Technology has made factory farming the predominant form of animal agriculture worldwide. The global meat and dairy supply has never, in human history, been bigger. Agricultural advancements keep animal feed cheap and abundant. Antibiotics enable animals to be kept in close quarters and poor conditions. Modern supply chain infrastructure and logistics systems deliver animal products widely and cheaply. Next up: AI. 

These days, the talk of the agtech world is precision livestock farming. 

PLF is a suite of sensor technologies that collect farmed animal data to be processed through machine learning algorithms. The results help automate operations on the farm — think GPS trackers, collars that monitor feeding and resting, microphones to measure your cow’s coughs, thermal teat sensors, and recognition algorithms for everything from the faces of individual pigs to the lameness in dairy cows. This technology is in the early stages of adoption across the United States. But we don’t yet know what impact it will have on the over 90 billion land animals and many more aquatic animals used and killed for food every year. 

Throughout my life, I’ve looked for ways to mitigate animal suffering, most recently by co-founding a nonprofit to improve coordination in promoting the well-being of farmed animals. For the past year, I’ve been fascinated by the exponential advancement of AI technology. I wanted to learn more about the current state and future prospects of PLF from the perspective of the animal agriculture industry. To do this, I decided to attend the Animal AgTech Innovation Summit in March 2024.

Before the conference, I’d assumed the PLF industry was thriving. A 2022 report by Future Market Insights predicted that the PLF sector would have an annual growth rate of 8.14%. PLF was the buzz of trade journals and the subject of a rapidly growing academic literature. And the fundamentals looked good: Between the rapid advancement of AI technology and the increasing number of animals that come from large farms where individualized farmer attention to animals is less practical, PLF seemed like a clear technological step forward. The reality I encountered at the summit was far more complex.

Some attendees were excited about the revolutionary potential of PLF, but just as many pointed to the technological and structural barriers keeping farmers from adopting it at scale. An investor panel on the last day set a somber tone for the sector’s financial outlook. The main selling points highlighted for PLF were sustainability and profits, with a focus on helping producers. And while there were some mentions of animal welfare, perhaps unsurprisingly, it didn’t seem to be at the forefront of anyone’s mind. 

I came to the conference thinking I’d learn about what a world with widespread PLF would look like for farmed animals. But I left with a much more complicated set of questions about the economics of animal agriculture, the challenges of adopting new technology, and how we might be able to leverage it for a more humane future.

Hardware hiccups 

PLF systems are AI, but unlike ChatGPT, they involve a complex blend of hardware development and software customization to meet each farmer’s needs. At the conference, every company I saw offered both hardware and software, and startup founders consistently emphasized that each farmer has unique requirements. Monitoring equipment needs to be custom-installed on site and endure harsh farming conditions. Sometimes the animals have to wear it, and such wearables will need periodic replacement. 

PLF technologies are still developing. They will likely require several iterations to perfect, which is complicated by the fact that animal products have long sales cycles: Animals need to be born, raised, and slaughtered before profits can be evaluated and PLF systems can be adjusted. Farmers, already overwhelmed with daily tasks, are reluctant to troubleshoot new technology during its trial phase. 

This is assuming the farmers are technically equipped to do so — and they may not be. According to a USDA report, 27% of American farmers used some kind of precision agriculture system in 2023. ¹ From the same source, we learn that just 69% own a computer and 85% have access to the internet (the national averages are 92% and 95%, respectively). Industry publications report the same concerns: A report from swineweb.com listed digital literacy and rural broadband and connectivity issues as two major barriers to PLF adoption, along with the challenge of standardizing data between the bewildering variety of available products and tools. ²  

The need for secrecy

During the conference, it became clear that many producers are not so eager to install cameras or sensors. This is in part due to the logistical challenge: Many are still in the process of digitizing their data from analog records (remember, a surprising number of farmers still don’t have access to the internet). Others object to having their animals monitored at all. 

It is larger farms that stand to benefit the most from automated PLF technology. They already rely heavily on mechanization over manual labor, resulting in fewer workers directly monitoring each animal. They are also frequently the subject of undercover investigations that reveal inhumane living conditions. Since 2011, six states have passed ag-gag laws, making it illegal to record, possess, or distribute photos, videos, or audio on a farm. This raises the question: If producers are hesitant to allow others to record on their property, why would they be open to installing cameras themselves? Indeed, a poll of Midwest farmers conducted by South Dakota State University found that data privacy was the second-biggest concern about PLF technology, behind only cost. 

