Hello, this is Baz to the Future and I am Baz. Long-time listeners won't know anything has changed, but people who catch these clips on the internet will notice that I am uncharacteristically in a suit, tie, and a mustache. My guest, Shyam Sankar, has a lot to do with both of those factors. This podcast goes deeper into important topics, often technological, sometimes societal. I'm in DC taking meetings, which is why I'm in a suit. A year and a day ago, Shyam and I accepted commissions into the US Army Reserves together into Detachment 201. Last night was the commissioning of the second wave of lieutenant colonels. We were in uniform, and the mustache is the only approved facial hair in the Army. I could shave it, but it would be a disrespect to our NCOIC who fought for this regulation mustache.
I agree. You got to keep it.
No, it's like the squad has become the mascot.
Exactly. I was just going to say that.
Yeah, it's for the squad. In addition to being a fellow lieutenant colonel, Shyam is the CTO of Palantir. Rather than reading your Wikipedia page, what do you think is the most important thing for people to know about you?
Well, the through line to Palantir, the army, and other things I care about is my family. We fled violence in Nigeria to settle in the US. I grew up in the shadow of the space coast with a sense of the possibilities of science and technology. I'm profoundly grateful to be in America and interested in dedicating my life to American greatness. Two, I love puns.
Listen, that's the important things we need to know about you: a profound life story and that you like puns, and also bad taste in wordplay. Let's start with Detachment 201. Almost two years ago, you called me and asked if I wanted to join the army. I said probably. We're a year and a day in. What have we learned?
We've learned a lot. First, interacting with uniform service members is compelling. The army is drowning in talent. It's a question of how we mobilize them and get bureaucracy out of their way. On the other hand, American industry has a lot to contribute to the military. Your experiences building Meta over 20 years can benefit the army, even if it's just learning from mistakes. You've delivered things at enormous internet scale that the army should benefit from.
It's interesting to think about the internet as one of the few things of a scale bigger than the army. Private sector companies have solved things at huge scale applicable to the army. I couldn't agree more. It's incredible to spend time with a formation doing what they do best. They're the best in the world and care immensely. Every soldier is super capable. It makes me feel more acutely how we've let them down in public-private partnership. They're not equipped with even what a private citizen would have.
Yeah, that's right. I've told people the culture in the army is a get-it-done culture, which is admirable. They get it done with what they have. It's amazing they accomplish so much, but it's made harder than it needs to be. I've been delighted that we've been invited to real conversations and can contribute. We're doing this entirely separately from our professional roles; it's a church-and-state thing. For you, given Palantir, that's limiting, but you're still finding interesting areas to contribute.
Principally, I've been helping on software talent strategy. We have uniform service members who are enormously capable developers, from embedded systems to enterprise applications. The industrial era model for talent has an impedance mismatch with the information age. How do we help the army rethink career trajectories and productivity expectations? The army sometimes thinks great people leave because they can make more in the private sector, but they knew what they signed up for. They're leaving because they can't be productive. We can solve these problems. It's been rewarding to see how much the army has adopted these recommendations over the last 6 months.
Very little to do with what Meta has to do with, so it's a broad range of skill sets we're trying to bring to bear. I'm curious: you were the person who ushered not just me but Kevin Weel and Bob McGrew into the detachment. You've been such a champion. I'm not usually hyping books, but I believe in this. You called me and sent the link to your 18 thesis. Now you've followed it up with this book. You have a broader construct of how to take the industrial mindset that created the army and move it forward. What's the major thesis?
It starts with the idea that the military as we think of it today is a consequence of winning the Cold War. The military that won World War II and the early Cold War looked radically different. We had founders then. In World War II, we direct commissioned 100,000 citizens to be officers. The entire media communications department was run by Disney employees. We brought the best of industry into the military. The military is a monopsony – there's only one buyer for tanks and aircraft carriers. It's hard to be a monopsonist. Everyone, including the Chinese and Russians, has given up on communism except for Cuba and the DoD. We have a five-year central planning process. We don't have robust competition of ideas. In World War II, we built 154 different types of aircraft. 10 really mattered, but that's the point of the free market – you couldn't know which 10.
That's right. They got quite a few of those wrong. Some that weren't thought important ended up critical. We went from building thousands a year to hundreds of thousands in the final years, an incredible turnaround.
Could we do that again today? We've taken the wrong lessons from history. It wasn't just flipping a switch; it was led by people like Bill Knudsen, who was commissioned a three-star general to run manufacturing. He previously ran GM. We had founders inside government, like Hyman Rickover who built the nuclear navy. Oppenheimer told him it wouldn't work, but he did it anyway.
Yeah, that's right. There's none of that almost because it's become so institutional and because we've been at peace for a long time. We've been a hegemon for so long that it feels fait accompli. Now we have challenging geopolitics. The South China Sea would be a tough place to engage in conflict without continued investment.
If you think about the purpose of the Department of Defense, it's to deter war. Why do people start wars? Usually because they think they'll win. How do you convince the adversary they're not going to win? You buy time.
I didn't appreciate when I joined the army the degree to which force projection is a path to global stability. It benefits everybody in the absence of conflict and in stable relationships and commerce. To do it, you have to be so obviously the better military that it's the long-term optimal strategy for others to adapt around the status quo. It's not about American hegemony; it's about global stability.
America is unique. After winning World War II, we invested our own capital to rebuild Germany and Japan. We did that to create prosperity and conditions for global stability. That was a lesson from after World War I.
You just talked about Rickover, and you have this great framework of Heretics and Heroes. It's not just in the military context but in all creative contexts. Talk more about that.
