The Week OpenAI Picked Singapore
If you only read one tech story this week, here's the line that sums it up:
“<em>OpenAI committed S$300 million to Singapore and opened its first Applied AI Lab outside the United States right here — with 200+ Forward-Deployed Engineer roles attached. The same week, it launched a US$4 billion Deployment Company to embed AI engineers inside large enterprises globally. And EY and Microsoft pledged US$1 billion over five years for enterprise AI services.</em>”
Last week the story was AI agents showing up as employees. This week the story is the infrastructure being built to put them there — and Singapore is now sitting inside that infrastructure, not outside it.
Here's what happened, why it matters, and what your business should do about it.
The Big Three
1. OpenAI for Singapore: S$300M, an Applied AI Lab, and 200+ Engineers
On May 20 at the ATx Summit, OpenAI and Singapore's Ministry of Digital Development and Information (MDDI) signed an MoU committing more than S$300 million to Singapore's AI ecosystem. The headline pieces:
- OpenAI Singapore Applied AI Lab — the company's first Applied AI Lab outside the United States, hosting more than 200 technical roles over the next few years and making Singapore a global hub for OpenAI's Forward-Deployed Engineers.
- Sector focus aligned to Singapore's AI Mission priorities: public service, finance, healthcare, and digital infrastructure — exactly the verticals that anchor most of the SME economy around them.
- OpenAI Academy Singapore chapter, a Forward-Deployed Engineer training programme to build local AI deployment talent, and Codex for Teachers hackathons with MOE and GovTech, including Mother Tongue language tooling.
The framing matters. OpenAI didn't open a sales office. It opened the part of the company that takes frontier research and turns it into running production systems inside customer organisations. The same week, Google also inked AI deals at ATx, and President Tharman Shanmugaratnam hosted the summit's opening — signalling government-level alignment, not just a vendor visit.
2. OpenAI Deployment Company: A US$4B Bet on "AI as a Service"
Days earlier (May 11), OpenAI launched the OpenAI Deployment Company — a majority-owned subsidiary backed by more than US$4 billion in initial capital, with TPG, Advent, Bain Capital, Brookfield as co-leads and Bain & Company, Capgemini and McKinsey as integration partners. To kickstart it, OpenAI also acquired Tomoro, bringing roughly 150 Forward-Deployed Engineers in from day one.
The pitch: stop selling enterprises a model and start selling them an engineering team that redesigns workflows around the model. Forward-Deployed Engineers sit inside the client's offices, working alongside executives and frontline staff to ship production-grade AI systems — not pilots.
The signal: the frontier labs have read the same data the rest of us have. Most enterprise AI initiatives stall. (HCLTech's new survey of 467 senior executives this week put the expected failure rate at 43%.) The fix the labs are reaching for isn't a better model — it's people on the ground.
Now connect the dots: OpenAI's first overseas hub for those same Forward-Deployed Engineers is in Singapore.
3. EY ↔ Microsoft: Another US$1 Billion for Enterprise AI
On May 22, EY and Microsoft expanded their alliance with a US$1 billion, five-year commitment to help organisations scale AI across audit, tax, consulting and strategy. It's the same playbook as the KPMG–Anthropic deal from a week earlier: a Big Four firm standardising on a frontier model and pushing it through every client engagement it runs.
Separately, Salesforce rolled out Agentforce Coworker in beta on May 22 — an embedded AI teammate that retrieves CRM context and takes actions inside the tools sales and service teams already use. And Anthropic quietly shipped self-hosted sandboxes and MCP tunnels for Claude Managed Agents on May 19, closing the governance gap for companies that want agents but can't ship data outside their own perimeter.
The through-line of the week: the delivery layer for enterprise AI is finally being built, in public, at billion-dollar scale.
