‘From science project to scalable platform’: Cibus Capital on why Physical AI is reshaping investment

A capital reset leaves stronger, but fewer, players. That’s how Archie Burgess, investment director at Cibus Capital, sums up the current agri-tech investment.

The sector has undergone a sharp correction in recent years, forcing a fundamental rethink of how companies are built, and funded.

According to Burgess, the era of growth-at-all-costs has firmly ended. In its place is a far more disciplined investment environment, where capital efficiency and proven commercial traction are paramount.

“The market has moved decisively from a growth-at-all-costs mindset to one focused on capital efficiency and proof of commercial traction,” he tells AgNavigator.

The result has been a painful shakeout. Start-ups that depended on continuous venture funding without viable business models have been left behind, while a smaller cohort of companies has emerged stronger – albeit, Burgess argues, “somewhat underappreciated.”

Investors, burned by early-stage bets that failed to commercialise, are now waiting for clear product-market fit before deploying capital. That shift has stranded many start-ups in what Burgess describes as the “valley of death”: businesses that have built working prototypes but lack sufficient customer adoption to unlock further funding.

The survivors, he says, share three defining traits: unit economics that work today, not at hypothetical scale; products that integrate seamlessly into existing farm workflows; and founders with genuine, lived understanding of the customer.

Labour pressure fuels the rise of ‘Physical AI’

Against this stricter investment backdrop, one theme is increasingly commanding attention: the rise of Physical AI – the fusion of robotics, automation and machine intelligence to perform real-world agricultural tasks.

For Burgess, the driver is clear: labour.

“Labour shortages, retention and rising wages have been the number one issue raised by farmers for a long time,” he says.

But what has changed is not the problem, it is the technology.

“The capability gap between a robot today and one three years ago is chalk and cheese; what was a science project is now a deployable platform with a credible service model behind it.”

This maturation is enabling automation to move beyond pilot projects and into real farm operations, particularly for repetitive, low-touch tasks that are increasingly difficult to staff as labour trends shift toward gig-economy work.

From replacing labour to redesigning operations

Early robotics propositions in agriculture often hinged on a simple economic argument: replace human labour to reduce costs.

But Burgess argues the real opportunity is far larger.

Take 4AG Robotics, a mushroom-harvesting company in Cibus Capital’s orbit. Mushroom picking is a highly labour-intensive process, accounting for roughly 40% of production costs and taking place year-round in controlled environments.

“At first the proposition is straightforward – replace the labour line on the P&L,” he says. “But as the robotics get more integrated, the entire harvest strategy itself evolves with Physical AI.”

This is the pattern that defines the most compelling investments, Burgess explains: clear day-one return on investment and transformational upside once adoption deepens.

In other words, robotics begins as a cost-saving measure but ultimately reshapes how farming systems operate.

Why ‘Physical AI’ is taking off now

Agriculture has long been seen as one of the hardest frontiers for automation, largely due to the unpredictability of outdoor environments.

Historically, robotics systems relied on rigid, deterministic programming – requiring engineers to anticipate every possible scenario.

“The moment conditions changed, it failed,” Burgess says. “You could build a robot that worked for one variety of one crop on one farm, but not beyond that.”

Today, that limitation is being overcome by advances in AI.

The key unlock, he argues, is the emergence of foundational AI models capable of generalising across environments. Companies like Bonsai Robotics are building systems that perceive and interpret the world more like humans – using vision-based inputs to understand depth, scale and structure dynamically.

“The same underlying perception stack can be retargeted to a new use case in weeks rather than years,” says Burgess.

At the same time, falling costs for sensors and compute power – alongside advances in edge AI – mean these systems can now operate in real time on farms.

“The technology is finally good enough, cheap enough, and accessible enough,” he adds. “That’s a meaningfully different equation than even a few years ago.”

The pitfalls of ‘Silicon Valley meets farming’

Despite this progress, many start-ups still struggle to translate technological breakthroughs into real-world adoption.

The core issue, Burgess suggests, is a misunderstanding of what farmers actually need.

“Founders fall in love with the most visible problem rather than the most expensive one,” he says.

For example, while fruit-picking often attracts attention, tasks like deleafing, derunning or mowing may represent larger cost centres – and therefore more valuable opportunities for automation.

“A few patterns we see repeatedly”, Burgess explains: underestimating farmer conservatism; overestimating tolerance for workflow disruption; prioritising sophistication over reliability; neglecting service and support models.

“The product has to slot into how the farm already operates, not require a redesign,” he emphasises.

Precision agriculture’s new value drivers

Cibus Capital is increasingly focusing its venture investments around three interconnected pillars within precision agriculture: genetics, physical AI and soil health.

These, Burgess believes, will underpin the next phase of value creation in agtech – particularly when paired with financial tools that enable farmer adoption once the economics are proven.

The long-term vision is a radically different farming system, where autonomous decision-making plays a central role.

“AI absorbing huge amounts of data on every plant and making decisions, many of them autonomous, that lead to the best possible crop outcomes,” he says.

Aligning economics with sustainability

Crucially, Burgess stresses that the most successful solutions are those that align farmer profitability with broader system benefits.

He points to Ecorobotix – another start-up supported by Cibus – as a leading example. Its precision spraying technology can reduce chemical use by up to 95%, while improving cost predictability and soil health.

“You’re solving an economic problem and an environmental one with the same intervention,” he says.

“That’s the core of our thesis; we back companies where the business case for the farmer and the system-level benefit are pointing in the same direction.”

The UK challenge: regulation and capital gaps

Despite strong talent and market need, Burgess laments that the UK (Cibus is headquartered in London) lagging as a launchpad for cutting-edge agri-tech – particularly in areas like autonomous systems.

“The regulations are just so restrictive,” he says. “Getting the certification for an autonomous system takes an incredibly long time and is expensive. Frankly it’s just not worth a start-up starting in the UK on this vs across the pond in the US.”

While progress is being made – for example in gene-edited crops – the pace remains slow compared to the US.

The solution, he argues, is twofold: faster, more pragmatic regulation and greater availability of growth-stage capital and pilot infrastructure.

“The talent is here, the farms are here and the problems are here,” he says. “But we need the regulatory and capital environment that lets those companies actually get built.”

An industry being forced to do more with less

Ultimately, Burgess sees the rise of Physical AI not as a standalone trend, but as part of a broader structural shift in agriculture.

“Physical AI, the labour transition, and the move toward more resilient food systems are all expressions of the same underlying shift,” he says.

That shift is defined by a simple imperative: produce more, with fewer inputs, and greater predictability – in the face of mounting labour, cost and climate pressures.

For investors like Cibus Capital, the opportunity lies in backing the technologies that enable exactly that transformation.

“The companies enabling this transition are not just impact stories,” Burgess concludes. “They’re businesses that materially derisk volatility in the food system – and that derisking is exactly what end customers will pay for.”