Animal Health AI Hits $2B as Halter Closes $220M Round

Daniel Mushrush runs cattle on his family ranch in Chase County, Kansas. He is the fifth generation of his family to do it, and this year he is doing it differently. Every cow on his property wears a solar-powered collar. The collar feeds behavior data into a platform called Halter, whose engine — its founders call it a cowgorithm — decides where the herd grazes next, when a cow is in early labor, and which animal is about to go off its feed. Mushrush says he is getting hours of his day back. His calves are coming in heavier. The collar is billed as a flat monthly subscription per animal, priced the way you might price software and deployed the way you might deploy livestock feed.

Halter — a New Zealand company founded in 2016 by Craig Piggott, who grew up on a dairy farm — closed a $220 million Series E this week at a $2 billion valuation, led by Peter Thiel's Founders Fund. That is the largest agtech funding round reported in recent memory and the clearest public signal yet that animal health AI has moved out of the research phase and into scaled production.

It also did not arrive in a vacuum. In the same five-day window, a cultivated-meat company extended its regulatory footprint in Singapore to a second animal species, the global animal health market got a fresh long-horizon forecast, ag lenders got a compressed AI deployment playbook, and researchers documented a federal food recall system that is still structurally running years behind the food it is supposed to protect. Animal agriculture and the infrastructure around it all moved in the same direction at the same time. The crop side of this story has been familiar for a while. The animal side is where the capital and the regulation are both finally catching up.

Halter's Round Redraws Livestock AI

What Halter sells is a solar-powered smart collar. What it sits on top of is a deep record of cattle behavior — grazing, walking, standing, ruminating, recovering from birth — collected across working farms in New Zealand, Australia, and the United States. The model trained on that record is what the company actually charges for. The collars are hardware. The subscription is software. The edge — what a new vendor cannot easily replicate — is the behavior corpus itself.

That matters for two reasons. First, pricing a farm platform per cow per month only makes sense when the product is already deployed at scale. Halter is in the field on a large installed base across multiple countries, and Piggott has named the UK, Ireland, and South America as the next markets on the expansion schedule. Charge per cow, deploy per ranch, and the software subscription becomes the unit economics of a farm.

Second, the round is not priced against a narrow cattle-collar category. It is priced against the broader animal health AI category that livestock operators will need if their entire herd is going to run on models rather than memory. That is the bet Founders Fund made. When Thiel's firm leads an agtech AI round at this valuation, livestock AI moves into the same bucket as priced categories with public comparables — not a whitespace thesis being tested.

The customer case on the Mushrush ranch is worth pausing on because it is not a tech-versus-farmer story. Mushrush is a working fifth-generation rancher. The collar is built into his daily operation. The outcomes — less time spent gathering cattle, heavier calves, fewer missed health events — are the kind of operational wins that sell into ranching culture without pitching against it. The company's founding story (young New Zealander on a dairy farm who built a collar, then a model, then a company) is color. The business story is that a cattle platform now commands late-stage Series E money and cross-border deployment economics.

The Animal Health Market Is Growing Up Around It

Halter's valuation sits inside a category that was already on an upward slope before the round closed. Research and Markets published its long-horizon animal health forecast this week, and the picture it paints is of a mature, pharmaceutical-heavy global market with steady projected growth through the mid-2030s. The incumbents it tracks — Bayer, Elanco, Merck, Zoetis, Boehringer Ingelheim, Ceva, and the other majors — are not AI-native businesses. They are pharma, diagnostics, and imaging companies that have been compounding revenue in animal health for decades. What the forecast explicitly names as the innovation vectors inside the category, though, are advanced diagnostics, imaging, and precision medicine. Those are the exact surfaces that a livestock AI platform or a veterinary AI product would plug into.

The state-level picture lines up. Pennsylvania's cooperative extension reporting documents broad AI uptake across the state's agribusinesses, with livestock and dairy operations identified as the near-term growth surface. Penn State's vision-guided mushroom harvesting work is cited alongside it as a template for what happens when sensors, models, and robotics are wired into a specific species — the rigor looks the same whether the subject is a mushroom, a dairy cow, or a calf on pasture.

Viewed together, the forecast and the state-level adoption data say the same thing in different dialects: animal-focused AI is now a priced category with incumbent customers, not a pitch deck with whitepapers attached.

The Money Layer Catches Up

The Halter round alone would have anchored the week. But the financial and data plumbing around agricultural AI moved alongside it, which matters for anyone who has to write the checks.

In AgFunderNews, Cameron Burford of Growers Edge laid out what he described as a compressed AI evaluation cycle for ag lenders. The framing is straightforward: treat AI applications the way a farm treats a crop year. Scope in weeks, run a short pilot, then either scale it or kill it. Applications that cannot prove their economics inside a single deployment cycle do not graduate to the next one. Budgets sit at the application level rather than at an enterprise-software contract level. The implicit challenge to ag-lending's default SaaS-contract cadence is direct: if your AI vendor needs a year to show value, the business case probably is not there.

This is where horizontal agent infrastructure earns its keep. Running AI on a short cycle at a per-application budget means somebody has to enforce the budget, hold the audit trail, and make the spending legible when the risk committee asks. Budget caps, audit trails, and per-run spend controls — the kind AgentPMT builds — are what a short evaluation cadence actually runs on. Without them, "treat AI like a crop year" quickly turns into open-ended cloud bills and vendor invoices no one can reconcile.

