Blimps, Bots, and the Future of Cargo

Blimps, Bots, and the Future of Cargo

One of the most important economic facts of the last fifty years is that the cost of transporting goods over the ocean has, relative to early history, dropped to roughly zero because of container ships. One paper in 2013 ($, Economist) found that bilateral free-trade agreements increased trade by 45% over a twenty-year period, while containerization increased it by 790% over the same period. This has had profound effects across the supply chain—it's enabled every step of the manufacturing process to take place in whichever country has a comparative advantage, for example—and one of its other effects is that the cost of later stages of transportation is a relatively bigger piece of the total.

Two parts of that last stage of transportation are benefiting from the same trends: battery prices dropped 97% from 1991 to 2018, while software for autonomous vehicles keeps improving. So any transportation solution that doesn't need a human in control but does use a battery gets cheaper by the year. Autonomous vehicles have all sorts of costs and limitations, but many of these arise because of their interactions with non-autonomous vehicles and with pedestrians. Go up ten feet, and you've eliminated a huge category of risks.

Drone delivery is one exciting part of this. For any product that has a high cost-to-weight ratio and benefits from timely delivery, drones are a win. This is happening now in a few places, most notably Australia, where the neighborhood of Logan has hit almost 4,500 deliveries a week. Annualized, that's 0.78 deliveries per person per year—not quite revolutionary yet, but promising.

Drone delivery's regulatory status is driven by two pretty obvious features: drones go up, and they can fall down. Going up means that they have a de facto right-of-way wherever they're going; unlike other delivery methods, they can pretty much travel in a straight line, so effectively utilization can be a lot higher. (Traveling in anything but a straight line imposes a cost, and one good illustration of this is how dedicated UPS is to avoiding left turns, which cause higher gas consumption and lead to more accidents.) But drone accidents do have consequences in a dense city, so the default approach is cautious.

One interesting question to speculate on: how much do drones substitute for cars, and how much do they increase total deliveries? Assuming there's some substitution effect, taking cars off the streets will reduce accidents. Right now the rough rate is one death for every hundred million vehicle-miles traveled in the US. But that implicitly assumes a linear relationship between miles traveled and risk. 53% of deaths are from single-vehicle crashes, and the risk of a multiple-vehicle collision is presumably lower for any given car if there are fewer cars on the road. It's complicated to model, but the upshot is that drone delivery's risks will look different on a net versus gross basis.

Drone delivery has some interesting potential to fill in gaps in urban transportation, both through its direct benefits and by reducing the ~$88bn cost of traffic congestion in a normal year. It won't displace that large a volume of delivery vehicles, because the goods in question are small, but it will reduce stops and reduce the frequency with which people drive to a store to make a single last-minute purchase.

Filling out another end of the transportation system is a rather different technology: airships. They're not that useful for dense areas, but very promising for spread-out ones; Buoyant is working on vehicles that can deliver a 650-pound payload in a 20 foot by 20 foot loading area (for comparison, a standard parking spot is about 10 feet by 19).

The general rule is that the smaller a payload is, the more cost is a function of distance traveled rather than total volume shipped; big ships can be almost an order of magnitude cheaper per container shipped than small ones. But that cost is really two costs: the operating cost of moving goods and the fixed cost of the infrastructure necessary to transport and receive them. Rail is 79% cheaper than trucks per ton-mile, but that's a marginal cost, not quite including the fixed cost of the tracks and roads themselves.1

For airships, the fixed cost is the vehicles themselves. Their most direct competitor right now is small planes and helicopters, which are fast but painfully expensive. In some circumstances, that's fine; for disaster relief, for example, a helicopter is still a very good deal. But for frequent small deliveries—restocking a Walmart on an island or in a remote part of the country, or making periodic bulk parcel deliveries—they can be highly competitive. There are still FAA regulations, but these are much relaxed in areas that have low population density, especially if they have few roads. Conveniently, the business is most legal in exactly the places where it's most needed.

The airships themselves are large relative to the payloads they carry, which is not the case for trucks. But this turns out to be less of a disadvantage in the air—three dimensions give you a lot more room to go around obstacles, and three dimensions in a remote area mean there's not much to collide with, either. And this size actually turns out to have some advantages, if you solve backwards from "total volume" to "surface area facing the sun." Solar-powered blimps that can operate all day without recharging are not here yet (or rather, they are possible if you're willing to wildly overpay for bleeding-edge solar panels), but the cost of solar has been declining at around 20% annually for the last decade.2

Airships right now are interesting as a cost-effective substitute for less efficient cargo transportation that's already happening. But they can be much more: one of the barriers to manufacturing exports in the developing world is that exports require infrastructure, and infrastructure generally requires spending, which needs to be funded, ideally from exports. If the fixed cost of participating in global supply chains drops, the beneficiaries will be people in the poorest parts of the world, who will have a shot at joining the global market. Transportation costs are high in the developing world, which amounts to a tax on imports and exports, benefiting no one. If countries can skip some of the expensive and time-consuming process of building road and railroad networks, and focus on ports, they can jump over a significant barrier to higher incomes.

