Published by Digigen | Google Cloud Partner for Thailand and Singapore
Seven announcements that change how enterprises in our region should think about deploying AI in production. Here is the Digigen take.
Last year at Google Cloud Next '25, the question was about imagining a new future for AI. At Next '26, the question is sharper. How do you actually move AI into production across your entire enterprise?
Google Cloud CEO Thomas Kurian's answer is straightforward. You need a unified stack. Chips designed for models, models grounded in your data, agents and applications built with those models, and the whole thing secured by the infrastructure underneath. As Sundar Pichai put it in his opening, Google's discipline is to always be customer zero for its own technologies. That same stack runs Search, YouTube, Chrome, and Android. Now it is being shipped to enterprises.
For Singapore and Thailand businesses, this Day 1 keynote was dense, fast, and worth slowing down for. Below is everything that was announced, why it matters in our region, and the Digigen take on what to do about it.
Throughout Google's unified stack, Gemini Enterprise is now positioned as the connective tissue between your data, your people, and your goals. Not a chat tool with enterprise controls bolted on, but the operating layer that ties intelligence and automation together across every layer of the business.
For our clients in Bangkok and Singapore, this framing matters. Most enterprises we work with are not stuck on which model to pick. They are stuck on how to deploy AI safely, observe it in production, prove it to the audit committee, and improve it over time. Gemini Enterprise is Google's structured answer to all four.
Gemini Enterprise Agent Platform is where you go to build, scale, govern, and optimize agents. As the evolution of Vertex AI, it sits on top of Google's leading infrastructure and integrates deeply with their data and security capabilities. This is the foundation of the agentic enterprise.
Choose your environment, from the low-code visual interface of Agent Studio to the code-first logic of the upgraded Agent Development Kit (ADK). AI-native coding capabilities help ship production-grade agents faster.
The re-engineered Agent Runtime supports long-running agents that maintain state for days, backed by Memory Bank for persistent, long-term context.
Centralized control with Agent Identity, Agent Registry, and Agent Gateway. Every agent gets a trackable identity and operates within enterprise-grade guardrails, including agents from your partner ecosystem.
Agent Simulation, Agent Evaluation, and Agent Observability give you full execution traces and a real-time lens into agent reasoning so your agents always hit their goals.
For MAS-regulated institutions in Singapore and Bank of Thailand-supervised institutions in Thailand, the governance pillar is the real story. Agent Identity with cryptographic IDs plus Agent Gateway with Model Armor protection plus the Security Command Center dashboard add up to a defensible architecture you can put in front of a risk committee.
For our complete deep dive on Agent Platform with dedicated Singapore and Thailand sections, read the master guide.
Thomas described the Gemini Enterprise app as the primary environment where your business actually operates. It is where employees, especially non-technical ones, can ask questions of enterprise agents, generate media, engage with prebuilt agents, and create their own with conversational interfaces. All with governance, compliance, and security built in.
What is new in this foundational interface:
To illustrate the point, three-time Olympic gold medalist Shaun White joined the stage. He described the old way of training: camcorders and guesswork. You land a trick, watch the tape, and try to figure out what to fix. Alongside Google Cloud Tech Lead Jason Davenport, the demo showed a model built with Google DeepMind that tracks an athlete's body in 3D space from a single 2D video, helping Shaun understand the physics of his tricks rather than relying on intuition alone.
For Thai and Singapore enterprises that have been weighing Microsoft 365 against Google Workspace, the Microsoft 365 export interoperability inside Canvas removes one of the last "stay on M365 because the team uses Word" objections. Combined with Workspace Intelligence (covered later), the migration math has shifted materially. We are already fielding inquiries from Thai customers reassessing the decision.
Amin Vadhat, SVP and chief technologist for AI and Infrastructure, walked through the upgrades to AI Hypercomputer, the integrated supercomputing layer underneath every AI workload on Google Cloud. The headline is the eighth-generation TPU, which Amin called "a thing of beauty." Because the demands of training and serving have completely diverged, the new TPU family ships as two purpose-built chips.
New Inter-Chip Interconnect (ICI) technology scales up to 9,600 TPUs and 2 PB of shared, high-bandwidth memory in a single superpod. Three times the processing power of Ironwood and up to double the performance per watt.
New Boardfly topology directly connects 1,152 TPUs in a single pod. Three times more on-chip SRAM and a Collectives Acceleration Engine deliver 80% better inference price-performance than the prior generation.
The rest of the AI Hypercomputer story:
Most enterprises in Bangkok and Singapore are not training frontier models. The headline number for our region is the 80% inference price-performance improvement on TPU 8i. That flows directly to the bottom line for any high-volume inference workload, from Thai-language customer support to Singapore retail recommendation engines.
Combined with Google's asia-southeast1 (Singapore) and asia-southeast7 (Bangkok) regions, the cost equation for production AI in our region just got materially better.
As Karthik Narain, chief product and business officer, put it: reasoning without context is just a guess. Google has rebuilt the data platform around the idea that agents now need data the same way humans do, with semantic context attached. They renamed it the Agentic Data Cloud.