As a result, agtech startups are designing their systems to prioritize data security so that they can assure farmers that no one, not even the PLF companies themselves, will be able to access the on-farm monitoring data without express permission. However, this emphasis on increased data privacy makes it harder for auditors, the government, and the public to obtain reliable information about how the animals inside these facilities are treated.

A harsh funding landscape

The biggest update for me from the Animal AgTech Innovation Summit came during a panel of four investors, including representatives from venture capital firms S2G, Anterra Capital, and Fulcrum Global Capital, and the pharmaceutical giant Merck. They reported a significant decline in agtech venture capital investments, from over $50 billion in 2021 to $15 billion in 2023, with the animal agtech space experiencing the largest decrease. A VP at Merck stated that the era of easy money has ended. The panelists predicted that animal agtech startups would receive $2 billion per annum in five years and that it would take until at least 2030 for another unicorn to emerge in the sector.

In part, this decline reflects the broader slowdown in investment across the tech industry, partially driven by higher interest rates. But it is also due to a mismatch between the typical venture capitalist model, which expects returns within two to three years, and the unique needs of the animal agriculture industry, where successful exits can take a decade or more. Earlier venture capitalists were “tourists” who wanted to add agriculture to their portfolio. They are no longer in the picture. The remaining venture capitalists are more specialized in agtech, but they mostly want to invest in middle-stage companies that have already gained some commercial traction. 

Still, the funding landscape looks bleak, with many investors from the panel recommending that startups consider seeking strategic investments from family offices, grants, or government funds.

Unclear returns 

I spoke with a pork producer who revealed they’re currently losing $40 on each pig they raise and sell. 

He is doing worse than the national average — but not by much. Pork farmers in 2023 lost an average of $31.57 per head. Other producers in attendance told me that different sectors of the animal agriculture industry frequently go through cycles of financial loss, and producers just have to endure it. These periods of loss are further exacerbated by the monopolies held by major meat processors, which make it difficult for farmers to capture profits. 

Farmers operate in a precarious industry with thin (or even negative) profit margins. And while PLF has the potential to save them money, it also requires expensive, finicky infrastructure. All this means that even if PLF can reduce the suffering of farmed animals, farmers won’t be in a position to adopt it unless they can justify the cost. 

Some animals are more equal than others

The conference reached a dramatic peak when protesters interrupted a talk given by USDA acting Deputy Administrator Jeffrey Silverstein. Resisting efforts by organizers to remove them, they yelled into a microphone, “Stop the bird flu bailout!” The protesters emphasized that the USDA has given over $1 billion in taxpayer money as indemnity payments over the past two years to shield the egg and poultry industry from economic losses due to bird flu outbreaks. These ongoing outbreaks have resulted in the deaths of more than 96 million chickens, often through inhumane methods like ventilation shutdown, which can cause the birds to die slowly from suffocation over two to three hours. 

One of the potential advantages of PLF technology is to detect early signs of illness in individual animals before infections like bird flu can spread across the entire flock. Indicators like rising body temperature, vocalization changes, and altered motor activity could help identify a sick bird for quarantine. However, current standard practices assume that once bird flu symptoms appear, the entire flock will be affected, resulting in the mass culling of all chickens. 

However, structural incentives make it less appealing to develop PLF technologies for chickens and other small animals — the vast majority of farmed animals in the world. Cows and pigs are more economically valuable per head than chickens, turkeys, or fish, making it more cost-effective to install expensive systems to monitor the status of individuals. Smaller animals usually can’t carry sensors and are difficult to individually identify via cameras in crowded conditions. And with government indemnities covering these outbreaks, farmers have little financial incentive to adopt proactive technology to avoid losing their flocks to disease.

Agtech can help animals

PLF has a slower, rockier road to adoption than I initially assumed. But most of the major barriers for PLF — like low investment, technical challenges, and thin profit margins — matter the least for the large, technologically sophisticated producers who collectively compose the largest share of animal agriculture in the United States. PLF might not be coming tomorrow, but it is coming.

So where does that leave animals? Most of the people I posed this question to responded that improved welfare would be a natural byproduct of PLF systems. I’m usually skeptical of these kinds of claims, but there was one case study from a seasoned venture capitalist in the animal agtech space that particularly resonated with me.

In 2021, the Red Tractor certification scheme in the UK initiated a new rule. Dairy farms with the Red Tractor quality mark, which applies to 95% of milk produced in the country, could no longer routinely euthanize male calves. Male dairy calves aren’t profitable: They can’t produce milk, they don’t have as much muscle as beef cows, and UK consumers were buying less veal. Prior to this rule, many male dairy calves were killed at just four days old. 

