FDE, the zeitgeist of Silicon Valley, is the real bubble: Part I
It's not datacenter capacity that is the real bubble — it's FDE capacity. Why consulting-heavy, forward-deployed models keep losing to self-service and design.
3 min read
The Palantir model is suited for US defense and the Department of War at large, but in civilian and enterprise segments, it is a captive-to-vendor strategy that is neither customer- nor partner-friendly.
Software and design cannot be replaced with messaging and consulting. To replace repeatable human work with even more engineers is a travesty!
The Yahoo! versus Google lesson
I remember the time when Yahoo! used to boast of big advertisers and publishers, big budgets, active human (manual) bid management, and as a corollary, a total disdain for Google's casual self-service ideology, which had heavily influenced the latter to design for the autonomous long tail. Overture/Yahoo! had a model that was built to extract maximum value from a finite set of big accounts; Google's was built to profitably absorb an effectively unlimited number of tiny ones. Both companies had products in ad-buying and in context-matching (pages to ads), with one of them heavily dependent on expensive hardware, high-end software, and human negotiation, while the other using commodity infrastructure and algorithms to achieve the same.
The same story, over and over
We all remember the early days of cloud between 2011-2015, when the large enterprise bought heavily from IBM SoftLayer, while the long tail was happy using AWS. Amazon had built an entire operating system culture to "stream" infrastructure to anyone and everyone, while the hardware sellers kept feeding a forward-deployed culture to build clouds. Sun Microsystems' Solaris had made the same choice with an army of consultants and administrators, while Linux was quietly serving those that pined for self-service. Siebel against Salesforce was the same battlecry — in fact, the entire movement of SaaS was a rebellion against complex (long-running) forward deployments.
At Nutanix, we had seen the same story unfold in front of us. Large consulting coalitions were formed between incumbents to advance private clouds: vBlock, FlexPod, PureFlex, Matrix, ExaLogic, and a myriad others that boasted of big hardware, big budgets, and big consulting to bring it all together. As Nutants, we championed 1-click, self-serve, and start-small — so-much-so that our bête noire, VMware, continued to be flummoxed by our anti-zeitgeist strategy.
What they all had in common
There is something in common between Yahoo!, Solaris, IBM SoftLayer, Siebel, VMware, and converged infrastructure: they don't exist anymore. More importantly, their over-promised and consulting-heavy cultures were superseded by algorithms, design, and a basic human need for self-determination. Consumers want self-serve, but so do enterprise users. What worked for Palantir in defense eventually does not work for either the large buyers or the mid-market and the long-tail.
True operating systems operate in all environments because they enable people to work autonomously, without being blocked on other people. Rest everything is a blip on the tech radar.
The real bubble
AI diehards think they can pour money at the frontier to buy this basic human desire for autonomy. And that is what is wrong with today's markets. It's not the datacenter capacity that is the real bubble, it's the FDE capacity...
Next in the self-service series:
Multi-tenancy: How long tail thinking provokes design for self-service.
