Welcome to the latest edition of the CIO In The Know newsletter. Insights that move CIOs forward.
First, be sure to subscribe!
This week felt like déjà vu. Let me explain. Over the past several months I’m finding a recurring pattern in many of my conversations with CIOs and executive teams in that the same issues are repeatedly coming up. Why is that? Simply put, there is a gap between the problem and the solution that is not getting closed.
Addressing the gap mentioned in the past couple of newsletters alone is a lot to consume and takes time. Unfortunately, for the CIO, time is not your friend. However, there is an upside. Read on…
Aside from the repeat appearances, let’s cover what else is coming up in conversation
In Discussion: What is on the minds of CIOs?
This week’s topics could be captured under the headings of infrastructure, data, AI, security and relationships. While I cover each of these from a topline, the nuance of each is incredibly fascinating.
New generation of data center infrastructure
In last week’s CIO In The Know newsletter, I talked about the challenges CIOs are facing with budgeting and hardware acquisition. Beyond those aspects, I’m fielding questions about how data centers are changing to support the new generation of hardware from companies like Lenovo, Dell, HPE and Cisco.
The data centers themselves need retrofitting (if possible) as well to support the dramatically increased power, cooling and weight densities required for today’s infrastructure. For example, a typical 5-10kW rack is no match for the 50-100kW rack density on tap today.
Beyond the data center itself, let’s look at a few examples of how each of the major vendors are shifting into the new AI-driven requirements.
Lenovo is leaning heavily into AI data center infrastructure with purpose-optimized solutions including their Neptune Liquid Cooling. For some time, the heat produced by newer, faster processors and memory was getting to a point where free air cooling was not adequate and liquid came into view. Neptune addresses this by bringing liquid cooling inside the server.
Dell launched their AI Factory with Nvidia as a platform for AI-optimized infrastructure based on Nvidia processors. Dell’s edge-to-data center approach provides support for running AI workloads on-premises. One interesting standout is that Dell’s PowerEdge XE9780 takes an air-cooled approach. However, a fully populated rack could consume over 50kW.
HPE recently increased the emphasis on their networking products in the wake of their acquisition of Juniper Networks. I shared some thoughts on the merger at the time of the close here. HPE is somewhat unique with its strong operational software portfolio while also providing AI-optimized infrastructure for enterprises with their HPE Private Cloud AI solutions.
Cisco is well known for their networking and security products. In the past 12-18 months, Cisco has accelerated work on their integrated management console Cisco Cloud Control as well leveraging AI to the max. It really is impressive how far Cisco has leveraged AI for the benefit of the network and security paradigms. I’ve been taking a deeper dive with Cisco over the past few months and impressed. More to come as some discussions are under NDA.
Folks have largely discounted infrastructure as the classic stacks of server, storage and network. Today, AI is driving a massive resurgence of AI-enabled infrastructure, and the opportunities are significant for enterprises.
Tokenomics
One of the reasons for the resurgence in on-premises infrastructure is token consumption. An increasing number of major technology companies are announcing they have exhausted their token budgets.
Uber announced it exhausted their entire 2026 annual AI budget in just four months! They’re not alone. Microsoft, Walmart and Royal Bank of Canada (RBC) all announced they are burning through AI budgets more quickly than expected and have had to pull back on consumption.
For enterprises, token-based pricing is a challenge that I wrote about here. At the same time, CIOs are finding ways to ensure that the right models are used based on the cost, efficiency, performance and that outcomes align with the work performed. While many would prefer to encourage token use, they are looking at ways to engage guardrails to maximize the value of their consumption.
Tokenomics is such a hot topic that the FinOps Foundation announced the formation Tokenomics Foundation and shifted the name of their annual FinOps X conference to [tokenomicon] + finops x.
The complex state of the data estate
Interrelated with tokens is data. One CIO this week referred to the data estate as ‘messy’. For many, that is an understatement and presents a significant challenge for those looking to leverage AI.
Fundamentally, AI requires organization within the data estate. Just access to data is not enough. For AI to truly accel, it needs context. This is especially true when it comes to customer data leveraged for customer experiences (CX).
At the AWS Summit in NYC last month, AWS announced AWS Context. AWS Context presents an interesting opportunity for the system to build a knowledge graph of your data that agents can leverage. Others have their own solutions. For example, SAP’s Knowledge Graph creates a knowledge graph based on your business data within SAP.
Agentic AI challenges
One common thread coming up repeatedly with both vendors and end-users alike is a focus on the positive and capabilities that AI brings. What is missing is a healthy, balanced conversation of the risks associated with using different tooling. Of course, risks can turn into opportunities for growth and improvement too.
