You're not hallucinating: Generative AI is helping IBM's mainframes grow Generative AI's powers extend to helping the ancient concept of a proprietary enterprise OS and hardware stack to thrive, if IBM's Q2 2024 results are any guide. Big Blue on Wednesday announced [PDF] quarterly revenue of $15.8 billion – a two percent jump that would have been four points if not for exchange rate shifts. …
"That may shrink cloud bills, or with Red Hat and on-prem tech shrink bills further still."
IIRC, the initial hype around The CloudTM was that it would make your on-prem IT less expensive. Highly paid consultants in suits were hyping CEOs everwhere (well, essentialy at Fortune 1000 companies where the business meal is still a notion), chanting about how The CloudTM would bring vast economies compared to all those expensive admins.
So, it looks like we're back on track to talk once again about how on-prem is lowering bills.
And the wheel ever turns . . .
My colleagues, a grumpy North Londoner nearer to my Dad's age than mine opined back in 2011 that at scale the cost model with Cloud would always be more expensive than your baseline on-prem workloads. Because they are buying much the same kit and licenses and then applying a markup.
Now, if you have very "bursty" workloads, cloud costs may well work for you. But if you run at scale and at a fairly even tempo then on-prem is your friend. And always will be.
The centralisation / decentralisation of computer hardware is an ongoing cycle that has persisted 50 years.
After selling centralised mainframes to corporations and service bureaus, the manufacturers spearheaded decentralisation by selling minicomputers to departments. And then PC-class machines to individual business units.
Then they invented enterprise computing and sold server-class machines to the corporations.
Now they have invented cloud computing – which means they sell mainframe-class machines to the new generation of service bureaus.
The next wave of selling direct to the corporations is now upon us.
What next? I predict puddle computing – powerful department-level machines that cache and process the output from the clouds.
Why not? Highly paid consultants will prove that agile puddling is the cheapest option.
Consulting revenue grew by two percent, which is less than IBM wants. Krishna attributed that to "factors such as interest rates and inflation impacted timing of decision making and discretionary spend" – issues that have been present all year.
He's not going to admit that consulting depends on lots of people and that at least some of them need to have long years of experience nor that IBM's core HR competence is getting rid of such expensive items.
One of the points I would have made in this article would be to offer some historical insight into why "Investors weren't offered guidance" for next quarter or through the end of the year?
"Z series revenue is very cyclical: it peaks when a new box debuts and then tails off a year or two later." Yep, this is a statement of the obvious . . . but IBM can also vary its revenue by several percentage points in a single quarter by pulling in revenue from out quarters and years. IBM executives have been playing this game since Louis V. Gerstner took charge (well, actually John Opel and John Akers)
Unfortunately, the only way IBM can do this is at the expense of "long-term" profits: by offering several very large customers extremely lucrative long-term deals to pull in their renewals for "short-term" revenue and "shorter term" profit gains.
Of course, IBM's customers know this better than anyone and the smart ones "play the game."
Anyone agree or disagree?
You can't do that! Revenue recognition is a sensitive topic, and any salesmen worth his salt will want as much revenue recognised as early as possible -- those yachts have to be paid for. But there are severe penalties for recognising revenue too early. It's true for hardware you can claim it once the box is in the loading bay at the customer's site. But it has to be there. You can't say it's in the post. Salesmen and their managers expend great effort to ensure revenue can be recognised correctly, and the quarterly deadlines are very busy times -- but if they fuck it up plod may come a-knocking.
In short, recognising any revenue too early is fraud.
Ah RegGuy1, good to see the discussion.
Hardware and software have different "rules of the game" eh? Software can be gamed with ELA's and hardware, well yes it can be gamed too. Examples follow:
IBM has a very long history of gaming enterprise license agreements--going back to at least the 90s. I sat in the rooms when the negotiations took place. Ask any very, very large customer--the ones who, with a single deal, can move IBM's quarterly or yearly software revenue such as A.I. on "mainframes" by several full points. Also take a look back in history at IBM's transition from a "lease-rental" hardware customer base to "purchase" mainframe base. Opel took the "cash cow" then, which happened to be IBM's mainframes that were on lease and converted them to purchase at such a discount that our salesmen of the time were screaming --- "Wow! I don't even have to sell this; it is less than a two-year break even point. I just put the contracts on the table and the customer sees it!"
Based on that Opel promised the 100 billion and 180 billion IBM -- as if the conversion curve would never run dry. Essentially he was pulling in all the profits from decades into his time in the corner office. When he left the corner office, the recipient should have been wary. Unfortunately for John Akers, he had to admit he couldn't make the 100 billion IBM in his first press conference--much less the 180 billion--as the revenue/profit hardware well ran dry. His career/legacy ended on day one holding the bag for Opel. Don't feel sorry for him though as he wasn't deaf and dumb and most likely knew what was going on in the board rooms as they were implementing a policy that Thomas J. Watson Jr. fought for decades--converting the lease base to purchase. When the government forced IBM to do it -- offer purchase as well as lease-rental -- the annual report documents that very, very few customers didn't want it. They wanted IBM to carry the "obsolescence factor" in hardware, not themselves.
Also, having been part of Software Group, I saw our "profits" trimmed tremendously to get customers to take our boy's hardware that was falling behind along with our boy's consulting which was almost unsellable at the rates they were charging. Ah, it looked like hardware was doing gangbuster while software was carrying the load. It is so very easy to move revenue and profits around so "A.I." is doing gangbusters, eh? Legal ... surely. Ethical ... not in my opinion. Make each division perform at its top with a little give and take, but not playing to the stock market where "A.I." is the buzzword.
IBM has become very good at doing what is legal over the last few decades to avoid the plod (one example is its worldwide pension plan raiding which "the plod" should have investigated but never did because it was legal but the height of unethical -- the plod might care about the unethical but they can only press onward with the illegal, eh?)
What is interesting to me is that what the feds took IBM to court over (forcing it to make purchase an option to rental-lease) decades ago is now the basic practice of most corporations--at least in software. Most of my software I can only get on yearly renewals even though companies like Adobe don't do much of anything to earn the additional money with enhancements and investments. Legal? Absolutely. Ethical? Not in my opinion.
Cheers mate, looks like we might have a great discussion.
Any others care to join in?
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
