The Reversion of IT

Rob Thomas
4 min readJan 23, 2020

The struggle between centralization and decentralization is at the core of American history. -Anthony Gregory

Information Technology (IT) spending growth has hovered around 3% growth per year, over the last 20 years. There are cycles, but in aggregate, that has been a reliable metric. At the same time, IT as a % of GDP has been relatively constant (defined as Information and communication technology expenditures) in the 5–7% range. It takes a lot of spend to really move the needle on % of GDP.

I believe we are at the start of accelerated expansion in IT (both in terms of growth and as a % of GDP). A recent Bernstein report indicated a similar view:

The key point here is that total IT spend is accelerating, and IT spend excluding Cloud (which is defined as pure public cloud) is driving much of that increased spend. In part, this is driven by larger addressable markets (i.e. IT is no longer available to just the largest companies in the world). But, I believe there are two underlying factors, which may not be as obvious:

  1. Decentralization
  2. Open Standards

If there are 2 things that have remained true throughout the last 50 years, it is the fact that IT wants to be decentralized and open.


Nearly every industry has seen a modulation between centralized and decentralized architectures over time. This is about economics. Take linear TV as an example. The first television stations appeared in America in the late 1920’s and 1930’s and individual stations operated for much of the next 40–50 years. This was a decentralized architecture. Around 1980, ESPN, CNN, and MTV launched and the cable bundle began to take off (centralized architecture) and has only begun to unravel recently with the launch of Netflix, Disney+, and other OTT alternatives (a decentralized architecture). As each decentralized alternative gains steam, consolidation will be inevitable, and we will then be back to a centralized architecture.

In a centralized architecture, a few actors will feast on the profits afforded in that model. Over time, the participants are inclined to seek alternatives, which drives the reversion to the mean. It’s all about the money.

Let’s turn to IT. For much of 1960–1980, mainframe was the dominate architecture in IT; very much a centralized approach. In the late 1980’s, IT began to decentralize with client/server, and then accelerated with the internet and mobile. The dawn of the public cloud era is a return to a centralized architecture: walled gardens, proprietary capabilities, with the landlords extracting maximum profits. This too will revert back to a decentralized architecture, and you can see it in things that are happening with containers, Edge computing, and AI right now.

TV and IT are just two examples, but there are many more: currency (fiat vs. crypto), retail (Walmart vs. e-commerce). As centralized landlords extract profits, the market seeks alternatives. This is, in part, what is driving the optimistic growth outlook described in the Bernstein report.


The notion of open is often misconstrued. It could mean open source, but is not necessarily confined to open source. Let me clarify the distinction. The internet as it exists today, is largely a product of TCP/IP, which is a set of open protocol standards, that are freely available and developed independent of any specific hardware or software configuration. That openness enabled wide spread adoption and interaction, which propelled the internet forward. This was accelerated further by things like W3C, which were standards for an Open Web Platform (think HTML, XML, etc.) An analog example of open standards would be things like electrical plugs, which are typically standard (at least within a country). The open architecture and standard approach enables accelerated adoption, which in turn has second order effects on innovation.

Open Source is different, but in the same zip code. Open Source refers to how software itself is developed, which is best captured in these ten principles. Open source changed the traditional software model, which used to be exclusively about a single organization developing proprietary code, and then dictating the terms (price and rights) for how that could be used. While that model continues in full force today, the rise of open source has augmented it, driving speed, re-use, and proliferation of software at a never before seen rate.

But, open is not just about standards and software. It’s an ethos and an approach to how technology is deployed and used in any company. And for that reason, IT will always drive towards open.

Jeremy Grantham has risen to prominence as an investor over the last 30 years, with over $100 billion in assets under management at GMO. His investing philosophy is based on a simple observation: markets always revert to their long term averages. Grantham calls this phenomenon Reversion to the Mean, and GMO will often take positions betting on this phenomenon, when asset prices deviate from their historical ranges.

This phenomenon also holds for IT, which will always revert to open and decentralized architectures. Even though we know this to be true, sometimes the hardest thing to predict is the timing.



Rob Thomas

Author of ‘The AI Ladder’, ‘The End of Tech Companies’ & ‘Big Data Revolution’