In the valley, we call them founders. In the history of defense innovation, everything that worked was a heretical idea. Winston Churchill built the tank. Andrew Higgins built shallow draft boats. The Navy tried to steal his IP but failed. Higgins boats landed in Normandy. The Sidewinder missile was a rogue project. These are beautiful stories. I wanted to elevate the discourse for those heretics lying in wait: remember the great Americans who came before you faced this. Stick your head up, the bureaucracy will attack you, but retrospectively people will realize how heroic your effort was.
It's funny how many things like the A-10 Warthog came about through malcontents who saw opportunity and pushed for it. One of our jobs is finding that talent and making it safe for them to take risks. The AI space is changing rapidly. Some people have become candescent in their capabilities because of AI tools. There's a set of other people for whom this is scary. The willingness to embrace change is one sphere; to accept change is another. How do you prevent the sclerosis that infects every large, long-running thing? You hit moments of stability. In industry, we had a long period of stability from 2012 to 2020. When you have a quiet period, you consolidate gains and become efficient, but then you're out of position for the next shift. Can you have something at scale that's also dynamic, or do you need a war equivalent?
The entropy of the universe is against you. You can either have micro revolutions or encourage heresy. That's your constant pushback. Occasionally, macro conditions make rebellion unnecessary. Then you need something more structural. One advantage the military has is the primacy of winning. There's one winner and one loser. In business, consequences can seem less severe, so the impetus for change requires more to muster.
That's exactly right. Sheryl Sandberg used to say companies don't fail by missing ambitious goals; they fail by hitting their goals all the way down. They last 10 or 15 years after you'd say they're dead, but people still go to work. It's not the same caliber of consequence as conflict.
Part of what led me to write the 18 thesis and the book is that we were managing our way down. The goal of the department is deterrence. We had the annexation of Crimea, militarization of the Spratly Islands, Iran's breakout capability, a pogrom in Israel, the Houthis holding the Red Sea hostage, and the invasion of Ukraine. It's hard to say our adversaries feel global stability is in their best interest.
Another thing you talk about is artists calling out of factory. What does it mean? I want to embrace the creativity of people. We have so many founder myths, but I believe the collective creativity is what we experience in products.
Completely agree. The premise of the artist colony is to maximize the individual. You don't yell at Dali to paint more like Monet. You create psychological safety and space because art doesn't monotonically improve. Being marked to market is anti-creative. You create an environment that maximizes human potential over a long arc. Focus on strengths. The error between you and your role is experienced by your weaknesses. Build a role around you. Working on weaknesses is largely a waste of time. Your impact comes from your superpowers. It's not just one founder; it's a team. Like the X-Men, uniquely flawed but powerful together.
For you, psychological safety is crucial. Otherwise, you'll be inclined to put people in a factory role because it gives you certainty, and they'll gravitate into it for the same reason. How do you create that safety when there's so much change and uncertainty? Even the tools are changing. Some people are enthusiastic, some aren't. How do you foster it over the long term?
Get them excited about doing great things rather than getting tunnel vision. You need a plan, maybe something resembling an OKR, some structure. But the structure must have an external greatness towards it, an ambitious thing to create. Then create a space that serves that. The military would call it task organizing. The more often you do that, the more you fight the entropy of the factory.
It's perverse when people are excited about a vision and run into bureaucratic barriers. The first-order effect is they can't do what they want. The second-order effect is learned helplessness. They stop thinking about how to do better.
That's exactly it. The first-order effect is the inability to do what you want. The second-order effect is learned helplessness.
This happens a lot in risk analyses. It becomes someone else's job, and you accept the reporting. Functions become disconnected. The person assessing risk has no vision for what the product could accomplish. The person who knows the potential has no ownership over risks. You end up with very conservative corporate constructs. It's heartbreaking to anyone who cares about creating value.
These are not laws of physics; they're the organization of man. In Middle-earth, Palantir is a word from Lord of the Rings. The orcs spawned at the seams. All problems institutions have are the seams between teams. Inside a team, it's probably functional. The seams between teams are where things get pathological.
Scale is hard. Great work requiring many people creates and necessitates seams. One of my great enthusiasms for AI is that when software gets cheap, process gets expensive. Today, process doesn't feel expensive because coding takes longer. But when coding time goes down by 10x, process becomes the long pole. AI exposes those things. Now you can attack them or obviate the need for them. Things we thought important, like not building infrastructure twice, now I say build it twice. We'll reconcile later. One of my hopes for AI is to transition how we work as an organization. I'm thinking about process, psychological safety, empowerment, great vision, and tools. Give people space to take risks. AI will be a huge game-changer.
I agree. In many ways, AI is the antidote to the 20th century's managerial revolution. There was no way to scale except through process and management. Now the most valuable people are those with specific knowledge.
Yeah. And getting them levered up to do these things.
Yeah, that's a great way to put it. We've never solved the quadratic overhead of an organization chart.
It's a very uneven revolution, the AI revolution. It's very jagged. Where are you finding tremendous success with it and where do you think it's surprisingly behind?
Well, there are two observations. First, you're surprised by how much specific, vocational experts at the front line are able to accomplish with this. But not every one of those people is that human, so empowering them to build things their colleagues can use. The second dimension is that this is like a vector with magnitude and direction. One challenge with AI is it makes you so productive, you have to make sure the direction is correct. There's a hedonic loop where you feel productive and magical, but the onus is on you to think about whether you're doing something valuable.
My experience with coding sessions is that I used to be an expert and it felt meditative. Now AI collapses hours of work, but I only spend time on the hard parts. I'm making the hardest decisions about what the product needs to be. Previously I figured it out as I coded, but now I get a sharp picture of a terrible thing I don't want to ship. I need more specification up front and more testing. I'm intellectually cooked after a vibe coding session because I'm doing only the work I uniquely can do.
