NVIDIA’s #GTC Berlin keynote ends with amusing robots from Pollen Robotics joining NVIDIA CEO Jensen Huang on stage!
I created a short, amusing video montage from the keynote in Berlin for you all!
Jensen delivered another compelling and brilliant keynote, focused on the ecosystem, with a big message for European technology and government leaders.
Google Cloud has introduced Gemini for government services, now in preview, to help federal, state, and local agencies apply AI to everyday operations. The focus is on work shaped by regulations, sensitive information, legacy systems, and human accountability.
The offering combines government-specific instructions, connections to existing agency systems, partner-built agents, and centralized governance. Google says agencies can connect their systems without migrating data while preserving existing access permissions.
Government use cases are a useful test of enterprise AI because producing an answer is only one part of the job. That answer must reflect the applicable rules, respect access controls, cite supporting evidence, and fit an established approval process.
This announcement reinforces a broader shift I see across the industry: AI value increasingly comes from connecting models to institutional knowledge and operational systems. The model matters, but so do integration, information architecture, governance, and the people responsible for decisions.
From a systems economics perspective, agencies should measure the full cost of completing a task. A faster document review has limited value if employees spend the time saved correcting errors or checking unsupported conclusions.
The measures of success should include:
Because there is no one AI workload, there is no one KPI for AI. Government adoption will need to demonstrate value workflow by workflow. This preview gives Google an opportunity to show how governed agents can translate technical capability into better public service.
The platform supports connections across Oracle, Salesforce, ServiceNow, Google Workspace, and Microsoft 365. Its security foundation includes an option to deploy within Google Cloud’s Assured Workloads FedRAMP High Data Boundary. Google also says agency information will not be used to train its foundation models without permission.
More from Google on this announcement here.
I believe domain-specific AI will become better, safer, and more economical for the enterprise work it is designed to perform. That is why I have repeatedly praised Google’s focus on industry-specific AI.
At Gemini at Work on October 8, 2026, Thomas Kurian reinforced that direction. For me, the significance goes beyond another product announcement: it points to where enterprise AI can create lasting value. We had an analysts Q&A session with Google execs after the announcement on October 8 which brought some additional clarity around the announcements.
Google’s Gemini at Work announcement identifies Financial Services and Legal as in preview, with Government, Healthcare, and Retail coming soon. These are the availability labels announced at the event, rather than a claim that all five offerings are generally available.
The direction matters. Enterprise buyers need AI that understands the context of their work: its terminology, information sources, workflows, permissions, and standards of quality. A capable model is the starting point. Putting it to productive use requires the surrounding system.
This is where systems economics becomes central. Token pricing alone does not tell an enterprise what AI costs. The full cost includes data preparation, integration, inference, review, rework, and ongoing operations. A cheaper response that requires repeated corrections may produce a more expensive workflow. A more capable system may justify its cost if it consistently reduces that work.
My preferred question: What does it cost to complete a useful, verifiable business task? Domain-specific AI earns its economic advantage when it improves that answer after the cost of specialization is included.
I see Google’s combination of frontier-model development and hyperscale cloud delivery as a significant strategic advantage. An industry focus gives it a way to connect those capabilities to the environments where enterprise work happens.
The opportunity is to make more of the required infrastructure, integration, and workflow knowledge reusable across customers. Execution will determine how much of that opportunity becomes a practical advantage.
I have praised Google AI’s Industry/Vertical focus, repeatedly since they have announced it. In this week’s announcements, Thomas Kurian reiterated that focus with latest Gemini At Work announcements from Mountain View!
This is Google’s huge differentiator as a Frontier Lab and a Hyper Scaler/Cloud Provider. I am sure that they will add more industries as time progresses, tech matures, and more industries come forward to collaborate.
Google AI’s Industry Focus (so far)
– Financial Services (in preview)
– Legal (in preview)
– Government (coming soon)
– Healthcare (coming soon)
– Retail (coming soon)
My conviction remains: domain-specific AI can turn broad model capability into better enterprise outcomes. Its value will be proven through the quality, safety, and economics of the work it completes.
Adapted and expanded from my original LinkedIn post. Graphic reused from that post. Product availability references reflect Google’s October 8, 2026 announcement.
