Here’s a slice from an a16z podcast with co-CEO and founder Ryan Petersen of Flexport:
The fundamental problem in supply chain is that we’re a chain. There’s many companies. In fact, on a typical transaction where we’re moving let’s say a pallet of goods from a factory in pick a city – in Hangzhou, China to St. Louis, Missouri. On a transaction like that, you’ll have at least 12 companies that either touch the goods physically or are involved providing capital, like trade finance, or a letter of credit, cargo insurance, bank payments. At least 12, and as many as 20 companies, are involved in that transaction. You get information bottlenecks.
The goods bottlenecks are famous, and we see the ships waiting offshore and all the problems. But what’s happening in parallel is that the data’s not flowing. People at each stage are having to rekey data into a system, or not bothering, forwarding PDFs… If you had perfect information, you would make sure the truck is waiting at the airport when the plane lands or at the seaport when the container shows up. You would get it all dialed in. Every truck would be scheduled appropriately, and you would be re-planning every five minutes so that if there’s any delay, you’re automatically generating a new plan and reallocating containers. Almost none of that can happen because there’s not a system. It’s 12 different companies, each running their own IT system, or in many cases, guys in warehouses with clipboards and no IT systems at all.
There’s 2 possible solutions. One is that a giant mega corporation owns all the assets door-to-door and creates 1 IT backbone for everybody. The other is to use ML and data science and make it in the interest of asset owners to come onto the platform.
DoD’s supply chains are equally as gnarly, and DoD has terrible visibility down the tiers of supply chains. The amount of lower-tier components coming from China is probably quite high, just as it is coming out that Chinese, Russian, and Iranian defense products have components coming from the United States.
All this is exacerbated because the incentives to outsource in DoD are high. Defense primes used to perform up to 70 percent of the value of an end-item in-house. Now, 70 percent or more is outsourced. I think this has occurred because a mix of (1) just-in-time efficiency notions; (2) globalization; and (3) defense contract incentives that pay a percentage of profit on-top of total sales, not capital invested.
10% profit on a good that is 70% outsourced is not equal to 10% profit on a good that is 30% outsourced. The former is way more attractive to a company.
This part was interesting as well, and something DoD should probably think about in its own context:
I think Amazon’s the most interesting, important company in the world, frankly. They have figured out a lot around how to be entrepreneurial at scale… It’s pretty rare to see a big giant corporation that’s figured out how to really be a team of small companies, small business units. For me, it’s a wrinkle in the law of large numbers. The law of large numbers says, “you get really big, your growth rate slows down.” But if, in fact, all you are is an agglomeration of small companies, each one of those small units can grow really fast, then all of a sudden the whole thing can grow really fast.
There’s always a trade off. Simple, vertically aligned supply chains often result in significantly increased costs due to a smaller supplier base and less competition.