The Commerce Department Is a Hedge Fund Now

‍

‍

The Largest American Company You Don’t Know.

When Harold Rogers, Coupang’s interim CEO, took to the stage, he made a bold claim that established his unique place in the world of cross-border commerce and logistics: “Coupang is the largest American company you don’t know.” It’s in the Fortune 150, is headquartered in the US, and generated $35B in revenue in 2025. The company is also the second-largest employer in South Korea, and operates in 190 regions and countries. Coupang is arguably the rarest kind of consumer brand that acts as a bridge between the US and companies with US interests. 

Coupang sells against speed, selection, and price, the three trade-offs every retailer pretends not to be making, and refuses to compromise on any of them. “Rocket Delivery” moves millions of SKUs from order to doorstep between midnight and 7 a.m., with no delivery fee, often at prices about 20% less than the local grocery store. AI predicts what you'll order before you order it, forward-deploying inventory within 5 miles of your home before you hit "buy."

Rogers revealed that Coupang’s cross-border business model has a much broader halo effect. The company sold $5 billion in US goods to Asia last year and has been building an $84 million AI startup portfolio since 2023. One of the most exciting investments was a $10M check to Texas-based Contoro. This robotics shop provides autonomous robots with suction-cup arms that seamlessly unload trailers and containers.

A US-headquartered operator that’s fluent in Eastern-hemisphere demand, Coupang is quietly underwriting the next layer of American logistics infrastructure.

‍

‍

A 2009 Platform is Secretly Driving the Creator Economy.

2025 was Kickstarter’s best year—record revenue, record GMV, and record users. And 2026 is already pacing 52% ahead of it. Not the headline you expect from a 2009-era platform that the market keeps trying to bury.

Kickstarter’s CEO, Everette Taylor, believes that the surge is half macro and half cultural. Layoffs are pushing more people into independent work. Every social platform is pulling the creator economy forward. And finally, a meaningful slice of Kickstarter’s growth is now represented by consumer products and AI-enabled hardware that are looking for a pre-launch audience more than a Series A. Some of the platform’s most successful campaigns have raised $20-40M through pre-orders and audience engagement plays, not traditional crowdfunding campaigns. 

“It's been a place where a lot of people are now [realizing] there are millions of people using this app and website every day,” Taylor said. “And also, this is an age of independence due to the economy. People are being forced into being creators in entrepreneurship, or they just want to be independent now.”

While many platforms are wishy-washy on their AI policies, Kickstarter has a concrete stance to “protect creative people,” Taylor said. Creators must disclose AI use in their projects; if they don’t, they’re removed from the site. Internally- and externally-built software helps monitor all projects, but the community also has a vested interest in supporting what they believe is “authentic craftsmanship.” The goal, he explained, is to provide tooling that serves both creator sets well. 

‍

The New Railroad Rush Transports Agents. 

If you only steal one number from this entire summit, steal this one: 51% of internet traffic today is generated by autonomous workers. The Age of Autonomy is already driving major traffic shifts on the open web, and Kate Johnson of Lumen Technologies believes the network limitations are creating an “existential matter” for the US. 

“It’s all about the competitiveness of the country, about productivity, economic growth, innovation, quality of life, but also cybersecurity. That’s what’s at stake,” Johnson said. “The United States and our economy have to lead now.”

Companies need to move data in real time, which, Johnson explained, is not how legacy telcos operate. Her thesis is that networking has been the line item pushed to the end of a project for 30 years. But when you constrain bandwidth for cost control, you’re constraining data flow. Agents proliferate everywhere, GPUs sit everywhere, and data lives everywhere. “Data flow is what returns value on GPUs, which have the shortest economic half-life of any asset on planet Earth,” she said. “Why constrain data flow when, in fact, those GPUs should be washing in it all the time. The focus should be on the productivity of the AI system, with the network considered at the same level as GPUs, power, energy, data, talent, all the other models, and everything that you think of from an AI system top to bottom.”

Lumen has gone from 12 million fiber miles when Johnson took the job in late 2022 to 58 million planned by 2031, all of which is pre-contracted by hyperscalers, neocloud companies, and technology companies. This is the new railroad rush. But there is a bottleneck: permitting. An inline amplifier can be installed in two weeks, but it takes about 18 months to get the appropriate permits approved. 

“Right-of-way permitting, all the fees associated with it, they’re all negotiated, community by community, local municipality by local municipality,” Johnson explained. “From a policy reform perspective, the faster we can go, the better, because the digital layer is important, but the physical layer has to be there. Building it out as quickly as possible is existential.”

‍

Your Next Officemate May Be a Data Center.

Diane Hoskins of Gensler and John Santora of WeWork sat down for what was billed as an office-vacancy panel, but it turned into a philosophical discussion about how organizational cultures are pivoting back toward in-person connection as AI permeates their workforces.  

Hoskins’ data is the part to underline. For the first time in two decades of Gensler workplace surveys, she’s seeing a meaningful spike in in-office learning activity. Employees want to gather, in person, to learn how their work is being reinvented by AI. “There is no manual,” she said. “It’s being invented as we speak.” Presence is now an AI adoption requirement, not a culture preference.

The supply side is being rewritten in parallel. Gensler launched an algorithm to help building owners and city planners determine whether vacant buildings can be repurposed. Suburban office parks are quietly converting to residential, data-center, and university uses. Santora is pushing landlords to adopt more flexible lease structures that allow tenants to add or shed floors as the company's size and shape change. In his words, if you think you know what you need over the term of the lease, “you’re wrong.” 

“Space usage needs to change and adapt to meet the market,” Hoskins said. “Our buildings are as much infrastructure in the built environment in our cities as the roads and all other component parts.”

Care by Design. 

The Christensen overshoot framework we ran through in Wednesday’s newsletter found another compelling application when Marc Tessier-Lavigne of Xaira Therapeutics and Jim Davis of Quest Diagnostics discussed the disruption happening in healthcare. The folks driving it are a distinct category of operator who refuses to play by the same trial-and-error rules.

Tessier-Lavigne is two years into a $1 billion build at Xaira, an AI company aiming to rewrite the 13-year, $4B timeline it takes to get a drug FDA approved. 90% of clinical trials fail during this journey, two-thirds because the drug was matched to the wrong patients. Xaira uses AI to skip screening millions of compounds and go straight to designing a molecule against the target. “You design the needle,” he said, “instead of looking for it in the haystack.” He’s deliberately aiming at the targets conventional methods can’t reach, what he calls the high-hanging fruit, because that’s where AI moves the success curve.

Quest is on the consumer-facing side of the same story. In 2025, Quest tested 240M people in the US, with 65% of those tests associated with chronic disease and 35% tied to general health and wellness. Davis openly stated that he wants to invert that ratio by bringing Apple Watch, Oura, and Whoop biometrics into a single AI engine that gives patients an integrated read before an appointment even takes place. 

The goal, according to Davis, is to get ahead of these chronic diseases and provide early insights into what’s happening inside the human body. That way, patients can “be the CEO of their own human body and take the appropriate action.” That is the longevity-economy thesis Future Commerce has been writing all year.