Duplicate » admin by request

The Concentration of Power: From the Internet to Identity Fabrics

Older man with gray hair and beard wearing a striped shirt and black glasses, speaking and gesturing while seated indoors.

In the mid-1990s, the arrival of the internet was hailed as a revolutionary moment for humanity. Ordinary people would gain access to the world’s pool of knowledge, and a new age of enlightenment beckoned, one where information was finally free to everyone.

Many of the pioneers behind it were driven by ideals rather than profit. Tim Berners-Lee gave the World Wide Web away. Stewart Brand and the creators of the Whole Earth ‘Lectronic Link (the WELL) built community on an explicitly altruistic ethos. The early days of Wired, and Steve Jobs cutting his teeth as a phone phreaker, belonged to the same period: independent, idealistic, suspicious of incumbents by instinct.

After three decades, the internet is dominated by a handful of massive corporations and their billionaire founders. Individuals, with no small irony, have traded their privacy and their data for “free” search and “free” apps. The iPhone, once a genuine wonder, is now most often used to stare blankly at an endless stream of advertising and trivia.

The security market is following the same path

As an analyst for ten years, my job was to watch cybersecurity market trends and rate vendor solutions. I noticed the market consolidating in much the same way the consumer internet did, with the tech giants bundling every facet of security into their platforms. That doesn’t mean it’s what buyers want, but then, nobody asked for Meta, Amazon or Google either. This is done by aggressive acquisition (Palo Alto swallowing CyberArk) or in Microsoft’s case, trying to bend the market to their will.

From this summer, Intune Endpoint Privilege Management will be folded into the E5 bundle. The subscription rises by $3 per user per month — from $57 to $60 — and Microsoft is loading in a whole tranche of previously paid add-ons to justify it: Security Copilot credits, Cloud PKI, Enterprise App Management, and EPM.

The prevailing narrative is that this is a pricing play, Microsoft undercutting the specialists. It isn’t. Nobody buys E5 for the EPM; they buy it for everything else, and the EPM arrives in the box like the complimentary toothbrush in a hotel room, so deep in the bundle that it isn’t even the feature Microsoft leads with to sell the increase. It isn’t chosen or considered, merely present, and for most buyers that is enough. Usage will be low. Microsoft are hoping that E5 subscribers will think they have EPM covered and look no further. We are here to ensure that doesn’t happen!

Identity fabric: smoke dressed as architecture

The concept of the “identity fabric” is frameworks-as-architecture in its purest form. Having watched these frameworks gain traction inside the analyst world, I can tell you the identity fabric remains more diagram than solution. Its function is not to secure anything. It is to lend a semblance of coherence to a patchwork of acquired products, and to foster the comfortable orthodoxy that the concentration was the plan all along.

It also, by happy coincidence, lets consultants draw highly complex diagrams (topologies, in effect) and charge lucrative advisory fees for the privilege. The pitch reduces to this: here is all the stuff you’ve already installed, and we’ll advise you on stitching it together better. There is nothing scientific about the identity fabric, but that hasn’t stopped it spreading like wildfire through the vendor and analyst communities, both of whom benefit.

And then the contradiction nobody sits with. The analyst firms tell you the identity fabric is how you weave your many separate tools into something coherent. The monoliths tell you they’ve already done the weaving: buy the platform and the fabric comes off the shelf. Both call it identity fabric. But they are selling opposite things. One is glue for a mess, the other is a reason the mess shouldn’t exist. They cannot both be right, which rather gives the game away. “Identity fabric” isn’t an architecture. It’s a label that means whatever the person invoking it happens to be selling.

The incumbents have an answer. The answer is the problem.

It would be lazy and wrong to say the incumbents have no answer for AI agents. BeyondTrust, CyberArk and Delinea have all responded. CyberArk shipped a Secure AI Agents product in late 2025 and now describes identity security as the “control plane” for the agentic era. They have an answer. The answer is the problem.

The approach treats the agent as a new identity class — discovered, inventoried, credentialled, governed with the same privilege machinery built for human and machine accounts. It is a genuinely impressive extension of what they already sold, and that is precisely the tell. When identity is the only tool you own, every new problem is reframed as another identity to manage. The agent is handed a name badge and added to the directory.

But an agent is not a who. It is a what, a transient thing, often alive for seconds to perform a single action before it vanishes. Issue it a badge and you are governing a label, not an act. The control fires after the agent has already moved, or it watches the identity while the action itself goes unseen.

One event makes the point better than any argument. On 11 February 2026, Palo Alto Networks closed its acquisition of CyberArk for roughly $25 billion — the largest deal in the history of the security industry. Palo Alto’s chief executive framed it as securing “every identity, human, machine, and agent,” and promised customers the end of identity silos. Read that again. The clearest strategic answer the category’s leader had for the next twenty years was to be absorbed into a larger monolith, which promptly declared identity the universal frame and the agent just one more thing to file under it.

Concentration, in other words, is not a risk the incumbents are warning you about. It is the thing happening to them — and they are selling you the framework that makes it look like strategy.

The AI Agent breakthrough: governing the action, not the actor

What the quadrants will never tell you is that the genuine breakthrough will not come from the incumbents. It will come from whoever stops trying to govern the actor and starts governing the action. Once the actor is a piece of software spun up for nine seconds to do one thing, identity is no longer a control surface. It is a label attached to something that has already gone, which, stripped of the cosmology, is all identity ever was.

This replaces the existing model rather than adjusting it. And paradigm shifts almost never come from incumbents, whose entire structure is built around the assumption the new model breaks. Adopting it would mean admitting that the thing they sold for twenty years was the wrong centre of gravity.

Which returns us to the backroom heroes. The internet’s pioneers, the people who built the protocols and the products before the monoliths arrived to enclose them, did not lose because they were wrong. They lost because the people drawing the maps had every reason to keep the map as it was, and the resources to redraw anything that threatened it.

The same maps are being drawn again, by the same kind of people, for the same reasons. The agent era will not be secured by whoever owns the largest directory. It will be won by whoever is willing to say, out loud, that the directory was never the point, and that by building on a flawed model of identity, the monoliths are doing their own customers a profound disservice.

A working analyst would never tell you that. A former one just did.

About the Author:

Picture of Paul Fisher

Paul Fisher

Paul Fisher is Head of Global Strategy at Admin By Request and has written about enterprise IT since 1991, first as a journalist and editor, then as an analyst. As Lead Analyst for PAM and IAM at KuppingerCole, he wrote the firm’s benchmark PAM market research and moderated the PAM track at the European Identity and Cloud Conference. He’s a regular speaker at major identity and cybersecurity events, and he advises organizations on PAM, IAM, and security platform implementations.

Author

  • Paul Fisher

    Paul Fisher is Head of Global Strategy at Admin By Request and has written about enterprise IT since 1991, first as a journalist and editor, then as an analyst. As Lead Analyst for PAM and IAM at KuppingerCole, he wrote the firm’s benchmark PAM market research and moderated the PAM track at the European Identity and Cloud Conference. He’s a regular speaker at major identity and cybersecurity events, and he advises organizations on PAM, IAM, and security platform implementations.

    View all posts Head of Global Strategy

Share this blog to your channels: