Follow the CAPEX: Keeping Up With The Clowns


IBM (left) and Oracle

Clown-1-1” (CC BY-SA 2.0) by americanbulldogbully007

Tl;dr: IBM is the Generalissimo Francisco Franco of cloud: still dead. Like the beleaguered consumer in these inflationary times, Oracle is spending more while falling ever further behind. A Trump Cloud?

The opposite of a cloud is a clown. We know this because CAPEX doesn’t lie.

Let’s pay another visit to the circus of perennial cloud pretenders, where platitudes are plentiful but CAPEX is scarce (previous visits: 2018, 2020).


Company motto: Got Financial Engineering?

Other than hacking up a big lump of their consulting business in the form of a hairball spin-off dubbed Kyndryl (not to be confused with Kyzyl, the capital of Tuva), not much has changed since we last checked in on IBM. The relentless march down and to the right continues. IBM’s cloud business is a “cloud” business.

IBM shifted into financial engineering overdrive with the spin-out. A worldbuilding frenzy resulted in an elaborate financial backstory for both IBM and Kyndyl, back to 2018.

They don’t seem to have manufactured or lost any CAPEX spending (unlike $2 billion in promised profits that – oopsie – vanished with the spin-out), but the allocation between IBM and Kyndryl gives us a little more detail on IBM’s CAPEX spending (or lack thereof). And, with the help of a magnifying glass, perhaps a glimpse of the level of investment into IBM Cloud. The 2021 numbers are full-year spending for the now separate companies while earlier years show the internal allocation within a singular IBM:

Global Technology Services is the division that was cast out as Kyndryl, minus random bits IBM couldn’t bear to part with, including “IBM Cloud IaaS” (I assume their PaaS services sit with Red Hat). IBM Cloud was moved to the Systems segment, now renamed Infrastructure. So IBM’s cloud infrastructure and the mainframe are finally united.

How did it take over a decade for IBM to start talking about a mainframe cloud? This is a company that has touted the mainframe as the place to run every imaginable and no matter how ridiculous new workload for decades (remember enterprise-grade Second Life?).

Given IBM Cloud was in GTS prior to the spin-out, we have a ceiling for IBM’s actual cloud CAPEX spending in 2018-2020. Total GTS CAPEX spending dropped from $1.45B in 2018 to $558M in 2021. Assuming this year’s Kyndryl CAPEX (KAPEX?) is closer to their baseline spending without a cloud to fund, it is hard to see even a solitary billion in peak annual CAPEX spend for IBM Cloud prior to that. Not that that is even remotely surprising, but the degree to which IBM has missed the generational transition to cloud computing remains astounding.


Company motto: The Ultimate Auditing Machine

Oracle, on the other hand, is at long last starting to put its money where its mouth has been (though the mouth’s lead looks insurmountable). The database vampire doubled their annual CAPEX spend in the last year (using their March 2021 through February 2022 results). The company promised to spend $4 billion in CAPEX in FY22 and it looks like will hit that mark by May. And they surpassed IBM in annual CAPEX spending this year.

Oracle may be investing in anticipation of being TikTok’s US data repository. Leaving aside the weirdness that Oracle “has not been party to the current negotiations”, hopefully it goes better than their Zoom windfall (and subsequent lack of capacity embarrassment).

Meanwhile in Cloud City

Now lets pan back and put the clowns on the same stage as the actual clouds:

Oracle’s CAPEX uptick edges them past IBM, but the two of them are still battling to differentiate themselves from the x-axis (I may never tire of that joke). Oracle’s lifetime CAPEX spend ($19.7B since 2001) is less than AWS spent in 2021 alone. Bottom line: IBM and Oracle continue to fall further behind in the cloud infrastructure race with every passing year.

New Clowns in Town

As yet another benefit of this free service, Platformonomics also tracks an emerging set of players – call them aspiring clowns – who talk up their cloud infrastructure ambitions but don’t appear to have the slightest understanding of the CAPEX required. They are blissfully unaware that the x-axis even exists.

Bank of America

Company motto: The Worldwide Leader in Financial Crime

Bank of America has paid over $82 billion in fines for various “banking” activities since 2000. They have been penalized as much as the next three miscreants combined.

It boggles the mind that kind of dominance has eluded antitrust scrutiny. It also boggles the mind to think how far that kind of money would go towards cloud infrastructure (that is about half of Google’s lifetime CAPEX investment).

