Agriculture: how AI will help Malaysian farmers.
What’s the problem?
Frequent pest-related damages.
Between 20-30% of crop yield is lost to pest-related damages in Malaysia each year,9 with damages for crops such as rice as high as 37%.10
Overreliance on traditional irrigation methods like surface irrigation.
The agricultural sector takes up nearly half (47%) of the national demand for water due to traditional irrigation methods, and this demand is set to dramatically increase by 103% by 2050.11
What are the bottlenecks to AI usage?
Fragmented agriculture ecosystem and ageing farmer population.
Malaysia’s agricultural sector is largely characterised by a vast number of smallholder farmers. Many of these farmers, particularly the older generation, are generally less inclined to adopt new technologies. Furthermore, the bureaucratic process of accessing grants and subsidies can be slow and complicated, discouraging farmers from seeking government assistance.12
Infrastructure Gaps.
Rural connectivity and digital infrastructure are inadequate in many farming areas, impeding data collection and Internet of Things (IoT) use. Malaysia will need better connectivity in rural areas, as over 10% of non-urban households lacked internet access in 2023.13
Limited Access to Agriculture R&D and Technology Transfer.
While significant agricultural research occurs, the mechanisms for effectively transferring these findings and new technologies from labs to the hands of smallholder farmers are often weak. This creates a critical gap between developing innovative solutions and ensuring their practical, affordable deployment on farms.
Maximising the opportunity.
To ensure Malaysian farmers realise the full benefits of AI and to respond to these challenges, the government could consider the following policy actions:
- Simplify the application process for grants and subsidies related to agri-tech adoption, making it more accessible for smallholder farmers. This includes implementing clearer eligibility criteria to allow for shorter approval times.
- Scale up practical, hands-on training programmes with multi-stakeholders specifically designed for farmers and especially older smallholders. These programmes should be culturally sensitive, multilingual, and offer ongoing support.
- Accelerate rural digital infrastructure development to enable the wider use of IoT tools. Crucially, to democratise access to AI benefits, the government should also facilitate the creation of national or regional platforms that can aggregate agricultural data (weather, soil, yield, pest outbreaks) and provide AI-driven insights to farmers, even those without individual smart devices.
AI could help enable precision farming, earlier detection of pests and smarter use of water. AI could help boost the productivity of the agricultural sector in Malaysia by RM 5.5 billion (USD 1.2 billion).
Malaysian researchers are using Google DeepMind’s AI system, AlphaFold, to tackle climate-related crop challenges and boost agricultural resilience.
AlphaFold, an AI system developed by Google DeepMind, is revolutionising how scientists understand proteins. Proteins are the building blocks of life, driving nearly every biological process. Yet their complex 3D shapes have made them notoriously difficult to determine in the lab. AlphaFold has transformed our understanding by predicting these shapes with remarkable accuracy, in minutes instead of years.
In Malaysia, researchers are using AlphaFold to solve urgent biological challenges. Dr. Su Datt Lam, a structural bioinformatician at the National University of Malaysia, applies it to study how mutations in proteins affect diseases, viruses like COVID-19, and in crops like rice. Traditionally, such work would require years of lab-based experiments. With AlphaFold, Dr. Lam can quickly model protein structures and gain insight into their function – saving time, costs, and opening new avenues for discovery.
One of his key areas of focus is agricultural adaptation. As Malaysia faces increasingly extreme weather patterns, Dr. Lam is using AlphaFold to study protein mutations that could make rice crops more resilient to climate stress.
“AlphaFold gave us access to an overwhelming number of high-quality protein structures, far quicker than what experimental methods like X-ray crystallography could offer. It’s completely changed how we study and understand proteins.”
Dr. Su Datt Lam
Structural Bioinformatician,
National University of Malaysia
Online retail: how AI can transform the sector.
What are the bottlenecks to AI usage?
High Implementation Costs.
Despite a high awareness of AI in Malaysia, its adoption among online e-commerce sellers is notably low, with only 15% having integrated AI across at least 80% of their operations—significantly lower than the Southeast Asian average of 24%. Nearly two-thirds (64%) cite costly and time-consuming implementation as the main barrier to greater adoption.14 Small and Medium Enterprises (SMEs), in particular, struggle to justify these substantial upfront investments and perceive the return on investment (ROI) as uncertain or long-term, which severely limits their willingness and capacity to adopt AI.15
Data and Legacy System Integration Complexities.
Many Malaysian retailers face significant difficulties in merging new AI technologies with existing legacy systems,16 which often requires substantial investment in upgrading outdated IT infrastructure and specialised training.
Internal Resistance and Organisational Readiness.
Even when the economic case for AI is understood, Malaysian retailers frequently encounter internal resistance to change and a fundamental lack of organisational readiness for AI adoption.17 Employees may harbour fears of job displacement or find it difficult to adapt to new workflows and skill demands, preferring established, manual processes. Indeed, nearly two-thirds (67%) of Malaysian online sellers concede their employees still prefer tried-and-tested ways of working.18
Insufficient Digital Infrastructure and Technical Expertise.
Beyond the challenges of integrating AI with legacy systems, many Malaysian retailers, especially SMEs, may simply lack the foundational digital infrastructure required to fully leverage AI. This includes inadequate computing power, technical workforce, storage capabilities, and robust internal networks.
Maximising the opportunity.
To ensure online retailers realise this opportunity in full, the government could consider the following policy actions:
- Facilitate public-private partnerships to drive change from both the top down and the bottom up. For leadership, this involves developing targeted programmes for business owners and C-suite executives that focus on AI strategy, change management, and building a compelling business case. For the workforce, where 67% of employees prefer established methods, a peer-to-peer “AI Ambassador” programme could be created.
- Expanding existing digitalisation grants like the Geran Digital PMKS Madani to cover the adoption of proven AI solutions in core retail areas such as inventory management, customer service, and personalised marketing. Secondly, to support retailers lacking foundational infrastructure, the government can explore subsidised subscriptions to pre-approved,
cloud-based AI platforms.
AI could make it easier for online retailers to create new types of advertising, respond to consumer questions faster and export to more countries in different languages. In total, we estimate that AI could help grow online retail in Malaysia by RM 5.9 billion (USD 1.3 billion).
Financial services: how AI can make the sector more efficient.
What are the bottlenecks to AI usage?
Fragmented data.
Many financial institutions operate with siloed IT architectures. Over 80% of Malaysian companies cite data integration challenges as a major obstacle to
AI implementation.20
Regulatory uncertainty.
Strict regulations around data privacy, security, and algorithmic accountability in finance necessarily create a more cautious environment for AI adoption.
Maximising the opportunity.
To ensure the industry realises this opportunity in full, and works through these bottlenecks, the government could consider the following policy actions
- Creating an AI regulatory sandbox, to allow firms to pilot AI-driven services. Establish an environment in which banks, insurers and fintech companies can receive time-limited regulatory waivers to test new workflows.
- Coordinating with the sector to develop new AI focused accreditation and training programmes. Convene a tripartite task-force—regulators, the Malaysian Qualifications Agency, and financial-services employers—to map out the most important skillsets and help develop targeted skills bootcamps.
- Explore the creation of new open banking standards to enable the secure pooling of data. Expand the current DuitNow API specification into a full “Open Finance Malaysia” framework covering deposits, investments, insurance, and Islamic-finance products.