3 Before 9

Why it matters for Asia

India has been the conspicuous absence in Asia's AI race so far, with most local enterprise demand defaulting to OpenAI, Anthropic or Chinese open-source models. Sarvam at 1.5 billion dollars puts a credible domestic alternative on the board for Indian banks, telcos and government departments that need models tuned for the country's 22 official languages, not just English. For regional buyers and South Asian conglomerates building agentic products on Indian-language workflows, this is the moment the sovereign AI narrative gets a balance sheet to back it up. Read more: [Indian Startup News](https://indianstartupnews.com/funding/sarvam-ai-set-to-become-indias-next-unicorn-close-to-raising-300-million-at-15-billion-valuation-11450641)

Why it matters for Asia

Until last week China's biggest social platform was running its consumer AI chatbot on a model from a rival upstart, an awkward dependency for a company with Tencent's resources. Pulling DeepSeek out of Yuanbao is Tencent declaring it can fund its own foundation model team rather than rent one. For Asian enterprises building on Yuanbao, ima or CodeBuddy, the underlying model has just changed beneath their feet, with implications for prompt design, cost and latency. It also signals that the BAT giants are no longer content to be downstream consumers of DeepSeek output. Read more: [South China Morning Post](https://www.scmp.com/tech/big-tech/article/3351101/tencent-unveils-first-flagship-ai-model-former-openai-researcher-helm)

Why it matters for Asia

Japan's third-largest automaker is staking its recovery on AI-defined driving rather than EV powertrains, a sharp divergence from the playbooks at Toyota and Honda. For regional auto suppliers in Thailand, Indonesia and Vietnam that depend on Japanese platforms, Nissan committing to a Wayve-trained foundation driving model means a re-architecting of the local sensor and silicon stack. It is also a rare big-bet validation of the end-to-end neural-network approach over the rule-based stacks favoured by Chinese rivals such as Huawei and BYD. Read more: [CNBC](https://www.cnbc.com/2026/04/14/nissan-to-trim-global-car-lineup-boost-use-of-ai-driving-tech.html) ## THE AI IN ASIA VIEW The week's three big stories - India funding its first sovereign frontier model, Tencent ripping out a rival's AI to install its own, and Nissan handing its product future to a British driving model - all describe the same underlying shift. The era of one or two American labs setting the pace and the rest of the world building on top is being replaced by a more fragmented map, where each major Asian economy is willing to spend serious capital to keep critical AI infrastructure in-house. For Asia-Pacific buyers and builders, this means more choice on paper but more complexity in practice. Procurement teams will increasingly have to evaluate sovereign Indian models, Chinese open-source releases, Japanese on-device stacks and Korean memory-bound architectures alongside the familiar Western names. Expect 2026 to be remembered as the year regional AI strategy stopped being a single decision and became a portfolio.