Skip to main content
AI in Asia
3 Before 9: April 17, 2026

3 Before 9: April 17, 2026

3 must-know AI stories before your 9am coffee. The signals that matter, delivered daily.

· Updated Apr 17, 2026 4 min read
AI Snapshot

The TL;DR: what matters, fast.

STT GDC and SuperX opened an AI Innovation Centre in Singapore, offering free two-week trials of Nvidia Blackwell GPUs for regional enterprises to develop AI solutions.

SoftBank, Sony, Honda, and NEC formed a new company with government backing to develop a sovereign Japanese AI model, targeting a one-trillion-parameter LLM.

Huawei Cloud officially launched its Model-as-a-Service offering across Asia Pacific at Jakarta's AI Boost Day, giving enterprises pay-by-the-token access to GLM, DeepSeek and Qwen models.

Who should pay attention: CIOs | CTOs | AI Developers | Policymakers | Regulators

What changes next: Debate is likely to intensify as countries refine their sovereign AI strategies and regulatory frameworks.

1. STT GDC and SuperX Open Singapore AI Innovation Centre with Free Two-Week Trials

ST Telemedia Global Data Centres and Nasdaq-listed SuperX AI Technology launched a dedicated AI Innovation Centre on Wednesday at STT's Singapore 5 facility in Tai Seng, giving enterprises free two-week access to Nvidia Blackwell GPUs for pilots, proofs of concept and model benchmarking. The centre pairs STT GDC's AI-ready infrastructure with SuperX's orchestration software and is aimed at regional businesses with data residency requirements, independent software vendors and companies still stuck in the AI pilot stage. Trials can be extended by a week on request, and tenants can graduate into full production, hybrid or private deployments inside STT GDC's global footprint.

Why it matters: Singapore wants to be the place where Southeast Asian enterprises stop running AI experiments on foreign clouds and start running them on sovereign infrastructure, and this centre is the most concrete answer yet to PwC's finding that only 20 percent of companies capture the majority of AI's value. For CIOs and CTOs across ASEAN who are under board pressure to prove AI return on investment, a no-cost slot on Blackwell silicon inside a compliant Singapore facility removes the usual procurement and data-transfer excuses for staying in pilot purgatory.

Read more: https://technode.global/2026/04/16/st-telemedia-global-data-centers-and-superx-debut-ai-innovation-center-in-singapore/^

2. SoftBank, Sony, Honda and NEC Launch Joint Firm to Build a Sovereign Japanese AI Model

SoftBank, Sony, Honda and NEC have formally established a new company, Japan AI Foundation Model Development, with each taking a stake of more than 10 percent and backed by up to one trillion yen of government support over five years from fiscal 2026. The venture is targeting a roughly one-trillion-parameter model trained on Japanese data and tuned for real-world physical control tasks, with Honda slated to deploy it first in autonomous vehicles and Sony applying it across robotics and gaming hardware. Nippon Steel, Kobe Steel and several major Japanese banks are in talks to come in as minority investors, signalling ambitions well beyond the tech sector.

Why it matters: Japan is pointedly refusing to rent its industrial AI stack from Silicon Valley or Shenzhen, and the consortium structure means the country's biggest factories, carmakers and financiers will share one domestic foundation model rather than each buying in from abroad. For suppliers, robotics firms and enterprise software vendors across the region, this creates a new Japanese-controlled distribution channel for physical AI and adds a third serious pole to the US-China race that Asian buyers now have to plan around.

Read more: https://asia.nikkei.com/business/technology/artificial-intelligence/japan-s-softbank-launches-unit-to-develop-homegrown-ai^

3. Huawei Cloud Rolls Out Token-Based Model Service Across Asia Pacific

Huawei Cloud used its AI Boost Day event in Jakarta on Tuesday to officially launch its Model-as-a-Service offering across the Asia Pacific region, letting enterprises pay by the token for access to six foundation models spanning the GLM, DeepSeek and Qwen families. The service runs on Huawei's in-house acceleration engine and is being positioned as a plug-and-play way for Southeast Asian firms to build agentic AI workloads without standing up their own GPU clusters. Huawei used the Jakarta stage to showcase customer deployments in banking, logistics and public services as proof that Chinese models are now enterprise-ready outside China.

Why it matters: Huawei is extending the Chinese AI stack into Southeast Asia on a pay-as-you-go basis, giving mid-market buyers in Indonesia, Thailand, Malaysia and the Philippines a lower-cost alternative to OpenAI, Anthropic and Google at a moment when token pricing is becoming the primary axis of competition. Enterprises weighing sovereignty, cost and supply-chain risk now have a concrete regional option that keeps data and billing inside Asian jurisdictions, which will force hyperscalers to defend their pricing and partner margins across the region.

Read more: https://www.thailand-business-news.com/pr-news/huawei-cloud-introduces-token-service-in-asia-pacific^