New Research: The CIO in 2030: Running the Enterprise Intelligence Economy thisisfrontier.ai/research/c… Fourteen operating imperatives for a strategic world where people, agents, software, compute, data, and trust are dynamically allocated to outcomes. #CIOChat

It does seem like the “plateau of productivity” was hit pretty quickly with MCP, despite all the naysayers.

Overheard: The transition to AI will take us from VUCA to BANI: - VUCA: Volatility, Uncertainty, Complexity, Ambiguity - BANI: Brittle, Anxious, Non-linear, Incomprehensible Maybe. Probably a mix of both.

A signpost on the way to the AI-powered Future of Work: At Anthropic, agent hours are now greater than human hours says their new employment report. This will also happen in most organizations soon! Source: anthropic.com/research/emplo…

This shows the vast interest there is in the de-globalization of IT across cloud and AI. This is not a small announcement from @Microsoft. Just some of the key points are: • Sovereign Control Panel is coming, consolidating sovereignty controls, risks and gaps into a single Azure portal experience. It also adds visibility into workload dependencies and portability to support deployment planning. • Data Guardian is rolling out in the EU and EFTA. It adds oversight by a trusted EU partner with European-resident personnel for qualifying Microsoft engineer remote access. Production Touch Ledger is also beginning a phased rollout there, providing tamper-evident records of that access and the actions performed. • External key management with Azure Managed HSM is now generally available, moving last year’s commitment into an available capability. Customers can retain the root encryption key on hardware they operate themselves or through a trusted third party outside Microsoft infrastructure. • Azure Confidential Clean Rooms is now in limited preview for multiparty analysis of sensitive data. It uses confidential computing to allow approved collaboration while limiting exposure of the underlying information. • Transitioning workloads from connected to disconnected operations is entering public preview in Sovereign Private Cloud. The new capability is the transition between operating modes; support for disconnected environments itself is already established. • DocumentDB and Blob Storage on Azure Local are entering public preview, expanding the data services customers can operate locally. SQL Server’s connected and disconnected availability is existing functionality, not new in this announcement. • GitHub Enterprise Local will become generally available at Ignite as a validated GitHub Enterprise Server appliance on Azure Local. Microsoft also describes GitHub Copilot inference using locally hosted models, although the text gives no separate availability milestone for that capability. • Foundry Local will become generally available at Ignite for connected and disconnected operations on Azure Local, with an expanded model catalog. Local inference and agentic retrieval will enable natural-language interaction with local enterprise data, with Microsoft 365 Local integration in preview for approved customers. • Confidential virtual machines on Azure Local are now generally available, bringing hardware-encrypted memory and hardware-based isolation to customer-controlled infrastructure. Workload attestation provides additional assurance before sensitive data or encryption keys are released. • Azure Local will gain multitenancy, separating resources, governance and management across tenants sharing the same infrastructure. This supports distinct internal security boundaries and locally operated cloud services, but Microsoft provides no availability date in the supplied text. • Microsoft is publishing new sovereign business continuity guidance and architecture patterns. These are planning and implementation resources for designing and testing resilience, rather than a new cloud service. This is a massive shift in focus for where cloud runs and how it runs. It also makes operating cloud more complex and almost certainly more costly. But now sovereign cloud is now at least practical.

This has always been the key question when adopting powerful new tech in the enterprise like AI: Are the platform admins within the company actually turning it on? It’s one of the least-known but biggest gates to tech adoption. CIOs: Ask this question for all your internal platforms! And be ready to support your admins when a flood of new oversight + support is required.

Agentic AI is now breaking SaaS’s seat-to-value link. @Salesforce and @Mondaydotcom now have agents on usage meters. No seats means pay for job, meaning more $. Who controls the billing controls the economics. Read our analysis + 3 CIO recommendations: thisisfrontier.ai/news/sales…

And for most orgs, that will be just 6-9 months. AI platform capability — even if you’re just bringing together other platforms together for your own enterprise — is already highly strategic. Because it drives so many aspects of the business:Cost, innovation, control, risk, etc.

CIOs unprepared for agentic workforce shift ciodive.com/news/gartner-sur… Supposedly, just 4% have plans for agents, but I believe that is low. Nevertheless, most plans aren’t ambitious enough IMO.

The march toward autonomous operations is now reaching the physical world. @Honeywell’s new Forge Cognition pushes AI beyond advice toward action. The real bottleneck isn’t intelligence. It’s who holds the kill switch. Our analysis, with 3 CIO moves: thisisfrontier.ai/news/honey…

Insightful #CIO graphic from @dhinchcliffe

We are starting to get clarity on the business models of AI-led organizations. A good breakdown that helps survey the space for digital officers looking for the best ways to build AI revenue streams…

Hearing from multiple sources including AWS that their compute capacity is likely sold out beyond 2028, due to AI demand. This shifts power further toward the hyperscalers. AWS customers face higher effective prices as older long-term contracts roll off and reprice closer to elevated spot rates, plus tighter allocation (new capacity logically goes to the highest-commitment buyers). So what’s a CIO to do now? • Lock in multi-year reserved capacity or commitments at today’s rates while you still can. • Aggressively optimize utilization and model efficiency — every unused GPU or inefficient inference run gets more expensive. • Most important: Build multi-cloud optionality and negotiate harder on term length and pricing escalators before the window closes. Enterprise AI routers can really help too. I’ll have research on this soon.

I’m looking at tools this weekend that let CIOs control agents and manage the downside. A new development: @Veeam has just launched Intelligent ResOps for @Microsoft365, including file-level AI Undo. The bigger shift? Agent governance is moving beyond permissions and kill switches to reversing the damage. Analysis and 3 CIO recommendations: thisisfrontier.ai/news/veeam…

Lots of employment data out there but this is one the more interesting: Entry-level jobs increasing in AI-intensive firms. It could just be boats rising, but we’ll see. This data will be watched closely for large-scale employment shifts. My take: It’s both boats rising + new AI-driven job types that need to be filled.

New Research: The future of work isn’t about making people 30% faster at yesterday’s jobs. I believe it’s about making entirely new kinds of work possible. One person directing swarms of agents. Small teams conducting groundbreaking, original R&D with frontier AI. Mission Engineers turning intent into potent, verified outcomes. This is a fundamental rewiring of the division of labor. And the hardest part won’t be the tech. It will be helping people and organizations evolve fast enough to capture the extraordinary upside. My latest new major report examines seven new ways of working, quantifies their potential, and lays out a 24-month CIO roadmap. The next great productivity revolution will come from vastly redefining what work can be: thisisfrontier.ai/research/t…

IT leaders still on the hunt for @VMware alternatives cio.com/article/4236116/it-l… There have been veritable explosion of alternatives over the last 24 months. My take: The real issue is finding a vendor they will trust again.

A growing list of major companies are blocking consumer AI agents from using their services, when they can even tell. It’s the same old tired mistake. It’s a key business opportunity in disguise (API revenue at absolute minimum), but they simply can’t see it that way. Design for loss of control!

#CIOChat A4: stochastic agents (time+context+user+event constraints) with infinitely variable linear regression algos. Think of it as Jev for IAM

Looks like another fun #ciochat topic coming up. Scorecards are scorecards, but our communication around them is one area where a little more multi-/cross-cultural sensitive can be a good thing. Sports metaphors don't always translate. 😇

A3. Aggressively deploying MFA (and I don't mean SMS LOL) & #zerotrust helps harden access points & identities. Short-lived credentials using Diffie-Hellman, etc., don't mean squat: a #QuantumComputer isn't just "fast at enumerating": it literally solves all cases at once via Quantum Superposition #CIOchat