A startup says it has cracked vegan cheese by routing production through existing microbrewery infrastructure. Separately, the Mexican government is rerouting a high-speed rail line after archaeologists discovered 16 cave drawings and petroglyphs along the planned corridor. These are, on the surface, unrelated stories. Below the surface, they are both about the problem of inherited infrastructure: what do you do when the thing you are building runs into something that was already there, and that prior thing turns out to matter more than the plan?
Working With What Exists
AuX Labs is not building new fermentation facilities. It is inserting its process into the existing distributed network of microbreweries, which have the tanks, the temperature control, the microbial know-how, and the spare capacity. This is infrastructure arbitrage, and it is a genuinely clever solution to the vegan food industry's persistent cost and scale problem. The train diversion in Mexico is the inverse: infrastructure ambition meeting cultural artifact, and culture winning. Both situations require a form of design humility that is rarer than it should be. You do not always get to build in a vacuum. For founders in the food-tech space navigating infrastructure costs, TurboFund's accelerator guide highlights several programs with specific agri-food and biotech tracks.
The Supply Chain as Archaeological Site
The Mexico cave art story is extraordinary on its own terms. The drawings date back thousands of years and were found along a route the government had already committed to build. The decision to reroute is not legally required under Mexican law in all cases, which makes it a genuine policy choice to let the past redirect the future. The microbrewery model makes the same kind of choice: instead of building new fermentation infrastructure, you map what already exists, find the gaps, and design a product that fits the existing topology. A 2025 paper in Nature Food by McClements and colleagues on precision fermentation scaling found that distributed production models consistently outperform centralized facility builds on both cost and carbon metrics when the distributed nodes already have biological containment infrastructure. Microbreweries, it turns out, are already biologically contained environments. The infrastructure was hiding in plain sight, like cave art under a train route.