Austin Vernon - Energy Superabundance, Starship Missiles, & Finding Alpha

Dwarkesh Patel introduces Austin Vernon as someone who writes online about engineering, software, economics and investing, but about whom little is known; the only identifying trace on the internet is a single JPEG uploaded with a recent paper. Vernon says his degree is in chemical engineering and that he has a lifelong love of engineering, including things like the Toyota Production System. He has worked as a chemical engineer in a large processing facility, where he did a lot of petroleum engineering, taught himself to write software, and now works on research and the early commercialization of CO2 electrolysis.

The first listener question comes from Alex Berger, co-CEO of Open Philanthropy, who suggested asking “Why so shady?” about Vernon’s pseudonymous online presence. Vernon recalls that Berger tweeted that he did not know who this person was or whether he was credible, but that his material was interesting, and he found this hilarious. His own position is that fame does not seem necessary: he is content for his ideas to be well known and to communicate them, but has little desire to be personally famous.

“I don’t know who this guy is or if he’s credible at all, but his stuff sure is interesting”

Vernon’s starting point is cost. Starship, he says, takes a technology and makes it about 100 times cheaper for cargo and 1000 times cheaper for people, and a change that drastic is hard to anticipate. Moon and Mars projects do not catch the general public’s eye, and people struggle to imagine point-to-point travel. Framing the technology as a weapon, he argues, signals that people should pay attention, and the United States should not end up second or third to get it.

Patel adds the cost trajectory: as recently as the 1990s, sending a kilogram to space cost around $20,000, SpaceX has brought that to $2,000, and the open question is what becomes possible at $200 per kilogram. Vernon describes the result as a bomb truck. The B-52 is usually thought of that way, but Starship could be cheaper still at delivering mass on target. Flying a B-52 from Barksdale in Louisiana across the world is expensive, whereas launching from south Texas or Florida would allow more missions per day (the transcript reads “emissions”, as transcribed). Reaching orbit takes a lot of fuel, but once in orbit consumption is low, so the advantage grows with distance, which is why point-to-point works best over long ranges.

Vernon argues there is a sweet spot where weapons are accurate but also cheap. He points to Russia, which has fancy, expensive weapons such as multi-billion-dollar cruise missiles but is short of $5,000 guided artillery shells, $20,000 guided bombs and guided rockets for multiple launch rocket systems. His explanation is that Russia had only a limited supply of chips obtainable from the US. A Starship could serve as a delivery platform for guided munitions in a shroud or for unguided iron kinetic projectiles, and if the cost per round is low enough, a ship cannot launch enough interceptor missiles to stop them; it is simply overwhelmed.

Asked to define terms, Vernon explains that “JDM” (as transcribed; he is describing the Joint Direct Attack Munition) is a fin kit added to old surplus Vietnam-era bombs. The kit costs about $20,000, which he calls cheap for a weapon because the bomb itself costs perhaps $3,000. It turns an unguided bomb that might land within 500 meters of a target into one that lands within two meters, cutting the number of missions needed by orders of magnitude. He presents this as a huge logistical advantage, achieved without building new bombs.

On chips, Vernon cites RUSI, which he believes is in the UK, as having torn down Russian weapons recovered in Ukraine and found American chips in all of them, even though Russia is not supposed to be able to obtain them. Russia cannot make many of its own chips, especially specialized ones, so it has somehow smuggled them in. Patel asks why Russia cannot source chips the way the US does, from Taiwan or China. Vernon says the issue is integration, not a single chip: the board and the wired-in programming, more like PLCs, come with guidance capability and must all work together. He concedes he lacks a really good answer, but says the key point is that these systems are hard to replicate.

Patel draws a downstream implication: India buys many weapons from Russia, so if Russia lacks access to these components, other buyers of Russian weapons lose access as well.

Patel raises Vernon’s speculation about keeping kinetic weapons in orbit, almost literally a sword of Damocles, and asks whether orbital decay and weapons falling on cities would make this too risky. Vernon answers that the benefit of weapons on orbit is speed. A rocket launched from Florida might need 30 minutes to arrive, and the target can move in that time, whereas weapons already in orbit can be spaced out to hit within a few minutes.

