Note

Malaysian Palm Oil Smallholders Produce Up to 40% Less Oil Per Hectare Than Estates. The AI That Could Close That Gap Is Priced for the Estate.

Smallholder oil palm farmers produce roughly 40% less oil per hectare than large plantation estates, according to peer-reviewed research published this year comparing smallholder and estate productivity. In Malaysia, that gap isn't a rounding error — smallholders and settlers under FELDA, RISDA, and FELCRA (the federal agencies overseeing land settlement, rubber, and land consolidation smallholders respectively) number more than 720,000 people, farming a meaningful share of the country's 6.5-million-hectare national oil palm area cap. Close that 40% gap even partially, across that many hectares, and the number gets large fast. The tools that could close it already exist. Almost none of them are built for a smallholder to use alone.

Why the technology exists and still doesn't reach them

Precision agriculture — satellite and drone imagery, GPS-guided fertiliser application, sensor-driven irrigation, the AI models that turn all of that data into a recommendation — is real and working on Malaysian and regional estates today. Researchers reviewing it this year for smallholder contexts specifically placed it in what they call the "re-innovation quadrant": technically sound, but economically incompatible with how a smallholder actually operates. The hardware, the software licences, the training, and the expert support needed to run it "place them beyond the financial and managerial reach of most smallholders," in the researchers' own words. Interpreting a drone survey into an actual fertiliser decision takes a chain of specialised expertise — data collection, analysis, agronomic judgement — that one farmer working a few hectares has no realistic way to assemble alone.

This is the same shape of problem the oil and gas vendors faced in 1993, just in a different sector: a capability that works, priced and packaged for an organisation with scale, sitting completely out of reach of the individual smaller player who could actually benefit from it.

What the researchers themselves recommend

Here's the part worth paying attention to: the same body of research doesn't conclude smallholders should be left to catch up on their own. It recommends what it calls "landscape-level service delivery models" — precision agriculture bought and operated as a shared service across many smallholder plots at once, rather than something each farmer purchases individually. That's not a policy borrowed from a Malaysian business book. It's agricultural scientists, working the problem from the ground up, arriving at the same structural answer: pool the capability at the level of the organisation that can afford to build it, and make it available to the many who can't, rather than expecting scale economics to somehow work themselves out one smallholder at a time.

The money is already flowing — just not to this

Malaysia's 2026 budget put close to RM2.4 billion toward FELDA, RISDA, and FELCRA combined, and roughly RM120 million more specifically for smallholder welfare across commodities. That's real, substantial support. But the slice aimed at the actual capability gap — mechanisation and automation — was RM20 million, alongside RM50 million for farm access roads. Most of the money is keeping smallholders afloat and connected. Very little of it is buying the shared agronomic intelligence that would let a smallholder's yield start closing in on an estate's.

The clock on this one isn't neutral, either. Malaysia's smallholder-planted oil palm area has been shrinking by more than 3% a year since 2019, as the national land cap, mandatory Malaysian Sustainable Palm Oil (MSPO) certification requirements, and an ageing farming population all tighten around the same group of people. Whatever gets built to close the productivity gap has a shrinking base of smallholders left to build it for.

Why this sector needs the coalition more than most

Compare this to Malaysian SMEs more broadly, where the earlier finding was that money wasn't really the binding constraint — most already had some form of AI in hand and said what they lacked was training and judgement, not capital. Palm oil smallholders are a different case. The economics of precision agriculture genuinely don't work at the scale of one farmer's plot; that's not a confidence gap, it's arithmetic. This is closer to 1993 than to today's typical SME story: without an anchor organisation buying and operating the capability collectively — the way Petronas built vendor capability, and the way these researchers are now explicitly recommending for agriculture — a smallholder mostly cannot get there alone, no matter how motivated or well-financed they are individually.

The question this leaves

FELDA, RISDA, FELCRA, and Sime Darby Plantation all sit exactly where Petronas sat in 1993 — large, capable organisations with hundreds of thousands of smaller farmers underneath them who could genuinely benefit from a shared capability none of them can build alone. The research already says what the model should look like. What's still missing is someone willing to run it — before another year of 3% shrinkage takes the decision out of anyone's hands.

Sources: Yield gap and precision agriculture barriers for smallholders, OCL — Oilseeds and fats, Crops and Lipids; Budget 2026 palm oil and smallholder allocations, New Straits Times; Malaysian Palm Oil Board industry overview.

← All notes