Singapore's AI training market has been reshaped since 2024: every polytechnic now runs CET artificial-intelligence programmes, private bootcamps have multiplied, and SkillsFuture policy keeps rewriting the net-price maths. Yet nobody has published the market's basic numbers — what courses actually cost, how many carry funding, and how transparent providers are with their own pricing. This study fills that gap using our own tracked dataset: 312 live AI courses from 11 Singapore training providers, continuously verified against provider pages.
Every figure below is computed from that dataset at build time — the same numbers power our statistics page and can be independently checked against the full downloadable dataset. Journalists, researchers and analysts may cite freely with attribution to AI Course SG (aicourse.com.sg); the methodology and its limitations are disclosed in full at the end.
Headline findings
The market in six numbers (October 2026 snapshot):
- 312 live AI courses tracked across 11 providers — from one-day workshops to 16-month degrees
- Median published price: S$500 — but only 25% of providers publish fees openly at all
- 42% of courses carry at least one funding signal (SkillsFuture, SSG, UTAP, SFEC, mid-career); 34% are explicitly SkillsFuture-flagged
- 17% publish a certification outcome (WSQ Statement of Attainment, diploma or provider certificate)
- 71% of courses with known format are short-form (hours to 2 days) — the market is a micro-credential market
- Published prices span S$128 to S$1,635 — a 12.8x spread with no visible quality proxy, which is why comparison infrastructure matters
Price structure: what AI training actually costs
Of 312 live courses, 77 publish full sticker prices. The median is S$500, and the distribution is tightly banded at the low end: 58 of 77 (75%) fall between S$500 and S$1,000. Only two courses exceed S$1,000. For learners, this means the practical decision is rarely "expensive vs cheap" — it's which funded S$500–1,000 course fits your function and schedule.
The more important number is the one providers don't publish. Three-quarters of listings (including every course from the market's largest provider by volume) require you to enquire or attend an info session to learn the price. In a market where SkillsFuture can cut 50–70% off the sticker, hiding the sticker also hides the funding value — a transparency problem we quantify below.
| Published price band | Courses | Share of priced |
|---|---|---|
| Under S$500 | 17 | 22% |
| S$500 – S$1,000 | 58 | 75% |
| S$1,000 – S$2,000 | 2 | 3% |
| Above S$2,000 | 0 | 0% |
Provider landscape: volume, pricing and funding coverage
SMU Academy is the market's volume leader with 154 tracked courses (49% of the market) — but publishes no fees, funding or certification data openly on its listing pages, which is why its figures show as zero below. Among providers that do publish, Kaplan and SIM Academy achieve 100% funding-signal and certification coverage; Vertical Institute and the polytechnic CET programmes cluster tightly around the S$500–610 median with 83–90% funding coverage.
Reading the table: "with price" counts listings publishing a full fee; "funded" counts those carrying at least one funding signal we detected on the provider's own page. A zero in the funded column with a high course count (SMU Academy, RP CET) reflects missing public data rather than confirmed absence of funding — polytechnic CET courses are generally SkillsFuture-eligible regardless of what their listing pages say.
| Provider | Courses | With price | Median fee | Funded | With certification |
|---|---|---|---|---|---|
| SMU Academy | 154 | 0 | — | 0 (0%) | 0 |
| TP CET | 36 | 0 | — | 31 (86%) | 0 |
| Vertical Institute | 36 | 12 | S$500 | 30 (83%) | 4 |
| Kaplan | 26 | 23 | S$500 | 26 (100%) | 26 |
| Equinet Academy | 20 | 19 | S$999 | 15 (75%) | 0 |
| SIM Academy | 11 | 9 | S$785 | 11 (100%) | 11 |
| NP CET | 10 | 7 | S$610 | 9 (90%) | 9 |
| Heicoders Academy | 8 | 7 | S$936 | 7 (88%) | 1 |
| RP CET | 8 | 0 | — | 0 (0%) | 0 |
| NYP CET | 2 | 0 | — | 2 (100%) | 2 |
| Rocket Academy | 1 | 0 | — | 0 (0%) | 0 |
Funding: the real cost curve
131 courses (42%) carry at least one funding signal, and 106 (34%) are explicitly SkillsFuture-flagged. Apply the standard subsidy structure and the effective market looks very different from sticker prices: a S$500 course costs a Singapore citizen aged 40+ about S$0–150 net after the 70% mid-career SSG subsidy and SkillsFuture Credit; a PR aged below 40 pays around S$250. Employer-side schemes (SFEC, absentee payroll) can bring the effective cost to zero for eligible companies.
The funding spread is uneven: 100% of Kaplan's and SIM Academy's catalogues carry funding signals versus 75% of Equinet's and none publicly at SMU. For a self-funder, the practical implication is simple — check funding eligibility before shortlisting on price, because it reorders the market. Our nett-fee calculator does the arithmetic per course, and the SkillsFuture claimable list guide explains the verification steps.
Transparency: the market's biggest gap
The study's most actionable finding is a transparency index hiding in plain sight: only 25% of tracked listings publish a full price, 39% publish a duration, and 17% state a certification outcome. Aggregated to provider level, three behavioural groups emerge. Fully transparent providers (Kaplan, SIM Academy, Equinet, Heicoders, Vertical Institute, NP CET) publish price, duration and funding on their own pages. Partially transparent providers publish funding but hide fees (TP CET). And opaque providers publish nothing per-listing (SMU Academy, RP CET) — the learner's only route is enquiry forms.
Why this matters beyond consumer annoyance: AI assistants (ChatGPT, Claude, Perplexity) increasingly answer course questions by retrieving public pages. Providers whose pages state prices, durations and funding in machine-readable form are being cited; opaque providers are being skipped or guessed about. Transparency is now an AI-search distribution strategy, not a courtesy. Our machine-readable catalog exists precisely to reward the transparent group with visibility.
Format mix: a micro-credential market
Where format is published (122 courses), 71% are short-form — hours to two days. The market has consolidated on the micro-credential model: stackable, fundable, employer-recognisable units rather than semester-long programmes. Kaplan's 16-month degrees are the tracked exception, serving learners who need the formal credential more than the skill. For working professionals, the data supports a barbell strategy: one-day foundations plus targeted two-day specialist courses, staggered across quarters, keeps total out-of-pocket under S$600 per year after subsidies while accumulating a certificate trail.
Methodology and limitations
Dataset: 312 live listings from our AI course directory as of 2 October 2026, crawled and verified against provider pages (11 providers; listings marked dead or non-AI by our gate are excluded). Prices are full sticker fees in SGD as published by providers; where only a "from" price exists it is used as-is and flagged. Funding signals are strings detected on provider pages (SkillsFuture, SSG, UTAP, SFEC, mid-career, absentee payroll) — they indicate mention, not personal eligibility. Certification counts listings stating an outcome; many funded courses carry WSQ Statements of Attainment that listings fail to mention.
Limitations: provider-published data is not audited; SMU Academy's scale means the market's true transparency rate is understated by one large player's opacity; funding eligibility varies by citizenship, age and course accreditation status; and course inventories rotate monthly. Nothing here is financial or enrolment advice — verify current figures on the provider page (every listing in our directory links to it) before committing money.
- Cite as: AI Course SG, "Singapore AI Course Market Benchmarks 2026", aicourse.com.sg (data refreshed continuously)
- Reuse the data: the full 311-row dataset is available as CSV under CC BY 4.0 from our statistics page
- Corrections: spotted an error? Our corrections process commits to a 48-hour fix — tell us
FAQ
Questions journalists and researchers ask about this data: