ANN/THE STRAITS TIMES – Local artificial intelligence (AI) start-ups looking to secure investor interest should focus on developing innovative applications that offer real-world solutions, particularly in sectors such as fintech, healthcare and analytics, industry experts said.
Rather than competing solely on cost, start-ups should leverage existing AI models to create tailored solutions that add proprietary value, according to industry players who spoke to The Straits Times.
Partner at venture debt firm Innoven Capital South-east Asia Ben Cheah noted that while cost efficiency remains a factor, investors are more interested in how start-ups build distinct applications on top of basic AI frameworks.
“Investors will likely focus on companies that can develop unique, high-value applications rather than just those optimising for lower development costs,” he said. “It’s not just about affordability but about defensibility and how well a start-up can create proprietary value.”
As AI models become increasingly accessible and affordable, experts suggest that differentiation through specialised applications will be key for local AI firms to stand out in a competitive market.
In January, Chinese AI start-up DeepSeek made waves with its chatbot of the same name, claiming it was built at a fraction of the cost of other chatbots by industry giants such as OpenAI with ChatGPT.
DeepSeek is also an open-source model, which means that its code can be freely downloaded, copied and modified by anyone to develop their own applications.
Cheah noted that AI remains a high-priority investment sector, but with lower-cost models like DeepSeek becoming more popular, investors are looking beyond AI development to how it can be applied in “transformative” ways.
“The most investable start-ups will be those that go beyond generic AI capabilities to build sector-specific solutions, whether in fintech, healthcare or enterprise automation.”
Co-founder and chief executive of generative AI start-up Addlly AI Tina Chopra said that early AI funding focused on foundation models, but investors are now gearing towards AI agents – autonomous systems performing specific tasks with minimal human intervention – as well as enterprise applications.
“Businesses need AI that integrates seamlessly, not just generic models – companies won’t pay for raw AI, but they will pay for solutions that deliver measurable productivity gains or revenue growth,” she said.
“It’s also clear that AI will not be won by the cheapest model but by the solutions that are most usable and scalable.”
Venture capital investment worldwide reached its highest level in seven quarters during the fourth quarter of 2024, driven primarily by growing interest in AI.
The AI industry accounted for the five largest deals of the quarter, led by the blockbuster USD10 billion (SGD13.3 billion) funding round by GIC-backed AI start-up Databricks, according to a January report by consulting firm KPMG.
In Singapore, investment activity dipped slightly in 2024 from a year earlier but remained robust. Deep-tech companies raised USD800 million across 96 deals in 2024, down from USD1 billion and 118 deals in 2023, according to a November report by Enterprise Singapore and PitchBook. These companies develop advanced technologies in areas such as the Internet of Things, quantum computing, and AI.
Chief executive of start-up accelerator Tribe Ng Yi Ming said investors here are focusing on start-ups that demonstrate a clear path to efficiency, scalability and profitability, with the rise of DeepSeek reflecting the growing demand for cost-efficient AI development.
This includes firms developing cost-efficient or “lightweight” AI solutions or specialised applications tailored to industries like predictive analytics and automation, he said.
But cost alone will not determine market success, and start-ups offering niche AI solutions for distinct industry challenges will continue to woo investors, even if their solutions come at a premium, said Ng.
“Practical applications, regulatory compliance, ethical considerations, and ease of integration into existing workflows will be critical success factors. There are also opportunities within the infrastructure and enabling technologies that support AI deployment, such as specialised hardware, data management platforms and cloud services,” he said.