While the rest of the financial industry are grudgingly accepting they are no longer the most sophisticated IT operators and distinctly subscale in the cloud era, Bank of America is headed in the other direction. In addition to a lot of “private cloud” handwaving and cloudwashing, they don’t believe just any public cloud will suffice and need their own. We have not heard any updates since their big announcement in 2019 with IBM (which is doubly inauspicious in both the time without a progress report and the choice of partner), but we can always check their CAPEX:

Hmm. Somehow B of A has gone a decade without any CAPEX. No bank branches, ATMs or data centers? Either banks play by a different set of financial reporting rules and/or there isn’t much going with their cloud efforts. Or we’ve stumbled across their next big legal fine.

With no spend at B of A and IBM in a CAPEX death spiral, there must be some serious 0 + 0 = 1 synergy to this collaboration. Imagine the board meeting where Bank of America decides IBM is the right company to partner with for public cloud. In the year 2019.

And lest you think IBM’s chicanery in this partnership is unreciprocated, IBM’s cloud pitch de jour is all about ensuring compliance. Yet those $82 billion in fines suggest compliance may not be job one at Bank of America. Or even top ten.


Organizational motto: You can’t spell bureaucrat without E and U

GAIA-X is a European industrial policy project, designed to protect Europe’s “digital sovereignty” in the cloud era. Spearheaded by Germany, it appears on par with other German sovereignty initiatives for energy, security, and economic growth (outsourced to Russia, the United States, and China, respectively).

After some fanfare and lofty rhetoric accompanying the initial announcement, GAIA-X is going about as expected, described as a “mess” and a “cautionary tale about the EU’s tech ambitions”. Not only have AWS, Google and Microsoft taken over the technical work, but they’ve let Huawei join to “boost Europe’s secure cloud services”.

Appealing as it may seem to government officials, we see yet again that product management is harder than it looks.

And there is no CAPEX, which means any European cloud is still some other continent’s servers.

Trump Media & Technology Group

Company motto: The Grift Never Stops

Rejected company motto : Execution Actually Matters
Rejected company motto : Competing with Big Tech is Even Easier than Building a Wall
Rejected company motto : Incompetence is Functionally Indistinguishable from Censorship

The Clown Heuristic™ looks at the gap between promises and actual CAPEX. I thought GAIA-X had the potential to be an “epic hyperclown”, but we’re in new territory as we move from the realm of overly optimistic corporate promises of the sort we’ve come to expect from Larry Ellison to unhinged populist politics ungrounded in any semblance of reality.

The Trump Media & Technology Group first intends to take on Disney, Facebook, Google, Netflix, and Twitter, and thereafter compete with the three hyperclouds plus Stripe (combined 2021 CAPEX approaching $150 billion).

With a mission to “counter the dangerous exercise of tech monopoly censorship”, building a foundational “Trump Cloud” to host their planned media businesses seems like a mandatory prerequisite. Building those apps without a platform to safeguard against “Big Tech” deplatforming seems like a risky strategy. Only a loser would do it backwards.

A big beautiful “Trump Cloud” would inject a yuge dose of entertainment into what is otherwise an increasingly mature and uneventful market. Just imagine the chunky gold font of the command line interface.

(IBM, if they’re not already actively pitching TMTG pursuant to their strategy of selling compliance-oriented solutions to customers who really don’t care about compliance, should offer any and all forms of assistance in the hope Trump Cloud displaces IBM from the center ring of the cloud clown circus.)

The level of snark demanded by TMTG is daunting (Matt Levine has taken a couple shots), but fortunately the endeavor largely speaks for itself. The pitch deck is quite a read, in particular:

Slide 21:

relates to The Trump SPAC Did a PIPE

Trump Cloud apparently doesn’t need infrastructure, just “structure”. And that experience from cloud powerhouses IBM and Little Debbie Snacks should take them far.

The Architecture slides (22, 23, 27) resemble the product of a barely trained neural net spewing out random web architecture concepts.

I am, needless to say, eager to see Trump Cloud’s CAPEX spend once the SPAC is completed.

4 responses

  1. Alison N Obrien Avatar

    I read this column regularly but seldom comment, typically because I am insuffiently technical. I’m a comm person. But this is one of the best informed really snarky columns you have yet produced.

  2. ‘@Alison – Thanks! Tragically Snarkonomics is taken.

  3. Alison N Obrien Avatar

    What a shame, that should have been yours. Saw an old conversation with you and Bobmu and he closed it with a comment about your snarkiness. This was on databases and his interest in Relational AI. aob

  4. Heck, I forwarded this to a friend before reading these comments and said “set volume to maximum snark”. What I like about your writings is, snakiness aside, you do an extremely effective job of showing how certain top execs are completely full of BS when it comes to their proclamations. In this one the peeling off of Kyndryl capex which was basically just buying servers for their clients under long term outsourcing contracts showed how woeful the capex situation is (and was).

Get Updates By Email
%d bloggers like this: