Episode #7, this episode aired on March 31, 2025
Analyst(s): Dion Hinchcliffe
Publication Date: April 7, 2025
Dion Hinchcliffe breaks down GTC 2025, AWS’s Trainium strategy, ServiceNow’s agent-driven Yokohama release, Celonis vs. SAP, and the Windows 10 end-of-life decision. This month’s CIO Pulse Report focuses on AI infrastructure consolidation, cost shifts, and SaaS data control. CIOs face critical decisions on platforms, performance, and portability.
What Are Covered in This Episode:
- Nvidia’s GTC 2025: Blackwell Ultra, Nemotron Models, and Dynamo Llama Agents
- AWS’s Trainium Pricing Strategy to Challenge Nvidia’s Dominance
- ServiceNow Yokohama Release and Rise of Autonomous Enterprise Agents
- Celonis Lawsuit Against SAP and the SaaS Data Portability Debate
- Windows 10 End-of-Life and the Enterprise Shift Toward AI PCs
For the full episode, please click on this link. Don’t forget to follow Dion Hinchcliffe in the CIO Pulse Report for the latest insights from top industry experts.
Nvidia Redefines the AI Stack with GTC 2025 Announcements
Nvidia’s GTC 2025 event delivered a clear message: the age of GPU-limited innovation is over. With the launch of Blackwell Ultra, Nvidia introduced the world’s fastest chip, designed not only for training AI models but also significantly boosting inference performance – a major cost center for CIOs. Alongside hardware, Nvidia debuted its suite of foundation models under the Nemotron brand and an AI agent framework called Dynamo Llama. These are not generic models – they are built to handle real-world enterprise tasks such as code generation, multi-step workflows, and complex reasoning. Importantly, Nvidia is packaging all this into a tightly integrated software suite aimed squarely at enterprise AI stacks.
The enterprise focus was evident throughout GTC, with deployments showcased by global banks, telecom firms, healthcare giants, and public sector agencies. These use cases highlight Nvidia’s ambition to deliver not just chips but a full AI operating system, complete with orchestration, simulation, and agent capabilities. The implications for CIOs are profound. The AI infrastructure stack is consolidating around vertically integrated ecosystems. This offers efficiency but increases dependence. CIOs must now re-examine their AI infrastructure strategy – especially the total cost of inference and the long-term viability of agent-based, vertically owned platforms.
AWS Offers Trainium at 75% Discount to Undercut Nvidia
AWS is leading a major pricing shift in the AI infrastructure market. Reports reveal that AWS is offering Trainium – its in-house AI training chip – at just 25% of the cost of Nvidia’s H100 in select enterprise deals. The move represents a strategic push by AWS to monetize its silicon, challenge Nvidia’s dominance, and create deeper customer lock-in up the stack. The goal is to pull enterprises into AWS’s AI ecosystem and challenge CUDA’s standing as the default AI runtime. This pricing gambit marks a clear signal: AWS wants to own more of the AI lifecycle, from silicon to the service layer.
However, Nvidia’s ecosystem still dominates the enterprise. Switching from CUDA to Trainium or Inferentia requires massive changes – retraining models, rebuilding pipelines, and retooling operations. These are time- and cost-intensive tasks that most enterprises aren’t yet prepared to undertake. However, for CIOs watching their GenAI costs scale rapidly, a 75% cost reduction in training infrastructure is hard to ignore. While it may not spark an immediate exodus from Nvidia, it introduces serious questions about AI infrastructure flexibility. The prudent move is to monitor Trainium’s performance benchmarks and software compatibility – and to build optionality into your long-term AI infrastructure roadmap.
ServiceNow Yokohama Release Brings Autonomous Agents to Enterprise Workflows
ServiceNow’s Yokohama release marks a significant leap forward in enterprise AI adoption. It introduces AI-powered agents that no longer just recommend actions – they now take them autonomously. Without human intervention, these agents can execute routine processes, manage approvals, and trigger workflows across IT, HR, and customer service functions. This is a transformative step beyond GenAI chatbots and signals the start of action-driven AI systems embedded directly into enterprise platforms. It’s a move toward autonomous operations – not just smarter recommendations.