Constructs a unified, dynamic context graph of your entire business. Smart Storage and the Object Context API tag and enrich files in Cloud Storage with metadata before an agent touches them. Integrated with Gemini's Deep Research Agent.
A Gemini-powered data science authoring experience across IDEs, notebooks, and agentic terminals. Intent-driven development in Python, Spark, and SQL.
A real-time, serverless Apache Spark engine up to 4.5x faster than open-source alternatives and up to 2x better price-performance than the leading competitor for large datasets.
Built on Apache Iceberg. Query data sitting in AWS or Azure without copying it. Zero-copy access for agents to your distributed data estate.
For Singapore-headquartered regional groups with operations across Indonesia, Malaysia, Thailand, and the Philippines, Cross-Cloud Lakehouse is the most useful announcement in this entire keynote. You can keep data resident where PDPA and other regional rules require, and still let agents reason across the full footprint.
For Thai enterprises with data scattered across SAP on-premise, Thai-language ERPs, Google Sheets, and legacy Oracle databases, Knowledge Catalog removes the need for a heroic data migration project before deploying useful agents.
Francis deSouza, COO and President of Security Products, was direct: human analysts cannot keep up with AI-driven attacks. Security must become an autonomous force, responding faster than the threat itself.
Three new agents in Google Security Operations:
Plus, you can build your own security agents with remote Google Cloud MCP server support for Google Security Operations (now generally available) and access them from the Google Security Operations chat interface (now in preview).
From Wiz, now part of Google Cloud, several new capabilities arrived:
For Thai banks and Singapore digital banking platforms facing rising deepfake and AI-enabled fraud, Google Cloud Fraud Defense is going to matter very quickly. Bot, human, and agent discernment is no longer a future problem. It is a daily operational reality.
For enterprises in our region without 24/7 SOC staffing, the autonomous Threat Hunting and Detection Engineering agents are a force multiplier. Singapore's Cyber Security Agency expectations are rising fast, and Bank of Thailand has signaled tighter scrutiny on AI risk management. The agentic defense stack maps directly to both.
Yulie Kwon Kim, VP of Product for Google Workspace, framed Workspace Intelligence as a unified intelligence layer that lives inside every Workspace app. It connects the dots across your work, your priorities, and the people you work with, then lets AI do the heavy lifting.
Synthesize information, surface insights, and query projects across Workspace from your Chat window. Includes proactive daily briefings and the ability to take immediate action like scheduling meetings or creating Docs.
Docs, Sheets, and Slides use Workspace Intelligence to synthesize information and create professionally formatted drafts that match your voice, style, and brand.
A personal, proactive inbox assistant powered by Gemini.
Instantly organizes your team's files and emails to manage workflows, generate content, and deliver answers based on rich project context. Drive becomes an active collaborator.
Executes complex, multi-step tasks across Google Workspace apps without leaving Gemini Enterprise. The bridge between your AI agents and your everyday productivity environment.
Combined with the new Microsoft 365 export interoperability and the Rapid Enterprise Migration that is now 5x faster, Workspace Intelligence reshapes the productivity decision for any Singapore or Thailand enterprise still on Microsoft 365. We expect this to be a live boardroom conversation in our region throughout 2026.
Carrie Tharp, Google Cloud VP of Go-To-Market Strategic Industries, framed it sharply: in the agentic era, an agent is not just a tool. It is a strategic extension of your business, built to expand reach, deepen engagement, and personalize service at scale.
Gemini Enterprise for Customer Experience now offers a full suite to enhance the customer journey from discovery through ongoing interactions:
Carrie's example was direct. If a customer moves from text chat to a phone call, the agent seamlessly remembers exactly where they left off. For consumer-facing brands in Bangkok and Singapore, this is the kind of detail that separates "we deployed an agent" from "we transformed our customer experience."
For Thai consumer-facing businesses, the Omnichannel Gateway does not yet include native LINE support. This is a real gap because LINE remains the dominant messaging channel in Thailand. The good news is that Agent Platform's MCP and native ecosystem integration patterns make a LINE OA bridge much cleaner to build than it was on Vertex AI.
For Singapore retailers, F&B groups, and hospitality brands, the WhatsApp-plus-voice unified context is exactly what most contact center stacks have struggled to deliver. FairPrice Group's Smart Cart program, called out by Google as a flagship case study, is worth studying closely.
Five concrete actions based on the Day 1 announcements:
From Day 1 announcements to your first agent in production.
Digigen helps enterprises in Singapore and Thailand turn Google Cloud Next announcements into deployed reality.
Talk to Digigen →MAS-ready governance, Cross-Cloud Lakehouse for multi-cloud, and 90-day actions for Singapore enterprises.
Read recap →TPU 8i economics for Thai-language inference, LINE integration, and 90-day actions for Thai CIOs.
Read recap →The complete deep dive on Gemini Enterprise Agent Platform with dedicated Singapore and Thailand sections.
Read full guide →Digigen is a Google Cloud Partner and Microsoft Solutions Partner with offices in Bangkok and Singapore (Digigen Pte Ltd). We help enterprises design, deploy, and govern AI, cloud, and productivity solutions across Thailand and Singapore.
Contact us at hello@digigen.io or visit digigen.io.
For our AI agent consulting and platform work, see The Agentiv.