Foodtech Stopped Trying to Invent Food and Started Trying to Run It

What's New

Fewer foodtech companies are raising money than at any point on record, and the ones that clear are commanding more. First half investment reached $3.6 billion across 359 deals, a pace that annualises to roughly $7.2 billion and would land 15.3% below last year's $8.5 billion, per PitchBook's H1 2026 Foodtech Report, published August 27. The correction that began three years ago has still not found a floor. Underneath the decline, though, sits a compositional shift the funding total hides entirely: median deal size has roughly doubled since 2022, from $2.2 million to a record $4.3 million, and the median Series A hit a record $9.7 million. The capital that remains is going to operations software rather than to novel foods, which changes what the category is.

Why It Matters

The category is repricing around what kind of business it wants to be. Capital is moving away from novel foods and toward software that runs food operations, which changes the return profile, the capital intensity and the buyer universe all at once. For GPs, that means the diligence skill set required is closer to enterprise software than to biotech. For platforms serving the sector, it means the assets being underwritten increasingly carry recurring revenue and software margins, which are far easier to value and to monitor than fermentation capacity.

Big Picture Drivers

• AI moved into operations, not invention: Demand forecasting, supply chain software, quality inspection and restaurant operating systems drew most of the AI native activity, rather than novel ingredient development.

• Labour cost is the underlying constraint: Agentic voice AI for restaurant front of house and computer vision robotics in the kitchen are the two fastest moving clusters, and both target the same thin margin problem.

• Concentration is the defining pattern: Median rounds rose at Series A, B and C simultaneously even as deal count fell to record lows, which is a flight to quality rather than a broad retreat.

• One listing carries the exit story: Sunmi's $1.3 billion public listing accounts for most of the improvement in exit value, and outside it the market stayed acquisition led with undisclosed terms.

• Alt protein is consolidating, not validating: NovoNutrients and BettaF!sh both changed hands without disclosed terms, which reads as tidying up rather than as capital returning.

• Personalized nutrition is the standout: The category raised roughly $256 million across 12 deals in Q2 and passed its entire 2025 total in three months, helped by GLP 1 driven demand.

By The Numbers

• $3.6 billion across 359 deals, in the first half, against $4 billion across 493 deals a year earlier, so capital fell far less than company count.

• 164 rounds in Q2, the fewest of any quarter on record going back to at least 2016.

• $49.5 billion at the 2021 peak, against a $7.2 billion annualised pace today, a decline of roughly 85%.

• $4.3 million median deal size, a record, up from $2.2 million in 2022 as investors write fewer and later cheques.

• $1.9 billion of exit value across 44 exits, already above the $1.3 billion recorded in all of 2025 but well below the decade's $4.8 billion median first half.

• $1.7 billion of North American deal value, against $0.9 billion in Europe and $0.7 billion in Asia, close to half the global total.

Key Trends to Watch

• The 160 deal floor is the test: PitchBook flags whether Q3 count holds above that level as the signal that the correction has bottomed. Another leg down would confirm structural rather than cyclical decline.

• Sunmi decides whether the IPO window reopens: One listing does not make a market, but its aftermarket performance determines whether other late stage foodtech names attempt public exits.

• Alt protein needs to finish consolidating: The field has to clear enough failed entrants to leave survivors that investors will fund again, and that process is not complete.

• Restaurant tech attracts both buyer types: Q2 produced seven exits in the segment across a public listing and strategic acquisitions, which is rare in a category this small.

• Software margin becomes the screening criterion: Capital intensive deep tech is losing to businesses with visible recurring revenue, and that filter will harden if the funding environment stays tight.

The Wrap

The most defensible reading is that foodtech is finishing a transition from a science category to a software category, and the funding decline partly measures the exit of the science. That is not obviously bad, since the businesses now attracting capital have better margins and clearer buyers, but it does mean the sector's addressable market is smaller and more contested than the 2021 narrative implied. Anyone building for this market should note the practical consequence: the assets are converging on ordinary enterprise software characteristics, which means the specialist foodtech tooling thesis is weakening at exactly the moment the category needs it least.