Closer to Home: What S$300M Actually Buys Singapore
A few things become true the moment OpenAI's lab opens here that weren't true before:
- Frontier deployment expertise is now local. The engineers who get OpenAI's hardest customer problems will be based in Singapore. That talent leaks — into local startups, into SI partners, into the SMEs they end up advising. It's the same dynamic that made Singapore a regional fintech hub a decade ago.
- The talent pipeline gets pulled forward. The Forward-Deployed Engineer training programme, the OpenAI Academy Singapore chapter, and Codex for Teachers all aim at the same thing: making sure there are people who can wire AI into local operations, not just tools. For an SME hiring its first AI-literate junior, the candidate pool just got materially better.
- The DEB stack is filling out. Stack what's now on the table for Singapore SMEs in a single quarter: the Digital Enterprise Blueprint (50,000 SMEs by 2029), the Grab AI Programme (10,000 F&B/retail SMEs via GrabAcademy + SUTD), DBS Spark GenAI (Start/Accelerate/Scale tiers, 380+ companies signed up, up to 50% grant cofunding via IMDA/EnterpriseSG), IMDA's pre-approved AI solutions catalogue, the SME AI Impact Awards 2026 (nominations 1 Jun–14 Aug), and now OpenAI for Singapore.
- The baseline keeps moving. The latest Singapore Digital Economy Report has enterprise AI adoption at 23.5% in 2025, up from 4.3% in 2023 — a 5x jump in two years. "Average" isn't where it was 18 months ago, which means waiting another year to start costs more than it used to.
The pattern across the week is consistent. Globally: frontier labs are investing in human delivery capacity, not just model capacity. Locally: Singapore has positioned itself to host that capacity rather than import it.
What This Means for Your Business
Three takeaways for SMEs in Singapore this week:
1. The "who deploys it" question now has answers
For the last two years the biggest blocker for SMEs hasn't been "is there a model that can do this?" — it's been "who's going to install it, train my staff, and keep it running on Monday?" Between OpenAI's Singapore Applied AI Lab, the consulting partners standing up Forward-Deployed Engineer teams, DBS Spark's tiered programme and IMDA's pre-approved solutions, that delivery layer is finally filling out for the local market. The next 12 months are when the supply of practical AI implementers stops being the bottleneck.
2. Treat "AI training" as a Q3 calendar item, not a vague intention
With the Grab AI Programme running through GrabAcademy, the OpenAI Academy Singapore chapter coming online, DBS Spark workshops, and the SME AI Impact Awards nomination window open from 1 June to 14 August, the next eight weeks are unusually rich. Pick one of your staff (ideally not the founder) and book them into one programme. Then look at one of your business processes and put them in charge of mapping where AI could touch it. That's the minimum viable move.
3. Pick the workflow before you pick the tool
The most useful idea inside the Deployment Company / Forward-Deployed Engineer playbook is structural: start with the workflow, not the model. Big firms are paying premium consulting fees for engineers to come in and ask, "what does this process actually look like end to end, and where does AI plug in without breaking the rest?" You don't need to pay McKinsey rates to do the same exercise at SME scale. Pick one repeatable, painful workflow — supplier reconciliation, after-hours customer enquiries, monthly reports, follow-up emails — write it down step by step, and then go shopping for the tool. The order matters.
The Practical Question
The headlines this week were S$300 million, US$4 billion, US$1 billion. The relevant question for most Singapore SMEs is smaller but harder:
"If the people who can actually deploy AI are now in our time zone, what's the one workflow I'd hand them first?"
The order book that lives in someone's head. The quotes that get re-typed into three systems. The customer onboarding form that always loses a field. The end-of-month numbers no one trusts until they've been checked twice.
This week's news didn't change what's possible. It changed who's available to build it, and where they're sitting. For Singapore SMEs, both of those answers got a lot more local.
At The Empyrean, we work with Singapore SMEs to find the practical, repeatable tasks where AI delivers value without disruption — and to set them up with the kind of governance that earns customer trust. If you're not sure where to start, we're happy to take a look at your operations and tell you honestly what would make sense.