Singapore SMEs are rapidly taking up AI: according to data from Singapore’s Infocomm Media Development Authority (IMDB) published in their 2025 Singapore Digital Economy (SGDE) Report, SME adoption rates roughly tripled in 2024 to 14.5%, with firms using a variety of solutions, from off off-the-shelf generative AI tools, to domain-specific AI-enabled solutions and customised or proprietary AI tools.
Yet these figures also indicate that around 85% have not yet adopted AI: more than double the number of the 37.5% of non-SMEs that are similarly reticent. Why is this the case, when what AI’s capability and why this technology matters has been made clear?
Michael Goh, CEO of IPI Singapore, comments on these findings in a special interview with SMEhorizon, ahead of the TechInnovation x AIMX 2026 event happening on on 26-27 August.
What are the current trends in AI adoption among Singapore SMEs?
IMDA data shows that AI adoption among SMEs tripled in 2024, from 4.2% to 14.5%. However, 85 percent of SMEs still have not made the move.
So, the trend I would point to is one of two poles. SMEs are not uninformed. Everyone knows what AI can do and why it matters. They are reading the same headlines everyone else is. Those who have clarity on what benefit they can extract from AI have started their journey.
But there are many on the opposite pole, unsure of when they can see meaningful ROI from integrating AI into their processes. That is the gap between awareness and adoption.
Knowing about AI and actually implementing it are two very different things. The question many are grappling with is “where do I start, and how do I know when it will pay off?”
That is actually the right question to ask. It tells us that business leaders are seriously considering AI adoption. But it also means the real work is still ahead for all of us.
Helping businesses move from knowing what to doing is exactly what TechInnovation x AIMX is built around this year.
What are some of the challenges Singapore SMEs face when adopting AI? What are the costs when adoption is unsuccessful or sub-optimal?
I think about this through what I call 4P plus 1R. Purpose, Process, People, Price, and Risk. Each one is a place where adoptions can quietly go wrong.
Purpose is where the challenge lies and it is here where many may lose the plot, starting with the technology and going looking for a problem to solve, rather than the other way around.
Process is the foundation work everyone dreads to do but skip it and AI will lock your inefficiencies in permanently and faster. You have to sort the house and get the foundation ready before you bring in any tool.
People is almost always the last on the list and first to cause a failure. Staff anxiety is real and if you do not address it early, the technology will land and your staff will not be inclined to use it properly.
Price has two sides; the cost of not adopting, a very real efficiency gap that compounds quietly, and the cost of adoption, cost can balloon out of proportion quickly if left unmanaged.
And lastly Risk: considerations about the contingency plans if the system fails, data leakage happens, PDPA and GDPR implications and with agentic AI the unintended consequences of giving systems too much autonomy too early. None of this should stop you from adopting AI. You just need to go in with your eyes open.
Thank you for sharing this helpful framework for successful AI adoption. Could you elaborate on the first step, Purpose?
The 4P plus 1R framework covers the full picture, but let me go deeper on Purpose since that is where everything begins.
The first thing I tell anyone: not everything needs to be AI-enabled. The pressure to adopt everything all at once is real, but automation is not AI, and sometimes a simpler solution (e.g. rule-based automation) does the job better and cheaper.
The question to ask is whether this use case is directly tied to the core of your business, whether it is solving the root cause and whether the return justifies the investment.
When SMEs ask me which AI tool they should adopt, I ask a different question first: where does your day lose time? Not theoretically. Go and look. Watch where things slow down, get re-entered, get missed, or get done twice. The invoice that takes three hours a day to process. The customer quote that arrives two days late. The machine breakdown that was entirely foreseeable. Those are the places where AI, applied carefully, can pay for itself.
That first successful AI deployment matters enormously. Get one thing right, build confidence, show your staff and your stakeholders that this works. Then build from there. SMEs often ask me whether to start small or go straight for something high impact.
My honest answer: you know your business better than I do. But whatever you choose, commit to it properly. A half-hearted pilot teaches you nothing. Do not be afraid that you may not get it right straight away. Very often it is a journey of discovery, keep an open mind but be very focus on what is the outcome you want to achieve. Keep working at it and be ready to take a slight detour to reach the goals you are aiming for.
What does it mean to get the Process right?
