There’s not a CFO in the world that isn’t talking about, thinking about, or paying for something called ‘cloud.’ And there’s not a technology concept more misunderstood than cloud. In order to predict to impacts of cloud for the Future Office of the CFO, it’s important to consider trends and developments of the past. In this column and the next, we’ll discuss the history of the Cloud from the CFO’s perspective, diving into uses, costs, and risks.
It Started With Mainframes
First, some history, because it’s important to understand context when advocating for change[1]. Once upon a time, firms built or retrofitted 5,000 sq ft to 50,000+ sq ft spaces full of air conditioners, water chillers, 400Hz power supplies, HALON fire suppression, and raised floors to house their mainframe computers. Those pioneers spent $5M (1970 dollars: $35M today) in Leasehold Improvements to house $5-$10M ($35M-$70M today) of mainframes, disk and tape drives, printers, and so on. Real-estate costs are, of course, extra and vary greatly. I worked at full-floor Datacenters in Wall Street area office towers whose cost would take most people’s breath away. Mainframes in the 1970s-1980s were hugely expensive, and both the investment decision (should we invest?) and the financing decision (how should we pay?) consumed a lot of the CFO’s attention.
CFOs also had to become intimately involved in complex ‘IT deal structuring’ activities. Suppose a firm didn’t want to commit to significant Capex investments. In that case, they could turn to Facilities Management or Outsourcing firms like EDS, or leasing firms like Comdisco, to optimize costs and cash flows. Lots of decisions were needed, most of them outside the skillset of the CIO!
From the Raised Floor to the Office
Over time mainframes got cheaper and easier to install (air cooling, standard power) and 5x-10x smaller. But upgrading a mainframe or adding another one still cost $10M-$20M in today’s dollars—still big enough to need the CFO’s involvement. Then startups like Digital Equipment Corp (DEC) and Data General arose. Their ‘minicomputers’ differed from mainframes in two important ways: they ran perfectly well in office environments rather than requiring raised floors, and you could buy one for as little as $250K (today’s dollars). Suddenly computer investments weren’t ‘Capital Committee’ agenda items: they were LOB ‘equipment’ budget items. And so began the ‘decentralization of technology’ (if you’re a business executive fed up with bureaucracy)—or the start of ‘Shadow IT’ (if you’re a CIO or CFO trying to retain central control of ballooning IT costs).
Each succeeding ‘generation’ of technology (PCs and PC networks, client/server computing, today’s cloud computing) further lowered unit costs and transaction size and thus further decentralized technology acquisition and maintenance. Lower prices/smaller transactions sound good to CFOs. But the corresponding loss of control sometimes more than made up for the apparent cost savings, as business units made ever-larger investments in ‘departmental technology’ (for good and for bad—a topic we’ll dig into in a future column).
As you see, a CFO’s involvement in the financial engineering of IT investments has lessened, but as firms have increasingly come to rely on IT, their total IT spends have skyrocketed (along with the firm’s productivity, quality, product features, etc.). The nature of that spend has significantly changed. Whereas 40 years ago, a CFO got dragged into IT’s request for a new mainframe every year or two, today IT and business units have IT spend embedded into almost every proposed project that comes before the CFO (a feature of The Acceleration Economy).
Make Way for ‘The Cloud’
For the last few years, CFOs have had to deal with something called ‘The Cloud.’ It’s touted as revolutionary, as transformational—in fact, as the very definition of ‘Digital Transformation’ (see my many videos & articles on the subject to learn why that’s nonsense). CIOs, CTOs, and other C-Suite execs beg the CFO to sign off on ‘Cloud Investments’ that they promise will right all wrongs, fix all problems, and make the firm ‘Digital’ (whatever THAT means). The trouble is, every one of those supplicants has a different definition of cloud and its costs, benefits, and risks.
I’m here to provide some clarity for CFOs (and CEOs) faced with big technical decisions that sound sensible but may be misguided. OK, here goes (N.B., if your BUSINESS involves selling a cloud-based service like LinkedIn or Netflix, or on operating at a truly massive scale like Amazon, this is NOT the article for you. My audience here is the CFO of an ordinary ‘user’ organization).
I’ll start by quoting something I heard Oracle’s founder, Larry Ellison, say about 10 years ago at Oracle’s annual OpenWorld conference: “The cloud is just someone else’s computer.” True, but it doesn’t sound especially useful for decision-making…until you consider the often quite different accounting treatments of Capex vs. Opex (more about that next time).
Is it ‘Hyperscale’ or is it ‘Fake’?
As this part concludes I leave you with a thought: I make a fundamental distinction between two classes of clouds that will anger some serious nerds who demand technical precision—but I’m writing this for CFOs, not nerds. When dealing with the many ‘Cloud Projects’ that get talked about by IT and business executives, remember two classes of cloud–each of which has several sub-classes–‘hyper scale cloud and ‘fake’ cloud. Hyperscale cloud is a massive assemblage of computing, storage, and networking hardware and software operated by vendors as a utility by means of tools and processes that allow customers to rent capabilities elastically (that is, demand can grow and shrink as needed).’ Everything else called ‘cloud’ by vendors or your IT folks is fake cloud. And while a CFO might want to fund a fake cloud project if the need is real and the economics pan out, it’s important to understand that fake cloud is almost always a short-term Band-Aid vs. doing a (much harder) cloud transformation
In Part 2 we’ll talk about the three layers of Hyperscale Cloud (IaaS, PaaS, SaaS) and evaluate various fake cloud approaches designed to confuse decision-makers.
References
- If you aren’t familiar with ‘Chesterton’s Fence,’ here’s an article: Chesterton’s Fence: A Lesson in Second Order Thinking (fs.blog) ↑