For example, let’s look at code creation with AI. This is a common use case where AI is being used to create and evaluate code. While the upside is velocity of code creation has accelerated, AI is not being used widely enough to ensure vulnerabilities are identified and addressed as part of this process.
Additionally, enterprises are largely thinking about AI agents like humans and applying similar rules and thinking. While I touched on agent identity in this post, the broader issue is starting to raise issues. I suspect it will come up again as the CIO and CHRO conversation evolves and they revisit HCM strategies around EX. More on that below.
Patch Tuesday is becoming patch now
This week there were a few conversations among CIOs lamenting about how we used to take several days between ‘patch Tuesday’ and when patches were applied. This provided time for teams to evaluate patches, test and understand their implications instead of blindly deploying them and hoping nothing breaks. It also allowed time to deploy during a maintenance window (typically a Friday/ Saturday night) providing recovery time should something go sideways.
Today, those days of luxury between patch Tuesday and the maintenance window are largely gone. Unfortunately, in the advent of AI, vulnerabilities are coming fast and furious. Oracle’s Critical Patch Update for July 2026 included over 1,400 updates! It is expected that this velocity will increase with more patches coming, more frequently. CIOs know they need to rethink the balance between risk of vulnerability and risk of disruption. Largely this means moving to automatic patching.
AI security concerns raising broader concerns about AI
Last week, one of OpenAI’s models broke free of its guardrails and created a security incident for Hugging Face. I wrote about it here.
This week, two more high-profile incidents happened between major AI vendors. Yesterday, Anthropic announced it is investigating three real-world incidents in their cybersecurity evaluations. The scary thing is that they didn’t realize this until they started looking deeper after the OpenAI incident. Today, during its own investigations, OpenAI found that additional autonomous agents escaped containment.
The concern is two-fold. First, if well-established companies like OpenAI and Anthropic are unable to manage and/ or detect when agents go awry, how can enterprises be expected to do the same? Second, these incidents are likely to amplify the intensity of the scrutiny coming to both companies, all model providers and more broadly AI. All of these will have implications for CIOs.
CX and the impact of AI slop
As if the incidents in the past two weeks were not enough, companies are starting to push back on how AI is being used. This is not a broad push back on AI as there is still strong optimism about AI’s opportunities.
However, how AI is used and where is coming up more broadly. This week, LinkedIn announced that will start reporting AI-generated ‘slop’.
Disclosure: No part of this post was written by AI.
As enterprises are looking to improve their customer’s experience, they are regularly leaning into AI as an enabler. However, they need to be careful to understand where and how to best use it…while leaving it out of other activities.
We all know that AI is generating a lot of buzz…noise? Slop too. The focus needs to be how to leverage the impact of AI’s functional differentiators versus just creating more noise.
The CIO and CHRO are talking platforms
This week, several executive conversations brought up the discussion of HCM platforms and impact to employee experience (EX). As expected, stalwarts such as SAP’s SuccessFactors and Workday were mentioned…but so were several other players looking to disrupt the space. CIOs are taking notice.
The backchannel says that there are significant gaps to be filled and opportunities abound for providers to fill them. Just focusing on HCM is no longer enough. Neither is just focusing on the CHRO. Based on conversations over the past few months, this is a space ripe for its next evolution.
Looking at the upside
CIOs have quietly been looking for a something to take the froth off the AI conversation. Not to stop it, but to take the temperature down a bit. The recent incidents at OpenAI and Anthropic wasn’t what most expected but I wonder if it is enough.
At the same time, vendors are rethinking their strategies to engage the CIO to help bridge the gaps and create a reasonable pathway forward. This is a good thing as past playbooks need not apply.
The net-net is that the challenges, while complicated, present a spectacular opportunity for CIOs to shine. For those that are business-minded, this is their time. Of course, the path forward is not trivial. But most CIOs I know that are oriented this way welcome the challenge.
What’s News
Here’s the latest from the blog and podcast.
New podcast episodes
Be sure to subscribe on Apple Podcasts, Spotify or wherever you get your podcasts, so you don’t miss a conversation.
Now It’s Your Turn!
There are three things you need to do:
- Subscribe!: Subscribe to both the CIO In The Know newsletter and to the CIOitk podcast on Apple Podcasts or Spotify to ensure you don’t miss an update.
- Join the Conversation: Like, share and engage. What’s your perspective?
- Forward the Newsletter: Help grow the community. Forward to a friend or colleague and encourage them to subscribe!
Look for more content coming your way…and be sure to subscribe!
Questions, comments, suggestions for the newsletter? I’d love to hear from you!
Discover more from AVOA
Subscribe to get the latest posts sent to your email.
1 comment