Digital Shōsha: How America Could Rebuild Its Industrial Trade Infrastructure
Erebor positions itself as a test case for linking finance, technology, and production
When Silicon Valley Bank collapsed in March 2023, it erased a key piece of the innovation economy’s financial infrastructure. SVB was more than a bank; it was the only major institutional lender designed for startups, filling the space between equity and credit. Its venture debt model gave early-stage software companies extra runway without forcing them to give up ownership too early. With SVB gone, the burden of financing innovation shifted almost entirely onto venture capital. But venture capital was never designed to meet the full range of needs in today’s tech economy.
That shift might have been manageable a decade ago, when technology mostly meant software. Today, it includes manufacturing, defence, energy, and AI hardware—sectors with far more complex and capital-intensive financial needs. Without institutions equipped to handle working capital, supply chains, and cross-border trade, these industries are left exposed. Japan solved similar problems through its trading houses after the Second World War. The US has never built an equivalent.
This brings me to Erebor, a new bank backed by Peter Thiel, Palmer Luckey, and Joe Lonsdale, which filed for a national charter in June 2025. The bank focuses on manufacturing, virtual currencies, defence, and AI—exactly the areas where institutional gaps have long held back American ambitions. But Erebor may represent more than a well-timed financial play. It might be an early version of something more ambitious.
Could we be seeing the emergence of a digital-age sōgō shōsha? The timing, investor base, and technology stack suggest the conditions are in place. If my hypothesis holds, Erebor could help define a new model for financing industrial revival in a fragmented and strategically contested world.
1/ The new bank will look nothing like the last.
The idea that a new paradigm would call for a new kind of banking is not new. One precedent I often cite is Henry Goldman, son of Goldman Sachs's founder Marcus, who seemingly invented the discounted cash flow to account for the rise of mass retailers. As Charles D. Ellis recounted in his landmark book about Goldman Sachs, The Partnership:
The public securities markets, both debt and equity, had always been carefully based on the balance sheets and the capital assets of the corporations being financed — which is why railroads were such important clients. To expand, United Cigar needed long-term capital. Its business economics were like a “mercantile” or trading organization’s — good earnings, but little in capital assets.
In response, Goldman “developed the pathbreaking concept that mercantile companies, such as wholesalers and retailers — having meager assets to serve as collateral for mortgage loans, the traditional foundation for any public financing of corporations — deserved and could obtain a market value for their business franchise with consumers: their earning power.”
Silicon Valley Bank, in many respects, followed the same idea. It pioneered new banking methods suited to the dominant companies of the age of computing and networks, underwriting loans partly on the basis of how much equity a borrower expected to raise, and building a warrant portfolio across thousands of venture capital-backed tech companies. That SVB failed should not be held against it. As with Napster and the music industry, or MySpace and social networks, we should always learn from the collapse of a pioneer and restart the effort on firmer ground.
But the world has moved on. The context today, that of the late cycle, is already different from the one that drove SVB's rise during the heyday of software eating the world. We do need a new kind of bank, but the working hypothesis should be centred on late-cycle dynamics: the shift toward hard assets and physical infrastructure, where software is no longer separate but increasingly embedded—merging with the physical world it once only managed from afar.
This brings me to today’s topic. Two recent articles announced the birth of Erebor, a new tech-focused bank backed by Peter Thiel, Palmer Luckey, and Joe Lonsdale. One piece appeared in the Financial Times, the other in American Banker.
We don't yet know much about Erebor, but in this essay I want to offer a hypothesis: with Erebor, we may be seeing the rise of a digital sōgō shōsha, built to do for America what the legendary original Japanese trading houses did for post-war Japan.
2/ Japan's post-war industrial boom was built atop the machinery of the trading houses.
The sōgō shōsha made the difference between two very different futures for manufacturing firms.
Without them, companies had to raise equity, build their first production run, and then often stalled. They were squeezed by suppliers who charged too much due to lack of scale, and overtaken by larger rivals who copied their ideas and moved faster, especially in foreign markets. Even with a good product, early clients, and investor backing, they struggled to compete.
With trading houses, the picture changed. Firms like Mitsubishi Corporation, Mitsui & Co., and Itochu acted as a mix of investment bank and global logistics operator. They funded production, moved goods, arranged supply chains, and took equity stakes across industries. They financed upstream assets like Australian iron ore mines and Brazilian soybean farms, while distributing Japanese products worldwide.
The classic sōgō shōsha model played out across many sectors of Japan’s post-war economy. A few examples show how these trading houses enabled Japanese firms to scale globally with far less friction:
When Toyota needed reliable steel in the 1960s, Mitsui wrote more than a cheque. It used its global reach to source raw materials across continents and secure downstream sales for finished vehicles. That let Toyota focus on what it did best: production efficiency, not navigating global logistics.
When Japanese electronics firms expanded into the US and Europe, trading houses like Mitsubishi Corporation financed upstream supply deals, managed shipping and customs, and helped place Japanese TVs, calculators, and radios on foreign shelves. They reduced the time and capital needed for manufacturers to go global.
When Japan's tuna industry sought access to international waters, companies like Maruha relied on Itochu and other trading houses to finance vessels, arrange joint ventures, and build cold storage capacity at overseas ports. That infrastructure gave Japanese fisheries a foothold in distant markets they could never have reached alone.
All in all, the trading houses delivered three essential services to the manufacturers partnering with them. First, they financed long-term commodity contracts, keeping input prices stable. Second, they built distribution networks to open foreign markets. Third, they gathered and shared intelligence that helped firms adapt products to regional demand.
Yes, Japan had the Toyota system, a radical new way of making things. But that alone wouldn't have changed the game. It's one thing to invent cinema. It's another to build Hollywood. What the sōgō shōsha added was the institutional infrastructure to turn technical advantage into national economic strategy. Japan's post-war boom was about more than better manufacturing. It was about building sophisticated intermediaries that linked domestic production to global markets.