Champions of AI: Singapore's 24 Firms and What SMEs Get

Deputy Prime Minister Gan Kim Yong opened SWITCH 2026 at Marina Bay Sands on Tuesday and announced three things: a S$150 million materials research platform, the first 24 companies in the Champions of AI programme, and a set of workshops for businesses too small to have a technology team. The event has grown from about 5,000 participants in 2016 to more than 25,000 from 100 countries, and AI is its central theme.

The announcements sit under the four sectoral AI missions from Budget 2026: advanced manufacturing, finance, healthcare and connectivity. Our State of AI in Singapore report tracks how that money has moved from pilots into sector programmes.

EnterpriseSG calls Champions of AI its national flagship AI initiative. The first cohort has 24 Singapore-based companies, among them DBS, OCBC, UOB, ST Engineering, Seatrium, PSA International, Singapore Airlines, Singtel Singapore, Keppel, Sembcorp Industries and CapitaLand Investment.

What they get is not cash for a pilot. EnterpriseSG helps each one build a long-term AI transformation roadmap, connects it to AI expertise and solution providers, and supports the unglamorous parts: change management, job redesign and retraining. The firms commit in return to developing their own workforce and to helping smaller suppliers in their ecosystems.

That last part is why this reaches a 12-person business. If your biggest customer is on the list, the way it buys, forecasts and audits will change, and the data it asks you to send will change with it.

EnterpriseSG ran a pilot of hands-on agentic AI workshops that reached more than 280 micro-enterprises and small businesses. Participants picked a use case and built a working prototype using global tools such as ChatGPT, Claude and Manus, in areas like marketing, sales and customer service. More workshops roll out in November, and the programme scales further in 2027.

An agentic tool, if the term is new to you, works through a sequence of steps toward a goal instead of answering one question. EnterpriseSG said its feedback showed many firms are experimenting but lack the in-house people to refine and integrate what they build, which is why it is also looking at support for last-mile deployment.

A Catalytic AI Solutions push begins with advanced manufacturing, trade and logistics. In logistics that means AI to allocate jobs across a fleet and optimise routes, with EnterpriseSG helping transport management system providers get access to new data sources to train on. In wholesale trade the same idea targets rice traders making procurement and inventory decisions. Support measures for that scheme are due by early 2027.

On SMEs Go Digital, more than 100 AI tools have been added across business functions, plus a Pre-approved Solutions Package Builder so a firm can configure a package to its own operations.

MARVEL-AI, short for Materials Accelerator for Research and Venture Enterprise Launchpad, is the S$150 million piece, led by A*STAR with the Ministry of Energy, Trade and Industry, EnterpriseSG, the Economic Development Board and the National Research Foundation behind it. AI models propose candidate materials and design experiments, automated laboratories synthesise and test them, and the results feed back into the models.

Four classes are covered: metal alloys, polymers and organics, ceramics, and composites, with applications in semiconductors, energy and chemicals. The throughput figures are the argument. About 75 polymer samples can be produced in 24 hours, roughly 25 times what researchers manage by hand, and up to 470 metal alloys can be screened in a day, more than 400 times conventional methods. Foxconn Industrial Internet and ST Engineering are named as industry partners.

Almost none of this is something a small firm buys. It matters because materials discovery is where the semiconductor and chemicals clusters compete, and the suppliers and engineers around them get the work.

Gan's line on workers was direct: "People must be part of our AI plans." He said job redesign and training should start alongside process changes rather than after them, so that AI takes routine tasks off people and leaves them work that needs judgement, creativity and expertise.

Read that as a signal for the next appraisal cycle. Companies on the Champions list have committed to role redesign, and in practice that usually shows up as new job scopes rather than fewer people. If your employer is not on the list, the same pressure arrives through a customer who is. The SkillsFuture AI subscription options remain the cheapest way to be early rather than caught out.

EnterpriseSG spent six months selecting this cohort, and the speech gives no public application route. The realistic path in is the SME workshops or your trade association: EnterpriseSG and IMDA are working with associations on support channels for companies with different starting points and will share more early next year. November workshop dates were not published at the time of writing, so watch EnterpriseSG directly rather than third-party pages.

Two details to hold loosely. The Catalytic AI Solutions measures are due early 2027, and the S$150 million is initial funding, which means MARVEL-AI is being built rather than finished. The Singapore AI agencies map helps if you need to know which agency owns the scheme you are chasing. The full text is in the METI speech transcript, and The Straits Times carried a same-day report.