There's a whole lot of land out there, but the good parts are mostly taken, and in the very long run a rich civilization with cheap energy is going to move as much transportation as possible into the air where it's out of the way (or underground, sure, but the that’s a whole new round of fixed asset investment, environmental regulation, etc.). This process can go on for much longer than it looks: declines in marginal cost are often a function of units produced, so every time there's a new use case for batteries and solar, the marginal cost of the next one goes down. Shifting more transportation-related capital expenditures from roads, railroads, ports, and airports to aircraft—i.e. switching from mostly moving land and building structures to building machines—means applying more spending to areas where the experience curve still applies. So the exciting thing about what's happening now is what's coming next.

Thanks to Ben Claman of Buoyant for talking me through the current economics and future potential of cargo airships. All errors mine. For earlier Diff thoughts on logistics, and a different angle on escaping the tyranny of two dimensions, see here.

Diff Jobs

Diff Jobs is our fairly new conspiracy to match interesting people with exciting roles. Some current positions we're working on:

• Blockchain engineers, and infosec engineers with exposure to or interest in crypto (remote).
• Growth and content specialists with an interest in the investment space (remote).
• Strategists who can explain systematic investment strategies to retail investors (remote).
• Business analyst interested in proptech (London).
• Data engineers with an interest in stopping fraud (remote, European time zone).
• Senior sales/business development, fintech experience (SF).

If you're interested in one of these roles, please reach out. And we continue to work on some roles that aren't listed here, so if you're looking for new opportunities in the tech and finance space, we're happy to chat.

If you're hiring (or about to be) and you suspect that the people you're looking for are readers of The Diff, please reach out.

Elsewhere

Bitcoin's Long March: Pensions

I've written quite a few times now about the "long march through the institutions" theory of crypto: every time a new category of investors decide to invest, that validates it for some of their more risk-averse peers. And that validates it for other categories. So if risk-seeking hedge funds take a position, and then more conventional funds do, the next step can be risk-seeking investment banks, or insurance companies. A recent step in this direction: the Houston Firefighters’ Relief and Retirement Fund bought a $25m position in Bitcoin and Ethereum, or about 0.45% of its assets. Every time crypto gets adopted by more cautious institutions, it's getting adopted by the ones that are more willing to hold for a long time. Flows from hedge funds or prop traders are transient, but pensions have a multi-year outlook and predictable outflows.

(Disclosure: like employees of Houston's fire department, I own Bitcoin.)

Last Mile

Amazon has announced "local selling," allowing Amazon customers to order on the site and then pick up the product at the seller's physical store. This is particularly notable because Walmart recently partnered with Adobe to sell subscription access to buy-online-pick-up-in-store technology to other retailers. For a while, it looked like Amazon had ceded this section of the market to other retailers, who were driving some cost and efficiency gains from this: to the retailer, it basically turns the store into a fulfillment center whose delivery drivers are all volunteers who pay for their vehicles and gas. But if there's a way to exchange money for physical products, Amazon will generally try to participate, too.

(Disclosure: Long Amazon.)

Google's Laffer-Style Tax Cut

I've written a bit about the 3D Laffer Curve of platforms: they want to "tax" at the revenue-maximizing rate, but they want to maximize the net present value of the revenue they get, not the amount they earn at any one instant:

The Singapore model [of owning real estate and profiting as economic activity makes it appreciate] is illuminating, because it's a good description of how tech platforms operate. "Digital real estate" is an apt metaphor: there is a market rent for the screen real estate Google or Facebook generates, and they try to charge a high price for it. But they're zoning it for maximum long-term profits, rather than for the highest immediate revenue.

Google has now made another interesting move in that direction, by lowering app store fees for subscription and media apps. For subscription apps, the prospect of paying a continuous tax to an app store has a material effect on their long-term economics, and probably grates on them, too—is the app store really providing a lot of value, relative to the app itself, if customers keep paying for it year after year? Rather than give them an incentive to only market to subscribers off-platform, Google is conceding that it will earn a higher net present value of future fees if it cuts those taxes and makes monetizing through the app store an obviously beneficial channel.

Sales

Perhaps the single most prolific salesperson in history was Airbus's John Leahy, who apparently sold over $1 trillion worth of planes during his career. But there are contenders: Li Jiaqi, a livestream salesman, apparently moved $1.9 billion worth of lipstick in a single day. The evolution of video and social media sales is an interesting one, because the model was almost perfected in the 80s by QVC and HSN, both of which ended up getting better at maximizing the revenue of their core demographic than at expanding it. TV drives plenty of purchase behavior (or, if it doesn't, the advertisers who spend on the order of $60bn annually on it have some explaining to do). Having a tighter feedback loop between ads and the purchases they cause tends to improve unit economics. The model is well-understood, but under-deployed in the US, and represents a pretty interesting growth and ecommerce convergence. (Stay tuned for a future Diff piece digging into this in more detail.)