Without talking to each other, legacy big-tech (likes of Google, Microsoft, Amazon, Nvidia) and neo big-tech (likes of OpenAI, Anthropic, xAI) will remain way off in their approaches in product releases and Go to Market approaches. That has and will create friction for the vendors and FUD (fear, uncertainty & doubt) in the market. The fact is, different operating philosophies eventually become market friction.
The September 29 White House gathering included executives from Nvidia, Google, Microsoft, Amazon, Meta, OpenAI, Anthropic, xAI, AMD, Broadcom and others—so it genuinely brought together what you call established Big Tech and the newer frontier-AI companies. Six AI companies also signed a voluntary accord involving internal controls, independent audits and board-level oversight.
Why Attend
SC26 is the leading technical program for the HPC community, offering a mix of cutting‑edge research, industry innovation, and community engagement. It is relevant to researchers, engineers, vendors, policymakers, and students seeking to explore HPC’s latest advances and applications. For the most current schedule, registration, and program updates, visit the official SC26 site: sc26.supercomputing.org
Technical Program: Peer‑reviewed papers, expert‑led sessions, panels, tutorials, workshops, and special events for early‑career researchers .
Keynote: Animashree Anandkumar (Caltech) on AI’s next frontier — understanding the physical world to accelerate discovery .
HPC Unites Plenary: Emphasizing collaboration and community impact .
Regional Tech Hub: AI and quantum research with real‑world impacts in medicine and HPC .
Awards: Includes the SC26 Test of Time Award for lasting GPU sparse computing contributions .
Students@SC: Volunteer programs, cluster competitions, mentoring, and networking .
SCinet: Large‑scale, high‑performance network connecting exhibitors and attendees .
See the KubeCon 2026 Agenda — KubeCon + CloudNativeCon North America is in Salt Lake City, November 9 to 12, 2026. See the KubeCon 2026 schedule, speakers and sessions, then register for your ticket. Workshops.
Highlights include:
Dinesh Nirmal – SVP, Software, IBM
Garrett Rowe – CEO, NeuralSeek
Suman Devarasetti – VP, AI Enablement & Payments, BANCORP
Chris Wright – CTO, Red Hat
Rodney Mullen – Skater, entrepreneur
Tim O’Reilly – CEO, O’Reilly Media
Bill Higgins – VP, AI Developer Relations, IBM
This event focuses on building real‑world skills in:
AI applications (watsonx AI, intelligent agents, automation)
Data management (integration, governance, trusted data for AI/analytics)
Hybrid cloud management (IBM Cloud, Red Hat, DevOps, SRE)
Infrastructure (IBM Z, LinuxONE, Power, Storage)
Security & governance (data protection, compliance, “secure by default”
There were two major, industry impacting announcements made by Nvidia in last 24 hours. One was the announcement of $150 billion boost for shares buyback program (a record), and the other one is announcement of Open Agent Safety Platform with several partners behind it (including Anthropic.
My POV. Announced on September 28, Open Agent Safety Platform is a master stroke in GTM strategy. Amidst (AI security related) high tensions in the market, Nvidia announcing this program shows the position Nvidia is in and how high the stakes are. It’s clear that this step will strengthen Nvidia’s moat further.
NVIDIA Open Agent Safety Platform consists of NVIDIA OpenShell open source software and the NVIDIA Sentry reference system design that enables full-stack governance and control across software and the hardware, compute and robotics systems that run agents.
More on Open Agent Safety Platform here: https://nvidianews.nvidia.com/news/open-agent-safety-platform
More on shares buyback program (announced on September 28) here: https://nvidianews.nvidia.com/news/nvidia-announces-a-150-billion-share-repurchase-authorization-increase
Third big AI related news coming for m Silicon Valley was AMD acquiring World Labs for $8.2 Billion.
AMD is acquiring Fei-Fei Li’s World Labs in an $8.2 billion, all-stock deal. Dr. Li joins AMD as EVP/Chief Scientist.
My POV: it’s a great move by AMD as they need AI research and software chips to push their platform to the market. Dr. Li’s reputation will give a big boost to AMD platform in research and frontier labs space.
