scaling agentic AI across Asia Pacific<\/a> is not model quality but operational infrastructure, and the data backs a steep investment ratio. NTT DATA’s own research shows 94% of APAC organisations treat sovereign AI as strategic, 93% are reconsidering where they host AI workloads, and 96% say legacy infrastructure is slowing adoption. Shetty’s 1-2-3-4 rule frames the actual spend profile: for every dollar on AI technology, organisations need two on adoption, three on ecosystem, and four on data preparation.<\/p>\nWhat this means for your business<\/h2>\n
The organisations most likely to stall are the ones that treated their first agentic deployment as proof of concept complete. Moving from a working pilot to enterprise-wide operation exposes every unresolved question about data quality, system interoperability, and who is accountable when an agent makes a consequential call without a human in the loop. If your AI estate is still largely a collection of standalone tools rather than embedded workflow components, Shetty’s argument lands directly on your roadmap, not as a future concern but as the reason your current pilots aren’t compounding.<\/p>\n
The regulatory fragmentation point is specific and underappreciated. The EU’s AI Act gives European CIOs a single compliance surface to design against, uncomfortable as that surface is. APAC CIOs are designing against a dozen different data residency regimes simultaneously, which means governance architecture has to be modular rather than centralised. That is a real engineering and legal overhead that doesn’t shrink as you scale, it grows proportionally. The 93% figure on infrastructure relocation isn’t just a sentiment number; it represents active capital allocation decisions happening right now.<\/p>\n
Shetty’s framing, coming from a systems integrator whose business depends on making exactly this kind of complex deployment work, predictably emphasises ecosystem investment over model selection, which is where NTT DATA competes. That tilt is real, but it doesn’t invalidate the underlying claim. The 1-2-3-4 rule is the sharpest signal in the piece: if your board-level AI budget is still weighted heavily toward model licensing and lightly toward data remediation, you are misallocating, and the ratio tells you by how much. The close I’d watch for is whether your next budget cycle’s AI line items reflect that distribution or still treat model spend as the headline number.<\/p>\n
Concept deep-dive: Agentic AI<\/h2>\n
Agentic AI refers to systems that don’t just respond to a single prompt but pursue multi-step goals autonomously, taking actions, using tools, and adjusting based on intermediate results, much like delegating a task to a capable employee rather than answering a single question. The governance challenge is that accountability for outcomes becomes diffuse the moment the agent acts without direct human instruction. That diffuseness is what makes Shetty’s insistence on embedded oversight a design requirement, not a compliance add-on.<\/p>\n
Based on reporting from Scaling agentic AI in APAC<\/a>, originally published 2026-07-26 23:00:00.<\/em><\/p>\n","protected":false},"excerpt":{"rendered":"Share with your CIO NTT DATA’s APAC Google Cloud GTM leader Shashir Shetty argues that the binding constraint on scaling agentic AI across Asia Pacific is not model quality but operational infrastructure, and the data backs a steep investment ratio. NTT DATA’s own research shows 94% of APAC organisations treat sovereign AI as strategic, 93% […]<\/p>\n","protected":false},"author":1,"featured_media":6781,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[142],"tags":[185],"tmauthors":[],"class_list":["post-6780","post","type-post","status-publish","format-standard","has-post-thumbnail","category-ai-agents","tag-cio"],"_links":{"self":[{"href":"https:\/\/workai.tv\/news\/wp-json\/wp\/v2\/posts\/6780","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/workai.tv\/news\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/workai.tv\/news\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/workai.tv\/news\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/workai.tv\/news\/wp-json\/wp\/v2\/comments?post=6780"}],"version-history":[{"count":0,"href":"https:\/\/workai.tv\/news\/wp-json\/wp\/v2\/posts\/6780\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/workai.tv\/news\/wp-json\/wp\/v2\/media\/6781"}],"wp:attachment":[{"href":"https:\/\/workai.tv\/news\/wp-json\/wp\/v2\/media?parent=6780"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/workai.tv\/news\/wp-json\/wp\/v2\/categories?post=6780"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/workai.tv\/news\/wp-json\/wp\/v2\/tags?post=6780"},{"taxonomy":"tmauthors","embeddable":true,"href":"https:\/\/workai.tv\/news\/wp-json\/wp\/v2\/tmauthors?post=6780"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}