Vernon thinks most designs need two stages. A very aerodynamic rod performs well in the lower atmosphere but would go too fast and burn up before arriving. Tungsten may be the only material that could go all the way through, which is why he likes the original idea of telephone-pole-sized tungsten rods, but tungsten is expensive and the rod concept limits what can be done. Other designs use a shroud that slows the weapon in the upper atmosphere and is opened once the velocity is low enough not to melt it. If a weapon simply fell from the sky, some might reach the ground but much would burn up, and what survives is mostly light material with a large surface area. He notes that Starlink satellites are designed to fall apart completely on de-orbit.

Patel notes that kinetic energy is mass times velocity squared and that velocity has an upper bound, so the destructive power of these weapons can be bounded. Vernon agrees, saying people who do not do the math imagine a nuclear weapon. By his estimate even a tungsten rod is on the order of 10 tons of TNT (written “T&T” in the transcript).

It’s a big bomb, but it’s not a super weapon.

Vernon’s conclusion is that advanced missiles of this kind are almost defensive weapons, able to keep a ship from operating in an area. They could bombard cities, but they cannot take ground or really police sea lanes, and regular ships would still be needed to enforce the rules of the sea under this air cover.

Patel asks whether a weapon loaded with shrapnel could explode next to an incoming missile or aircraft, which he found surprising. For ships, Vernon says, an initial strike to kill radars should come above the ceiling of the ship’s missiles. The question is how far the ship moves between release, when steering stops, and impact, which he puts at maybe 1000 feet, so the weapon can simply spread shrapnel over the area.

Aircraft are harder because the weapon would be steered all the way in, perhaps even more slowly than in a ship attack, using a specialized package. He calls it doable but pretty hard: satellites with synthetic aperture radar could see the aircraft and guide the bomb the whole way. A package could unfurl next to the target and launch heat-seeking or conventional missiles. He frames these as things that might be possible with real effort, not easy ones.

Patel says modern weaponry looks directly descended from World War II, unlike the change between 1900 and 1940, when weapons became a different class. Vernon replies that physical technology has not changed much, so it has made more sense to put better electronics in the same tanks. They look the same, but with more powerful, accurate systems, longer-range weapons and better sensors, they may have several times more killing power. He says the war in Ukraine, where many 40- to 50-year-old weapons are used, shows this especially. Patel responds that this is worrying given US stockpiles, which he suspects are maintained versions of decades-old technology, though he says he is not well educated on the topic. Vernon’s reply falls in the next segment.

Continuing from Patel’s worry about stockpiles, Vernon says the debate over military readiness depends on the situation: the US is more ready for some scenarios than others. In general, he argues, most people discussing the topic have an incentive to downplay American capabilities because they want more defense spending, so he suspects the US is more capable than an editorial in The Hill might suggest. He sees a hint of this in how successfully Ukraine has used the few weapons the US sent.

He adds a caution about uncertainty. In fighting, especially naval engagements where the US does not have many ships, bad luck can decide outcomes, so one should stay wary and not become overconfident.

Patel suggests that if the offensive technology sent to Ukraine is better than defensive technology, even a single ballistic missile from China or Russia might sink a battleship and kill the 1,000 or 2,000 people aboard, and he asks whether the same advances open avenues for defense. Vernon says the general consensus is that defensive technology has improved much more recently than offensive technology. He ties this to China’s anti-access/area denial (A2/AD) strategy, which rests on missiles having better sensors. That lets China keep American ships from getting close, but not challenge the US at a place like Hawaii. He stresses that this works both ways: Taiwan also has many missiles with the new sensors, so an operation near Taiwan is probably tougher for China than most people say.

When Patel objects that most commentary says China could quickly knock out Taiwan’s defenses, Vernon replies that it might, but that this is a guess surrounded by uncertainty. He gives several reasons for doubt. Taiwan has one of the world’s largest defense budgets, which he puts at about $25 billion a year plus an extra $5 billion recently, and it has bought many anti-ship and air-defense missiles, which he says Ukraine could only dream of, since Ukraine’s budget was about $2 billion (his figure, given with some hesitation). Taiwan is also an island, whereas Russia could roll across a land border.

Vernon adds that there have been very few successful amphibious landings in history, the most recent being American ones in World War II and Korea. The burden is on China to execute perfectly. If its ships’ air defenses are not quite as good as assumed, it could lose half its fleet within 10 hours.