Yokohama also upgrades ServiceNow’s Strategic Portfolio Management suite, enabling CIOs to align funding with business OKRs in real-time. This allows for dynamic reprioritization of projects and smarter resource allocation guided by AI-generated insights. Critically, ServiceNow’s architecture supports these changes with strong governance, transparency, and compliance – significant for organizations in regulated industries where GenAI adoption has lagged due to trust issues. This is not just a feature upgrade; it’s a shift in how enterprise software platforms operate. For CIOs, the release signals that enterprise vendors are embedding AI agents directly into the platforms where work happens. It’s time to evaluate how these capabilities can reduce manual workloads and accelerate throughput – and whether ServiceNow is evolving into your next operational AI layer.
Celonis Lawsuit Against SAP Raises Critical SaaS Data Access Questions
Celonis’ legal action against SAP highlights a deeper concern facing CIOs: data ownership in cloud-based platforms. The lawsuit claims that SAP restricted Celonis’ access to customer data and used that control to build and launch a competing process mining product. This situation underscores an emerging pattern: dominant SaaS providers increasingly use control over customer data as a competitive lever – not just a service feature. For CIOs, it raises a vital question: who truly owns your enterprise data when it resides inside someone else’s software?
The issue is not limited to SAP. Microsoft is charging premiums for API-based access to Copilot-compatible data, while Salesforce has limited data extraction through its licensing. This trend turns data access into a revenue strategy – and a barrier to composable enterprise architectures. For organizations pursuing best-of-breed solutions or custom AI workflows, this approach could severely limit agility and drive up costs. The Celonis-SAP lawsuit may not resolve quickly, but it could set a legal precedent for data portability in SaaS. CIOs should treat this as a wake-up call. It’s time to map where your most critical data lives, under what terms, and what it would take to extract it – before strategic lock-in becomes irreversible.
Windows 10 End-of-Support Pushes Enterprises Toward AI PC Decision
Microsoft will end support for Windows 10 on October 14, 2025, creating a major decision point for CIOs. The core question is: adopt AI PCs equipped with dedicated neural processing units (NPUs) or stick to a selective, cost-conscious refresh strategy. Microsoft is pitching AI PCs as productivity powerhouses, delivering enhanced Copilot responsiveness, on-device inference, and intelligent multitasking capabilities. The push is clearly on with major OEMs backing the move and Windows 11 positioned as the AI-native OS. However, the shift is not universally compelling. AI PCs come at a premium, and not all roles will benefit equally. Many business workflows – especially in SaaS-heavy environments – won’t immediately leverage the hardware’s full capabilities. However, the strategic value lies in standardization. A uniform hardware baseline enables enterprises to deploy future AI use cases more seamlessly, from secure offline assistants to advanced local inference and UI enhancements.
For many, this refresh cycle is the first since the pre-COVID era and intersects with hybrid work demands. CIOs must create a framework to guide adoption: prioritize AI PCs where ROI is clearest, test large-scale Copilot rollouts, and evaluate long-term productivity gains. The endpoint is evolving – and may become the next key AI delivery platform.
Click here to view the full webcast on x.com.
Disclosure: The Futurum Group is a research and advisory firm that engages or has engaged in research, analysis, and advisory services with many technology companies, including those mentioned in this article. The author does not hold any equity positions with any company mentioned in this article.
Analysis and opinions expressed herein are specific to the analyst individually and data and other information that might have been provided for validation, not those of The Futurum Group as a whole.
Other insights from The Futurum Group:
Gemini 2.0, AI Agents, and the Future of IT – A Recap from The CIO Pulse Report
Unpacking the Benefits of AI PCs – Six Five On The Road
The Six Five Pod | EP 254: Unpacking GTC: Nvidia’s AI Dominance and the Hyperscaler Challenge
Author Information
Dion Hinchcliffe is a distinguished thought leader, IT expert, and enterprise architect, celebrated for his strategic advisory with Fortune 500 and Global 2000 companies. With over 25 years of experience, Dion works with the leadership teams of top enterprises, as well as leading tech companies, in bridging the gap between business and technology, focusing on enterprise AI, IT management, cloud computing, and digital business. He is a sought-after keynote speaker, industry analyst, and author, known for his insightful and in-depth contributions to digital strategy, IT topics, and digital transformation. Dion’s influence is particularly notable in the CIO community, where he engages actively with CIO roundtables and has been ranked numerous times as one of the top global influencers of Chief Information Officers. He also serves as an executive fellow at the SDA Bocconi Center for Digital Strategies.
