A viral post claims agtech venture capital "spent $100 billion learning that plants grow at the speed of plants." The numbers are real. The conclusion is wrong. Here's where the money actually went, what it built, and why the return was never in farming — it's in food taking share from a $5.7 trillion healthcare system.
Interactive report · click any bar or card to expand · built to embed in crusonia.io
Nearly half of every agrifoodtech dollar ever deployed went to eGrocery, delivery, and restaurant software — the Instacart / DoorDash / Gopuff complex. Strip the delivery layer and the actual upstream farm-tech pool is a fraction of the headline. The debate is won by refusing to average across incommensurable pools of capital.
Eight segments, wildly different verdicts. "Optionality" = capital that financed durable capability, IP, cost curves, or cleared pathways still being harvested post-bust. "Waste" = a thesis that was structurally wrong and left little behind. Toggle to see how the picture inverts once you isolate the deep-tech that iSelect actually invests in.
Click any segment to see why it scored the way it did.
Per-segment dollars reconstructed from AgFunder category run-rates; optionality % is a defensible judgment, not a measurement. Sources: AgFunder, PitchBook, GFI, Insect Institute.
Most of the individual figures are real — pulled near-verbatim from PitchBook and the World Bank. The distortion is in how they're combined. Click a card for the verdict.
The equity died. The capability didn't. Venture's return to the fund can be zero while its return to the system is large — cheaper cost curves, cleared regulatory paths, platforms bought for parts by people who now move faster. That gap is the whole argument.
The entire US farm gate is ~$0.5T at commodity margins — Paul Daws is right that no venture return lives there. But food that prevents disease isn't competing for the food dollar; it's competing for the health dollar, an ~11x larger pool growing 5–7% a year.
GLP-1s already ran the experiment. A ~$65–70B/yr drug class rewired the American grocery basket in under two years — grocery spend down 5–9% after adoption, snacks and fast food falling, protein and fresh produce rising (Cornell/Numerator, 150k households). Nestlé launched a GLP-1 companion line in response. That is food demand re-rating to a health outcome, at national scale. The food-native version of that wedge — priced against $5.7T, not against corn — is the prize the boom built the ingredients for.
Diet-related disease costs the US ~$1.1T/yr (Rockefeller) — roughly what Americans spend on food itself. Reimbursement rails now forming: Food-is-Medicine Medicaid waivers, medically-tailored-meal savings modeled at ~$13.6B/yr (Tufts), FDA front-of-pack labeling and an ultra-processed-food definition in progress.
Real platform disruptions look like failures at year ~10 and compound after — the value hides in a platform early equity marks can't see. Ag/food's 2015–2022 platforms (gene editing, precision fermentation, ag-CV) are being written down today: the same point on the same curve. A dated 10-year fund can't hold the option long enough to exercise it. An evergreen one can.
Striped = "looks like failure" phase · solid green = first real proof/return. This is why the structure critique and the prize are the same argument from two ends.
10x-in-10-years can't underwrite biology. Food & water need infrastructure capital — evergreen vehicles, farmland, corporate balance sheets — not dated venture.
Nobody serious invested in yield. The boom revealed information and left reusable platforms (Hayek/Schumpeter/Romer). The 10x comes from food taking share from healthcare — and the fix is redesigned venture, not infrastructure capital.
VC funded capabilities investors imagined, not what producers demanded — complexity grew faster than value. The prize is the integrating architecture.
The synthesis: Concede the arithmetic (dated 10-year funds can't beat biology's release schedule). Keep the function (venture's job is information revelation + spillover — and on that scorecard the deep-tech boom worked). Fix the structure — evergreen mechanics, participation rights, information-priced optionality, and a growth-stage instrument that carries platforms from Series C to deployment, pointed at the one denominator big enough to pay: healthcare.
The ledger says the capability got financed and the value leaked. The companion essay is the forward thesis: the contracts, offtakes, and financing instruments that decide who captures it.