America's food is made from memory.

A formulation is not a manufacturable product. The knowledge that turns a recipe into a million reliable units lives in the heads of a few operators, inside factories that run dark.

The owner believes his plant runs at 70%. Measured, it runs at 45%. The schedule lives in one person's spreadsheet. The batch record is a 14-page paper packet. The margin on any given production run has never been computed — not once, in the plant's history.

The industry produces enormous amounts of food. It learns nothing from making it.

And now food is entering the largest reformulation cycle in its history — higher protein, higher fiber, less sugar, harder to make — assigned to factories that can't hit their current spec.

Food is going fabless. We're building the foundry.

Semiconductors had this moment: design separated from fabrication, and the foundry that learned from every customer's production became the most valuable company in the industry.

Food is at the same fork. Brands own stories — a $40M brand is 18 people and no factory. Retailers own shelves. Nobody owns intelligent production.

Standard Provisions is building the intelligence layer for food manufacturing — deployed the only way it can be: through ownership. Twenty years of software vendors proved these plants can't adopt. So the intelligence arrives with the deed — factories we own, instrumented in a day, run on measurement instead of memory.

The network that learns from every run becomes the default place where new food comes into the world.

Every run makes the network smarter.

Every production run teaches the system: formulation × ingredients × equipment × process → outcome. More products create more runs. More runs, better translation. Better translation, faster commercialization and better factories.

Better factories win more brands — and every new plant starts from the best run recorded anywhere in the network, not from guesses.

The asset we're compounding isn't factories. It's the only dataset of how food actually gets made.

The benchmark is simple: same line, same installed capital, more sellable product per hour.

Four forces, one moment.

• The reformulation shock. GLP-1s and nutritional density are redesigning what America eats — and stranding capacity in old categories while starving it in new ones.
• Labor is breaking. Food manufacturing productivity is declining while labor costs compound — the only major input getting worse.
• The technology is finally ready. Sensing and AI can now capture what only operator judgment held — installed in a day, at the price of a forklift.
• The entry window. Qualified capacity trades at roughly half its pandemic price, into a succession wave with no natural buyer.

Every prior attempt came from software — locked out. Or from private equity — inside, but with spreadsheets. No one has entered with both the keys and the brain.

We've lived both sides of this problem.

Yoga built Mezcla to a $40M run rate with 18 people — on these exact factories. He is the customer of the assets we'll own, and his family has already proven this playbook in defense manufacturing.

Aryan is an investor who built the intelligence-layer thesis — eighteen months of primary research inside the operations of food manufacturing — and runs capital allocation.

Incubated at Bain Capital. Backed by Dream Ventures.

The first plant is the proof: six months from now, this vision has a P&L.