Process is the part everyone wants to skip, and it is exactly why so many adoptions stumble.
Before any tool goes live or before you start talking to any vendor, go and talk to the people doing the work. Not to confirm what you already think is happening but to find out what is actually happening. The gap between management’s picture of operations and what happens on the ground is almost always where the real inefficiency lives.
Closing that gap before you deploy any technology is often the highest-return thing you can do, and it costs you nothing but time and honesty.
Are there other aspects crucial to the Process?
Data is the other piece. If your data is sitting in silos, incomplete or inconsistent, your AI will produce outputs that look credible but cannot drive real decisions. That is worse than having no insight at all, because you might act on something misleading. Get the data organised before you expect AI to do anything useful with it. Speak to someone who had deployed AI in their workflow and learn from their experience.
For those moving into agentic AI, systems that can take actions autonomously across your business, build in human checkpoints for decisions that carry real business impact. These tools are powerful. But give them too much autonomy too early and the consequences can ripple across your entire operation before anyone notices. Having a human in the loop is essential for decisions that affect matters that goes beyond your organisation.
You’ve mentioned how People often gets considered last, but is the first thing to cause a failure. What should SMEs be doing?
Bring your people in early. Not when the system is ready to launch. From the moment you are choosing and scoping the use case.
If roles are going to change, and with any meaningful AI adoption some will, tell the people affected before the rumour mill does. Sit with them. Work out together how those roles can be redesigned toward something of higher value.
That conversation is not easy. But leaving people to fill the silence with their own assumptions is much harder to recover from.
And be straight about the learning curve. In the early stages, your team will likely be doing more work, not less, while they get to grips with the new way of working. If you set that expectation honestly upfront, they will push through it. If you oversell the immediate benefits, you will lose them.
Here is what I always come back to. If your best people are spending 40 percent of their day on repetitive, low-judgment tasks, they are not being fully used. AI should absorb that weight so the same people can spend more time on what genuinely needs them: the customer who needs a real conversation, the relationship that needs a human in the room, the problem that needs human judgment.
That is an operational argument, not a technology one. And your staff will get behind it if you bring them along properly.
How can SMEs with limited resources and expertise achieve these goals?
For SMEs with limited resources, keep it simple. Communicate early, involve your people in the process from the start, and help them develop new skills so they are actually getting value out of the tools. Be mentally prepared that some people will not join you on this journey. You will need to have a HR contingency plan for that.
What is your advice for SME owners who may feel daunted by the need to adopt AI?
Start with the problem, not the technology. The SMEs that struggle almost always began by asking “how do I use AI?” The ones that succeed started by asking “what is the one thing that, if I fixed it, would make the biggest difference to my business right now?”
That shift in framing changes everything. AI stops being this overwhelming, expensive unknown and becomes a tool you are evaluating for a very specific job.
And before you go live with anything, ask yourself honestly: if this system fails for 30 consecutive days, can the business keep running? If the answer is no, you are not ready yet. Build the continuity plan first. Operational resilience has to come before operational dependence.
You also do not have to work this out alone. One of the things I am most proud of at IPI is what we have been able to do for companies like Roots Innovation, a Singapore startup that built Synstream, an AI-powered platform for integrating fragmented data and automating enterprise workflows. Through TechInnovation 2025, they were connected with Panasonic R&D Center Singapore.
With IPI’s support, that initial conversation became a full Proof-of-Concept that delivered two to three times faster model training readiness for Panasonic’s AI team. It led to a formalised MOU and a jointly co-developed product announced at SusHi Tech Tokyo earlier this year. A Singapore startup’s technology, validated and scaled by a global MNC. That does not happen by accident.
If you are an SME looking to take that first step, or a technology company looking to find the right enterprise partner, TechInnovation x AIMX on 26 and 27 August at Marina Bay Sands is where those conversations happen. Come with a problem. Come with a solution. Come ready to move.
For those still hesitating, I understand it. Done poorly, AI adoption is disruptive and costly. But the leaders I have seen who got this right are not the ones who moved fastest. They are the ones who asked the hard questions first, cleaned up what needed cleaning, and brought their people along before the technology arrived.
Move with intent, and you will not just have adopted AI. You will have built something worth keeping.