According to my friend the VC, the Red Tractor scheme led to a hockey stick increase in sexed semen sales. This is exactly what it sounds like: a technology that separates out male and female producing semen to decrease the number of male dairy calves being born. This technology has been on the market since the 1990s and now accounts for 75% of all dairy semen sales in the UK. Sexed semen has been credited with making the consistent year-on-year decrease in on-farm euthanasia of dairy calves possible. ³  

This demonstrates the power of technology to improve the lives of animals — and it shows us how actors such as third-party certifiers can use their leverage to drive the adoption of these technologies. Other actors also have leverage. Major food companies and retailers could use their purchasing power to encourage the adoption of animal welfare-enhancing technologies by their suppliers. They could set animal welfare standards that suppliers must meet and give preference to those using PLF solutions that demonstrably improve animal welfare outcomes. Regulatory bodies could require the use of certain animal welfare-enhancing technologies as a condition of licensing or permitting for animal agriculture facilities. Research institutions could test out promising proxy metrics for animal welfare such as emotional feature recognition in pigs or active walking in chickens. Animal welfare advocates could lobby policymakers, educate the public, and engage in dialogue with corporations to drive support from stakeholders. As the PLF industry evolves, many actors will have the opportunity to shape its trajectory. 

Aligning incentives

PLF could also help regulators and certifiers tie farmers’ incentives to the welfare of their animals much more directly. One way to accomplish this is through labeling schemes. Consumers are willing to pay a premium for humane animal products. Unfortunately, labeling schemes meant to designate humane or high-welfare products are often misleading and sometimes downright deceptive. Part of the reason is that these schemes rely on welfare inputs rather than outputs. A standard might demand that birds not be kept in cages or that they have access to a certain amount of outdoor space, but not that the birds are engaging in natural behaviors or actually going outside for meaningful periods of time. Because it is labor intensive to measure outcomes, most farms do not collect or report this data. But PLF could be designed to generate these metrics, and thus change the focus of animal welfare regulation from inputs to outputs. 

The UK Department for Environment, Food and Rural Affairs is currently considering mandating a requirement for domestic and imported animal products to be labeled on a five-tier welfare ranking system. Right now, the law rates welfare using method of production, but this and similar regulations could be written to account for direct welfare outcomes.

Retailers are another opportunity for the PLF industry to serve the cause of animal welfare. Many corporations have made commitments to improving the welfare of the animals in their supply chain. Walmart’s policy requires that each of their fresh pork suppliers “must have on-farm video monitoring for sow farms and will be subject to unannounced animal welfare video audits by an accredited and independent third-party.” If these video feeds were analyzed through PLF systems, reporting about the welfare of the pigs would be more reliable, transparent, and likely less costly than the intermittent auditor visits.

Slaughterhouses present another promising avenue for data collection. Several countries, such as England, Israel, and Spain, have introduced laws mandating the installation of CCTV cameras in slaughterhouses to deter animal welfare violations. In other countries, like the Netherlands, companies have voluntarily agreed to install CCTV and open themselves to audits, driven by collaborative agreements between industry and government.

A notable case study of this collaboration is the Dutch pork producer Vion, which has worked with animal welfare organizations and willingly provided video footage from their slaughterhouses to AI4Animals. Using this footage, the team developed a PLF system capable of detecting various issues, such as pigs lagging behind, which may indicate an injury during transport, or signs of consciousness after stunning and bleeding. When the PLF system identifies these issues, it saves the video footage clips to a dashboard, enabling the company to take corrective actions.

As more producers adopt PLF systems, the government could aggregate data to identify industry-wide trends and refine standards. Policies that properly account for the external cost of animal suffering, such as a mandatory welfare labeling system or a suffering tax, could help create incentives for PLF companies to actually design their systems to deliver benefits for their animals, rather than focusing solely on productivity or specific health metrics. This approach could encourage a cycle of scientifically informed transparency regarding farmed animal welfare.

Still, as with technological revolutions of the past, there’s no reason to expect that the introduction of PLF ends well for animals by default. When drastic technological shifts happen, animals are at the mercy of the tide, swept along by powerful social and technological forces they cannot control. Regulatory agencies, third-party certifiers, retailers, and consumers can each create leverage to align economic incentives with improved animal welfare. The future can still be steered in a direction that values the experience of all sentient beings.