The Week AI Came for Wall Street — and What It Means for Singapore
If last week was about AI infrastructure being built at gigawatt scale, this week was about who it's actually for: regulated industries with a lot of money and very little patience.
Banks. Defence. Pharma. The mood shifted from "can AI do this?" to "can it do this for us, today, inside our compliance perimeter?" — and Big AI delivered.
Here's the rundown, and what it means if you're running a business in Singapore.
1. Anthropic Just Built an Investment Bank for Itself
On May 4, Anthropic announced a $1.5 billion joint venture with Blackstone, Hellman & Friedman, and Goldman Sachs — a new AI-native enterprise services firm aimed squarely at deploying Claude inside the world's most regulated companies.
The very next day, at an invite-only briefing in New York, Anthropic shipped what that JV will sell:
- Claude Opus 4.7 for Financial Services — its strongest model on banking work, leading Vals AI's Finance Agent benchmark with a 64.4% score.
- A library of roughly ten pre-built AI agents for the workflows that consume the most analyst hours: pitchbooks, earnings analysis, credit memos, underwriting, KYC, month-end close, statement audits, insurance claims.
- Full Microsoft 365 integration and a fresh data partnership with Moody's.
Dario Amodei and Jamie Dimon shared a stage for the first time. JPMorgan, Goldman, Citi, AIG, and Visa are already in production on Claude.
A year ago, "AI for banking" meant building it yourself. This week, it shipped in a box.
2. SpaceX Quietly Powered Anthropic's Next Year
On May 6, Anthropic locked in the full computing power of SpaceX's Colossus 1 facility in Memphis — over 220,000 Nvidia GPUs and 300 megawatts of new capacity coming online within a month.
In practice, Anthropic doubled Claude Code's rate limits this week alone, and there's much more capacity behind it.
This is the second time in two weeks an AI lab has effectively bought out a hyperscaler's full output. The compute layer is consolidating fast. For everyone downstream, the practical effect is the same: rate limits go up, prices go down, and the model you're using next quarter will be more capable — and cheaper — than the one you're using today.
3. The Pentagon Lets Eight Vendors Inside
On May 1, the U.S. Department of Defense announced agreements with eight tech companies — SpaceX, OpenAI, Google, Microsoft, Nvidia, AWS, Oracle, and Reflection — to use their AI tools inside its classified networks.
Notably absent: Anthropic, after a recent policy disagreement with the administration.
The headline isn't really the contract value. It's the precedent. If frontier AI is now considered safe enough to run inside U.S. defence networks, the "is AI ready for serious work?" debate is, for most of the rest of us, effectively over.
4. Cohere and Aleph Alpha Merge
Canadian frontier lab Cohere announced a merger with German AI champion Aleph Alpha this week — combining North American and European enterprise AI footprints under one roof.
Frontier AI is starting to look less like a pure research race and more like the early days of cloud: a small set of well-funded labs, each tied to specific compute partners, each chasing a slightly different slice of the enterprise market. That consolidation is good news for buyers — fewer vendors to evaluate, more incentive for each one to win your business.
5. Microsoft Says the World Just Crossed 17.8% AI Adoption
In its annual State of Global AI Diffusion report on May 7, Microsoft pegged AI usage at 17.8% of the world's working-age population in Q1 2026 — up 1.5 percentage points in a single quarter.
That curve is no longer linear. We're in the part of the chart where late-adopter pain starts showing up in margins, not just memes.
Closer to Home: Singapore Becomes "Neutral Ground"
While the U.S. tightened its AI ring around defence and finance, Singapore went the other direction this week — leaning harder than ever into the neutral global AI hub card.
- On May 5, the SuperAI conference announced its first wave of speakers for SuperAI 2026, anchoring a full Singapore AI Week from 8–14 June. Fifty startups will showcase, with ten finalists pitching at Marina Bay Sands for a US$100,000 prize pool powered by OpenAI.
- ATxSummit returns May 20–21 at Capella Singapore — World Bank, NVIDIA, Google, Amazon, OpenAI, and 4,000+ leaders from 50+ countries. Minister Josephine Teo opens.
- AI Engineer Singapore runs May 15–17 with 700+ AI engineers, founders, and builders.
- Quietly in the background, Singtel's RE:AI–Mistral Applied AI Centre of Excellence continues building out, with Singtel itself as the pilot customer testing customer service chatbots, network operations copilots, and AI-assisted tender evaluation.
Three world-class AI events in three weeks. That doesn't happen by accident. It happens because Singapore has spent the last 18 months positioning itself as the place both U.S. and Chinese AI ecosystems can stand on without flinching.
For local businesses, that means something concrete: the people, partners, and tooling you need to deploy AI are about to be, for several weeks, all in the same square mile.
What This Means for Your Business
Three takeaways for SG SMEs from this week:
1. Pre-built AI agents are now a category, not a project
You no longer need to assemble an AI agent from scratch — entire workflows are shipping as products. The question stops being "how do we build this?" and becomes "which vendor's pre-built agent matches our process closest, and where do we customise?"
For an SME, that radically lowers the cost of starting. You don't need a data team. You need a process map.
2. The compute price curve is your friend
Every Anthropic–SpaceX, Google–Anthropic, OpenAI–Oracle deal pushes the unit price of AI inference down. Anything you decided wasn't viable to automate 18 months ago — re-cost it. The economics have moved.
3. Use the next three weeks
ATxSummit, AI Engineer Singapore, and SuperAI's lead-in events. Three world-class AI conferences in Singapore between now and mid-June. Even if you don't attend the main stages, the side events, hackathons, and meetups are where partnerships actually form. If you've been telling yourself you'll figure out AI "later" — later just landed in your calendar.
The Practical Question
Wall Street got a Claude model tuned to its workflows this week. The Pentagon let eight vendors into its classified networks. Singapore got positioned as the neutral ground where everyone wants to set up shop.
For the rest of us — the smaller businesses watching the headlines — the question hasn't really changed:
Which decision in your business is currently waiting on a person, when it could be waiting on a model?
That's still where the gain is. The infrastructure is just, finally, overwhelmingly catching up to make it easy.
At The Empyrean, we help Singapore businesses cut through the AI headlines and find the specific, repeatable tasks where it actually pays off — without ripping out the systems you already have. If you'd like an honest read on where AI could fit in your operations, we'd be happy to take a look.