Alongside the lender playbook, Datavault AI and AgSensor Solutions announced a consulting partnership aimed at identifying and tokenizing high-value agricultural data — soil sensing, carbon and sustainability data, regenerative ag ESG records, agricultural IoT feeds. Datavault's blockchain tokenization products (the company's own naming: Information Data Exchange, DataScore, DataValue) are designed to treat on-farm data as an asset class that can be valued, traded, and paid for. The company's CEO framed it plainly — data, he said, is the new crop for the modern farmer. AgSensor's co-founder pointed out that the volume of data already flowing off the field is still structurally undervalued.

Whether that framing holds up commercially depends on whether buyers actually materialize for the tokens. What matters for the larger picture is that the agriculture AI conversation is no longer only about models. It now has a parallel conversation about what the data underneath the models is worth. The two conversations happening in the same week is what it looks like when a sector is being built out rather than demoed.

The personnel signal rounded the week out. Carbon Robotics, which builds the LaserWeeder autonomous weeding system and crossed a revenue milestone last quarter, named Kevan Krysler — formerly a KPMG partner, VMware's chief accounting officer, and Pure Storage's CFO — as its new CFO in March. That is the kind of hire a company makes when it is preparing to look like a grown-up finance organization: audit committees, quarterly discipline, and the reporting machinery that institutional investors expect. Farm automation companies are beginning to look less like acquisition targets and more like public-company candidates.

The Protein Edge and the Recall System That Isn't Keeping Up

The same week pushed cultivated protein forward. Parima, the French biotech behind the Gourmey brand, received Singapore Food Agency clearance to sell cultivated duck. That extends Parima's existing Singapore approval for cultivated chicken and makes it the first cultivated-meat company cleared in that market beyond a single species. The company has additional regulatory filings active across the EU, UK, and other jurisdictions.

In parallel, the South African biotech firm Immobazyme, working with backing from the Council for Scientific and Industrial Research, announced domestic production of fibroblast growth factor 2. FGF-2 is the cell-culture growth factor that has historically sat on the cost line as one of the single biggest inputs for lab-grown meat. Local biomanufacturing of FGF-2 lowers the input-cost floor for the whole category.

Neither story is flashy on its own. Taken together they are what regulatory and supply-chain maturation look like in a category that has spent most of its existence shipping press releases and demo tastings. Parima's pipeline of pending approvals is a real regulatory posture. Immobazyme's FGF-2 production is a real input-cost change. Cultivated protein is starting to behave like a supply chain rather than a research program.

The food safety system at the end of that supply chain, by contrast, is not modernizing at the same rate. A Public Interest Research Group report this week documented a federal recall system that routinely takes months or years to warn the public about contaminated food. The report catalogs outbreaks where the recall process ran longer than the shelf life of most of the affected products: an infant-formula botulism case that dragged on for years, a Listeria case tied to frozen supplemental shakes that dragged on longer, and a ready-to-eat pasta-meal outbreak whose first public recall arrived well after the affected product had already moved through grocery chains. Companies that identify contaminated product are required to notify FDA and USDA. They have no corresponding obligation to contact stores, restaurants, or consumers directly. The FDA's Traceability Rule, which would tighten that loop, has been pushed back to 2028. PIRG puts the annual cost of foodborne illness in the United States at $75 billion.

This week's real-time recall snapshot is exactly what PIRG described. E. coli in a small-dairy raw cheddar. Listeria cases in Canada. A rodent-contamination recall in the UK. Glass-fragment contamination in a Hong Kong product. All currently active, all in the same week that Parima's duck crossed Singapore's regulatory line and Halter's cowgorithm cleared its new valuation.

AI food safety is the part of the picture that commercial AI has barely touched. Ranchers are running livestock on platforms trained on deep behavioral records. Cultivated duck is clearing regulatory gates in Singapore. Ag lenders are running AI on compressed evaluation cycles. The recall system that protects consumers from any of it is still, structurally, a paper pipeline.

What's Worth Watching

The headline is Halter, priced at livestock-AI scale and funded at late-stage Series E size. The broader read is that animal-focused AI — collars on cows, bioreactors in Paris and Singapore, the ag-lending cadence that funds both — all moved in the same direction in the same week. Three things will determine whether that continues.

First, whether Halter's new capital actually lands in the UK, Ireland, and South America on the timetable Piggott has sketched, or whether deployment economics slow down outside the markets where the company already has dense coverage. Second, whether Parima's pending regulatory filings clear in the EU and UK, or whether the EU's cultivated-meat approvals process drags the same way the U.S. recall process has. Third, whether the FDA's Traceability Rule lands as scheduled or slips again.

If the first two happen and the third does not, the production end of food agriculture will be running AI-native while the consumer end still runs on quarterly recall press releases. That is the actual shape of the gap — between the systems that make food and the systems that recall it when something goes wrong. AI food safety is the least funded, least watched, and most obviously necessary corner of the entire agriculture AI picture.

Sources

After growing up on a dairy farm, this Peter Thiel-backed founder is using AI to save cattle ranching — Fortune
Guest article: The ag lender's guide to AI investment — AgFunderNews
A tech CFO's next act: bringing AI to farmers — CFO Brew
Datavault AI and AgSensor Solutions Announce Consulting Partnership to Tokenize High-Value Agricultural Data Assets — Datavault AI Inc.
How Artificial Intelligence Is Changing Agriculture — Lancaster Farming
South Africa now makes critical molecule for lab-grown meat — Cultivated Meat News
Animal Health Market Intelligence Report 2026-2034 — GlobeNewswire / Research and Markets
Parima Gets Singapore Approval to Sell Lab-Grown Duck Under Gourmey Brand — Cultivated Meat News
Recalls and Alerts: April 16, 2026 — eFoodAlert
Report shows shortcomings in recall system — Food Safety News