I commented some time earlier on the new trend that some companies are adopting, of going back to earth from the cloud. That was about the choices that face individual companies today, and the tradeoffs of being in the cloud versus on-premise.

Today’s essay is about something else. It is about the inevitibility of the cloud, and the diminishing space for on-premise. 

Now, especially in the Indian BFSI sector where I do most of my work, it is considered a scarilegous to say that cloud is inevitable so I will try and clarify my thinking on this. 

In any major technology shift, the old continues to retain (for a while) some advantages over the new so people often point to these advantages and hold on to their existing older platforms. In my own working life, I’ve seen many such large moves

• company owned premises to shared data center
• private networking to internet
• mainframe to client-server to web-based
• and so on…

These were big moves. On-premise once meant inside a company’s premises (often proudly visible from the reception area) — today hardly anyone would try to build and run their own data centers unless they have genuinely exceptional needs (such as the National Stock Exchange) or huge legacy footprints, like HSBC. Some 60% of the world used to be on-premise data centers not so long ago, its down to 40% and shrinking. Even HSBC has announced a move to AWS. 

Private networks were once the mainstay of businesses — today most use the Internet as their communication backbone, talking to their customers, their suppliers, internally to employees on Slack or Teams or Gsuite (even Board meetings are held on Zoom). The ubituitius Cisco IP phone is hardly visible now. 

The point is, each of these older technologies had their advantages, but were overwhelmed by the progress of the later ones. As more people moved, suppliers of the old technologies moved away, partners died or shifted sides, employees with relevant skills could not be found and so on. It did not really matter what the advantages were, the world shifted and the ecosystem shifted along with it.

This happens in every change. You need to move or be left behind in an expensive island of uniqueness. Those advantages of the old  system— either the new system eventually provides them or they are forever lost ,  displaced by new advantages. A horse has very few advantages over a modern automobile, and the ones it does (cheaper feed fuel, etc) aren’t enough to allow it to remain in the market except in pockets such as Matheran.

Netflix shipped its last DVD in September 2023. If you are a fan of the DVD today, a tough world faces you — it’s nearly impossible to buy a cheap player, DVD stores are shut, new movies aren’t even released in the format any more. The advantages of the coveted Blu-Ray against streaming aren’t enough to let it remain a viable option for the masses.

This is what is going to happen with the cloud. There isn’t really a rush of people departing the cloud (as the revenues of the hyperscalers show). The trend is clearly headed one way — into the cloud — its just a question of holdouts, laggards and leaders. Occassionally a company goes the other way for its own reasons but there’s no sign of a rush for the exits.

What does this mean for an individual company? Being on-premise relies on the following major things

1. Availability of harware on-premise
2. Avalability of datacenter space
3. Availability of people who know how to buy, operate and manage the environment
4. Partners who offer on-premise versions of software
5. Networking and security experts who know how to connect on-premise to the rest of the world. 

In all these, the market is shrinking so there are fewer and fewer players. There was a time when ten companies were selling servers; today even the minimum requirement of three can’t always be met. HP reports falling server sales, while Dell servers and networking revenue has barely grown past its peak in 2019. Lenovo similarly reports a fall of over 17% in 2023 due to companies buying less. The hyperscalers now build their own servers, so Dell, Lenovo and HP are winners in a shrinking pond. Like the Economic Times puts it — as action moves to the cloud, server sales take a hit.

The data speaks rather unambiguously — companies are buying less on-premise and moving in larger numbers to the cloud. 

Even on the datacenter side, hyperscalers have been gobbling up most of the capacity at a faster rate than ever. As per the Synergy Group, nearly 40% of global datacenter capacity was with hyperscalers in 2022, and this was growing far faster than the others. 

And its not just that. Software OEMs are headed the SaaS way in droves, and often reluctant to support on-premise versions beyond a point — FinOne, the dominant loan management platform in India, is doing just that and so are many others. According to Gartner, global spending on SaaS is projected to mark a nearly 50% increase in 2023 compared to the expenditure in 2018, whereas a 7% decline in spending on on-premise solutions by 2024 is anticipated.

Then there are the skills. All of you know first hand how difficult it is to get a storage or networking expert for on-premise setups. Most companies have already given up and outsourced it. 

The timeframes in which technology change occurs is also speeding up. VHS ruled for thirty years; DVD barely managed fifteen. It took over thirty years for shared datacenters to become the dominant paradigm; it will take much less time for the cloud. 

Individual companies choosing to hold back or reverse course must introspect well — there is a real risk that they will be stranded with vintage tech. Are they really going to gain enough to make it worthwile? The argument isn’t really about right or wrong, its about changing with the times or being left behind. It takes a company at least a couple of years to move efficiently to the cloud, which really shortens the runway for this decision.

I still have a box full of DVDs, so of those films are impossible to get on streaming. Sadly, I no longer have a way to play them.