What 188 Sessions Tell You About Explore 2026

I have spent the last few days working through the Explore 2026 catalogue properly, all 188 recorded sessions, sorting them into themes rather than the four official tracks.

Doing that turns out to be more revealing than watching any individual session. The shape of a conference programme is a statement about where a vendor thinks the conversation is, and about where they would like it to be. Sometimes those are the same thing. Sometimes the gaps are the interesting part.

What follows is the analysis first, then the full catalogue at the bottom so you can work through it yourself.

The shape of it

Broadcom split the catalogue into four tracks. Cloud Infrastructure carries 127 of the 188 sessions, which makes it less a filter than a synonym for “most of the conference.” So I regrouped by subject:

AI and security together account for 55 sessions, close to a third of the programme. That is a fair reflection of where the market conversation actually is, and nobody should be surprised by it.

The bottom two rows are the ones I keep coming back to.

Five sessions on upgrade and migration

Out of 188.

Consider who is in the room. A large share of the installed base is on 5.2.x or 8.x, facing a major version transition with a new installer, a changed lifecycle model, a different identity architecture and an unfamiliar operations stack. For most of those organisations, the upgrade is the single largest piece of infrastructure work in their next two years.

Five sessions.

To be fair to Broadcom, the ones that exist are good. The path-selection session on choosing between upgrade, converge and import is the most useful hour in the entire catalogue if you have a 5.2.x estate. But five is thin against the size of the problem, and it means practical migration knowledge lives in documentation, KB articles and partner heads rather than on a stage where a thousand people hear it at once.

The counter-argument is that upgrade content does not fill a keynote hall, and conferences are marketing events as well as technical ones. That is true. It is also why the imbalance is worth noticing rather than resenting.

Seven on design and architecture

The Design and Architect track exists, and it is small. Sessions in it drew single-digit view counts, which says more about scheduling and discovery than about content quality.

This one bothers me more than the migration gap, because design is the discipline that determines whether any of the rest of it works. You can attend twenty sessions on individual features and still produce a bad design, and most of the expensive failures I have seen were design failures rather than product failures.

There is a reasonable argument that architecture is better taught in workshops than breakouts, and the new small-group interactive tracks introduced this year support that. But a catalogue where design is 4% of the programme sends a signal about what the industry currently values, and I am not sure it is the right one.

What people actually watched

Second was the vSphere deep dive at 88, the reliable perennial. Third, and more interesting, was the hardware crisis session at 47 — the one about extracting more from constrained infrastructure. The memory tiering economics session performed well above most feature content too.

Read those together and you get a picture of the audience that does not quite match the picture of the programme. The catalogue leans toward what is new. The viewing leans toward what is expensive.

The session that deserved better

Twenty-three views for the topology-aware CPU scheduler deep dive.

That session covers the engineering behind up to 30% higher SQL Server throughput, roughly 26% on Oracle, peak PostgreSQL latency halving in VMmark, and the SAP HANA certification at two VMs per socket. If you do capacity planning or sizing work, it is the most immediately monetisable hour in the catalogue, because it changes consolidation ratios on hardware you already own.

Twenty-three views. Meanwhile the announcement blog covering the same material was read by orders of magnitude more people.

There is a lesson in that for anyone producing technical content, myself included. Engineering depth does not travel on its own. It needs the announcement to carry it, and the announcement is what most people will ever see.

What the programme says about direction

The platform story has consolidated. Two years ago you would have seen separate tracks for the component products. Now it is one platform with capabilities, and the session titles reflect that. Whether you find that clarifying or reductive probably depends on which component you specialise in.

AI has stopped being a separate conversation. The AI sessions are not off in a corner discussing possibilities. They are about GPU scheduling, tenant isolation, memory economics and governance — infrastructure problems with AI attached, which is the right framing and a noticeable change from last year.

Cost is back as a first-class topic. The hardware crisis and memory tiering sessions did well because people are being asked to do more with hardware they cannot easily buy. That is a genuine shift and it will shape the next two years more than any product announcement.

If you only watch three

The full catalogue

All 188 sessions, grouped into the nine themes above, with session codes, speakers and view counts. Click any theme to expand.

How to actually work through it

The failure mode with a 188-session catalogue is predictable. You bookmark twelve, watch none, and by the time you get to them the release has moved on.

Pick three. Put them in the calendar as real appointments rather than aspirations. Then book an hour afterwards with whoever else on your team watched, with one question on the agenda: what changes in our estate because of this? If the answer is nothing, that is a legitimate outcome. Just say it out loud, because “nothing changes” is different from “we never got around to deciding.”

View counts are early on-demand figures from the public video library and will have moved since. Session titles, codes and speakers are as published by Broadcom. Sessions are findable by code at the official VMware Explore video search.

Views expressed here are my own.