government, 2/3 of that half is in defense, is actually more normal to historical kind of pattern”
Palantir does not fit into standard enterprise architecture boxes (as Gartner-style analysts categorize them) and instead solves the problems in the seams between them, so the heterodox approach is valuable precisely because it is heresy to the existing architecture, yet it works.
“We don't fit in any box. We fit in all the boxes... You're solving for all the problems in the seams.”
Palantir's own founders' AI-era bet: Palantir's ontology and platform, built since 2015 with Foundry leading to AIP, gives it a head start because the code is less valuable than the secrets behind why the code was built.
“there's the code and there's the secrets that actually is behind why we built the code.”
Organizations must create environments where people can make a serious mistake and be safe to disclose it; a culture that fires people for discovering their kryptonite destroys the learning that talent development requires.
“You don't want to create a culture which is like, 'You screwed this up. I got to fire you.'”
Sankar once made a consequential mistake at Palantir and sheepishly told Karp fully, and Karp valued the honesty even though he was in pain about the consequences; this experience taught Sankar that transparency is key to the environment.
“I sheepishly went and did Alex and was just completely honest... He was very appreciative.”
Hiring for culture should favor people who see their present job as a calling rather than as a stepping stone to the next job, because that mindset engenders a primacy of winning today.
“Some of them are just structurally mercenaries... you want to try to build your company with people who really think of the present job as the calling.”
Growing up in Orlando in the 1980s and 1990s, in the shadow of the Space Coast, instilled a structurally positive-sum view of technology that made building the frame for Sankar's life.
“growing up in Orlando in the '80s and '90s. It was this profoundly optimistic period... in the shadow of the Space Coast, you were kind of bombarded with this idea that technology was going to make the world better.”
Sankar's personal definition of greatness is an aspiration substantially bigger than oneself, focused on raising the aspirations of American workers and civilization, and he sees a nihilistic unstable equilibrium in civilization that greatness counteracts.
“An aspiration that's substantially bigger than oneself... There's this unstable equilibrium in civilization that pulls you towards nihilism. Uh, and greatness is the antidote to that.”
Forward deployed engineers are people obsessed with generating real-world impact who are willing to continuously solve backward, unlike sales engineers whose job is to close the sale.
“You're sending someone out there who is obsessed with whether it's going to generate an impact in the world or not”
Sankar's family fled violence in Nigeria, and his father nearly died in the process, which gave the family a deep gratitude for the US and a grounded understanding of the counterfactual, what makes the country special.
“we fled violence in Nigeria where my father we frankly almost died to kind of settle in the US”
Sankar's father's business failures and lack of a classic immigrant success story still produced graciousness and perspective, and Sankar argues that the immigrant success narrative is a 'facile' journey, meaning easy to romanticize.
“Dad started businesses that went bankrupt. It was very hard, but having that rooting experience that created a graciousness and perspective in him.”
Sankar's father, who gave everything to his family and could weather humiliation for his children, taught him subconsciously, and a major part of fatherhood is being a man who inspires one's family to be better.
“a lot of fatherhood is is is being a man that inspires your family to be better.”
Palantir's main critique is that its model requires a three-year crucible: retention is bimodal, and people who stay three years are very likely to stay a decade, so the business's future depends on helping the people who hit a roughly three-year wall, when they judge the chaos as a bug rather than a feature, to get over it.
“the probability that you'll stay for like a decade plus if you've stayed 3 years is incredibly high. But, you know, roughly around the 3-year mark, everyone faces this crucible”
Maximizing the potential of people requires a personality that is not threatened by others' successes and instead relishes them, which Sankar says is the parallel between Karp and his father.
“You actually relish in the other people's successes and that's the parallel between Alex and my father.”
Sankar's own superpower is the narrow intersection of forward deployed engineering and product, where he is effectively one on both sides of the interface and can act as a dictator to resolve it.
“the intersection of forward deployed engineering and product... It's like 1 in on both sides of that interface, like owning that almost as a dictator, that's my superpower.”
Palantir functions as a 'Founder's Factory' because serially putting people on problems they are not qualified for turns those with raw potential into people capable of far more than they expected.
“I think that's actually why Palantir is the Founder's Factory.”
Winning today matters most: a culture focused on winning the present job earns the right to win tomorrow, whereas focusing on how the world should be, an unmoored aesthetic, produces failure.
“If you win today, you have the right to win tomorrow.”
Sankar believes the US has made more change in the Department of War in the last year than in the prior 19 years, and that its first and second derivatives of progress are now positive, so the country is finally doing the right thing after exhausting alternatives.
“I've seen more change in the Department of War in the last year than the prior 19 years... the first derivative and even the second derivative is positive”
Palantir measures its progress by how fast it can deliver enterprise autonomy for customers; this interval fell from about 8 weeks two years ago to roughly one week now, and that rate is compounding dramatically.
“I'd say 2 years ago we could do that inside of 8 weeks. Now, it feels like a week”