Enterprise Compute Initiative (ECI): a 2026 guide for Singapore companies

In this guide

• The ECI at a glance
• What it funds
• Providers and partners
• Eligibility
• Applying, and whether it is open
• What it does not cover
• ECI vs EDGE vs the 400% deduction
• What to prepare
• Common mistakes

What is the Enterprise Compute Initiative?

A short-term programme that helps Singapore-based companies turn an AI use case into a minimum viable product (MVP) and build the team to keep it going. Prime Minister and Finance Minister Lawrence Wong announced it in his Budget statement on 18 February 2025, as EDB reported. Digital Industry Singapore, a joint office of EDB, Enterprise Singapore and IMDA, administers it, towards the National AI Strategy 2.0 goal of "anchoring AI teams across Singapore". It grew out of a Google Cloud pilot with 30 companies from October 2024. Note that it is a Budget 2025 measure: the 400% AI tax deduction came in Budget 2026 and works differently (comparison below).

Scroll the table sideways to see every column.

What does the ECI fund?

Two things, from two payers.

• Cloud credits, AI tools and training from the provider you choose: the provider's own offer, not a government grant.
• Consultancy from a pre-qualified partner for AI strategy, technical implementation and change management. DISG supports 70% of actual costs on a supportable total capped at S$150,000, so at most S$105,000 per company, and the work "must go beyond strategy consulting and focus on technical implementations".

The grant goes to the partner, not to you. According to DISG's ECI factsheet (updated 10 February 2026), you file no grant forms or claims; the partner does, and your 30% is settled under your own agreement with it.

A worked example

A partner quotes S$120,000: DISG supports S$84,000 and you pay S$36,000. At S$150,000 the split is S$105,000 and S$45,000, DISG's maximum. At S$200,000, on our reading of the cap, you pay S$45,000 plus the S$50,000 above it: S$95,000. Add your own team's time, which DISG expects you to commit for the whole programme.

Which cloud providers and partners take part?

Four, each with its own offer and its own list of pre-qualified consulting partners. Figures as DISG's provider page lists them, checked 16 October 2026; they are maximums, and some are subject to approval.

Scroll the table sideways to see every column.

Each provider's programme runs for one year from its own launch. Confirmed launches: Google Cloud's pilot in October 2024; Microsoft's on 1 August 2025, the third under the ECI according to the Ministry of Trade and Industry; Oracle's on 26 January 2026, per Oracle's announcement. DISG does not give AWS's date. The provider you pick is final, it need not be one you already use, and the partner comes from that provider's list: you cannot bring your own.

Who is eligible for the Enterprise Compute Initiative?

Any sector except gambling and betting, if the company meets every baseline criterion:

• registered or incorporated, and physically present, in Singapore;
• sponsorship at CEO or equivalent level;
• at least 10 staff based in Singapore;
• a tech team of at least two, such as software engineers, AI engineers or data scientists;
• experience building a custom AI solution, for a proof of concept or wider deployment;
• accessible data sets for the use case;
• the financial ability to finish the project, such as stakeholders' equity.

The use case must target process improvement, new products, R&D or the wider ecosystem, show impact, and lead to an AI team inside the company; complex, scalable cases are prioritised. Contractors do not count towards the minimums, a group enters once through its parent, and an earlier AI grant is fine if this use case is different. In practice the ECI is for companies that already employ engineers: a ten-person trading firm with no developers does not qualify, and EDGE or the AI deduction is the likelier route.

How do you apply, and is the ECI still open?

1. Apply through DISG's online form with Corppass and your latest audited financial statements, one main use case and your choice of provider.
2. Wait for the review: generally up to four weeks.
3. Onboard: your provider onboards you and assigns a consulting partner.
4. Build the roadmap and MVP over up to six months, tested by at least five end users who give written feedback.

Is it still open? DISG has not published a closing date. Its pages, last updated on 17 July 2026, still carry the application link, but with each track running a year from its launch, some may be at or past that mark, so ask DISG (ECI_DISG@edb.gov.sg) which providers are taking new cohorts. A rejected applicant may reapply with a meaningfully revised proposal.

What does the ECI not cover?

• Companies below the thresholds, or in gambling and betting.
• Strategy-only consulting, or a consultant from outside the provider's list.
• Your 30% of consulting, and anything above the S$150,000 supportable total.
• Life after the MVP: further support is a separate conversation with DISG.

How unused credits are treated, and who pays to run the MVP once they are spent, is not yet published. Ask your provider before you build on its credits.

How does the ECI compare with the EDGE Grant and the 400% AI deduction?

They are different instruments: the ECI subsidises a first AI build with a cloud provider, EDGE reimburses part of a vendor project, and the deduction cuts tax on AI spend you pay yourself.

Scroll the table sideways to see every column.

Sources: Enterprise Singapore and IRAS, checked 16 October 2026. IRAS excludes "any expenditure subsidised by a grant or subsidy from the Government" from qualifying AI expenditure, so a cost the ECI or EDGE paid for does not also earn the deduction; ask your tax adviser about the share you paid. More in the 400% AI tax deduction and what replaced PSG, EDG and MRA.

What should you prepare before you approach a partner?

A one-page use case, written before anyone quotes.

• The workflow. One process as it runs today and what it costs in hours, errors or missed enquiries. "Enquiries in three languages wait until morning" is a use case; "adopt generative AI" is not.
• The data. Where it lives, whether your team can reach it, and whether it includes personal data under the PDPA (see PDPA and AI agents).
• The success measure, and the five or more end users who will test the MVP.
• The owners: the CEO-level sponsor and the in-house engineers who will run it afterwards.
• The approval boundary: what the AI may do alone and what waits for a person, as IMDA's voluntary agentic AI framework recommends.

Common mistakes

• Treating a deduction like a grant. The 400% deduction pays no cash and is worth nothing without taxable income.
• Assuming eligibility. Only direct Singapore employees count, and prior custom AI work is a baseline criterion.
• Buying compute before you have a workflow. Credits are the easy part; the programme is judged on an MVP real users tested.
• Choosing a provider on the headline number. The figures are maximums and the choice is final.
• Forgetting the running cost of the MVP after the programme.

How Olano helps

Olano is not an ECI cloud provider or pre-qualified partner, and not a pre-approved vendor for any grant; we do not apply on a client's behalf and never promise a funding outcome. We help on either side of the funding question: scoping a workflow worth funding into the one-page use case above, which you can take to DISG, a partner or your adviser, and, where the ECI is not your route, building and running the agent system as a fixed-proposal studio engagement. Whether that work qualifies under EDGE or for the deduction is for Enterprise Singapore and IRAS to decide. See our services and grant funding pages.