You Have To Speak IBM’s Language If You Want To Be Heard
May 11, 2026 Timothy Prickett Morgan
Like many of you, I had been expecting for Big Blue to deliver the kicker to the entry level “Bonnell” Power10-based Power S1012 that was announced a few years back during the PowerUP 2026 conference. This long-awaited machine will be the only Power11 system that is offered for those many, many IBM i customers in the P05 software tier, and will presumably have a beefier hardware configuration that crosses over into the P10 IBM i software tier.
POWERUp 2026 came and went, and there was no Bonnell+ kicker and there also was not the expected Technology Refreshes for the IBM i operating system, either. And last week, IBM’s Think 2026 event for partners and customers alike came and went, and no entry Power11 machine, which we all presume will be called the Power 1112, was unveiled and as far as we know, the Technology Refreshes have also still not been rolled out.
Ahead of the Think 2026 event, we sat in on a pre-briefing with IBM chief executive officer Arvind Krishna and chief commercial officer Rob Thomas, hoping that they might give the IBM i platform some love and talk about the Power S1112 and the Tech Refreshes. The pair did talk about a bunch of interesting things, including a special set of security, encryption, governance, and observability software running atop the OpenShift Kubernetes platform from the Red Hat division, called Sovereign Core, which ironically enough is shipped by IBM atop Dell PowerEdge servers running AMD Epyc processors and AMD or Nvidia GPUs to support provably hardened AI workloads.
While watching this preview from Krishna and Thomas, all I could think of was the Sovereign Core should be something running on Power Systems machines using Power11 processors and supporting IBM’s own Spyre matrix math accelerators as well as AMD and Nvidia GPUs. Perhaps this will be coming, perhaps not.
We certainly hope so.
Krishna was very clear about how IBM was focused on hybrid cloud and AI, and that Big Blue had no interest any more in building foundation models and had partnered with Anthropic, OpenAI, and others to get the latest-greatest models. (That said, IBM’s own Granite models in the WatsonX stack are very useful for certain tasks.) Big Blue has its own IBM Cloud, of course, and within that there is the Power Virtual Server that makes Power Systems capacity available around the world on a metered basis. But when it comes to AI, even IBM is partnering with Amazon Web Services, Microsoft Azure, and Google Cloud as well as some of the neoclouds that just focus on AI workloads like CoreWeave. It is natural enough, then, to ask what IBM’s role in this whole GenAI boom is.
“We help put AI into the enterprise, whether that is using WatsonX Orchestrate to connect together multiple sources, bring agents from different places and provide a control plane and governance,” Krishna explained. “It’s how they know what’s happening and also make sure data exfiltration is not happening. In addition to that, we help our clients orchestrate between multiple models. So we believe in a fit for purpose model strategy. We use foundation models where appropriate. We use smaller models where appropriate, be it from Mistral or our own Granite. And we also allow on premise deployment of models on Sovereign Core, so this way people can mix it back and forth where appropriate. And that’s our strategy to go forward on AI.”
It would be easy to get mad at Big Blue for showing very little interest in talking about how these technologies will be ported to Power Systems and Spyre accelerators. There is no reason this could not happen given that Red Hat Enterprise Linux certainly runs on Power10 and Power11 iron and the entire Red Hat stack can, in theory, run on these processors as well. It is a matter of porting it and certifying it – which should be a lot easier with Project Bob and its underlying Anthropic Claude model. We would normally get mad about this because we want customers running IBM i and AIX systems to be able to add the Red Hat AI stack and its Sovereign Core extensions natively on the architectures they clearly love.
But clearly IBM has other ideas or it would have already announced Sovereign Core on Power on Red Hat OpenShift and Enterprise Linux with all the WatsonX Orchestrate goodies. It is just easier to admit that IBM i and AIX shops have plenty of X86 systems wrapped around their systems of record, and many of them already have Dell gear at that. So the easy way out is to just go hybrid with the systems and focus on the AI.
We wonder if customers will like this, or follow. Time will tell.
What we do know is that when you engage with Big Blue, you best have the hybrid cloud and AI talk down and you should walk the walk, too, and then you might also be able to get some of the Power Systems stuff you think you need. Like native Sovereign Core and a native Red Hat stack on Power.
RELATED STORIES
The Big Easy: Connecting The Dots On Big Blue’s AI Strategy For IBM i
