Follow the CAPEX: Cloud Table Stakes 2018 Edition

After recently receiving a non-ironic lecture on Twitter that CAPEX investment is the best tell for cloud competitiveness, I’m starting to think Follow the CAPEX™ is building a following. Beyond the circular wisdom of Twitter, awareness is growing of the importance of CAPEX as both a leading indicator of the ability to compete at hyper-scale and also confirmation of customer success. Tech earnings press coverage now regularly highlights CAPEX investments.

Despite becoming perhaps uncomfortably mainstream, here are the CAPEX numbers and some commentary for 2018 (see previous 2016 and 2017 installments for the hyper-scale clouds as well as the hyper-scale clowns). There are no surprises, just up and ever further to the right.

The three hyper-scale public cloud companies – Amazon, Google, and Microsoft – spent over $68 billion on CAPEX between them in 2018. That is not all for cloud infrastructure (and maybe not even a majority of it), as those companies have other, material CAPEX spend, but it is directionally indicative. All three companies again registered all-time CAPEX highs this year. The combined total is on the order of 5% of total non-residential fixed investment in the US (one wonders whether hand-ringing about relatively disappointing fixed investment across the broader economy is attributable at least in part to the economies of scale of the public clouds).


In 2018, Amazon spent $27.6 billion, Google $25.1 billion, and Microsoft $15.8 billion on CAPEX (including capital and build-to-suit leases for both Amazon and Microsoft). Those are year-over-year increases of 19%, 91%, and 39% respectively (Google went nuts on CAPEX in 2018).


The three companies’ cumulative CAPEX spend since 2000 is over $270 billion, with over $116 billion of it in the last two years.


After being almost identical last year, we see a big spike in CAPEX as a percentage of revenue at Google to over 18%.


Last year’s commentary on Amazon still applies:

The top line Amazon CAPEX number is pretty useless at this point for insights about AWS infrastructure spending. There are just so many other things consuming CAPEX – distribution centers, warehouse robots, grocery stores, cargo jets, office buildings, biospheres, a hyper-segmented line of talking alarm clocks, etc. – that AWS infrastructure is almost certainly a minority of the total company CAPEX.


This year I broke out both the capital leases and the build-to-suit leases in addition to the general CAPEX numbers. Historically, Amazon’s CAPEX numbers have overstated their CAPEX relative to their competitors because it has included capitalized software development costs. But Amazon has stopped this practice and now says this is “not significant”.

My theory has been the capital leases are the best proxy for AWS spending, as the lease term corresponds nicely with the useful life of servers. Amazon’s CFO on their most recent conference call lends some strong confirmation to this theory:

And then on capital, especially infrastructure capital, if you use the capital lease slide as maybe an indicator for what we have invested into our AWS business to support infrastructure and global expansion, that number grew 10% last year, when it had grown 69% in 2017.

We can compare the capital lease expenditures and AWS revenues, and the numbers line up reasonably well:


The build-to-suit leases are for buildings. These can include office space, warehouses and retail outlets, as well as datacenters. My guess is only a fraction of this expenditure goes towards data center buildings, but I have no hard data.



Google obviously was offended when I wrote this about Amazon last year:

But at a company level, the spend is staggering and is now bigger than any US company I can find except AT&T (and Amazon should pass them this year assuming very low double digit CAPEX growth).

While Amazon did pass AT&T in CAPEX spend in 2018 (there evidently are real efficiencies to only building a fake 5G network as AT&T’s CAPEX actually declined in 2018), Google nearly doubled their CAPEX spend from 2017 to over $25 billion, and almost caught up to Amazon. They bought a huge piece of real estate in New York City, but that was still a mere $2.4 billion. Google’s incremental CAPEX spend in 2018 is about what cloud pretender Oracle has spent on CAPEX throughout their entire history.

Google’s three to four year cycle for CAPEX remains intact, which looks like another case of the useful life of servers showing up in the financials. Further validating this cycle, Google’s CFO said growth of CAPEX spend will “moderate quite significantly” in 2019.

Google’s CAPEX for its “other Bets” continues to plummet to the point of irrelevance, clocking in at a mere $181 million, down from almost $1.4 billion two years ago. That bodes poorly for the prospects of a space elevator from Google and their opportunity to atone for pioneering the all-seeing surveillance economy.

I think search, ads, and YouTube continue to dominate Google’s CAPEX spending. Google Cloud Platform uses a small portion of their common infrastructure and it is hard to find a material revenue bump for it given everything else in the same reporting segment (Android app purchases via the Play Store and their hardware business which can add revenue, if not margin, in big chunks). Thus, it is hard to believe GCP is growing its share of Google’s overall infrastructure.



Microsoft still has the cleanest CAPEX numbers from a cloud standpoint (i.e. most of their CAPEX is for cloud infrastructure), though their distractions are growing with new hardware products likely in capital-intensive stages (HoloLens, Xbox, Surface) and a major campus building project getting under way. They continue to grow their investment materially, with a focus on having the broadest global footprint.

Missing the (Bamboo) Forest for the (Apple) Trees

Apple laid the blame for its market-mangling earnings miss on China, noting “over 100 percent of our year-over-year worldwide revenue decline occurred in Greater China.” The announcement roiled markets and unleashed the usual hurricane of evergreen opinions about Apple’s premium pricing and perceived pace of innovation, differentiation in the smartphone market, and the wisdom of Trump’s trade war. Equity analysts used the announcement to once again demonstrate their unrivaled ability to forecast the future with absolute confidence right on the heels of explicitly failing to predict the present (with a very few exceptions).

Despite Apple attributing the entirety of the miss to China and highlighting the softening economy there, the desire to relitigate eternal debates about Apple and the smartphone market obscures what I think is a broader and much more important inflection point: the Chinese economic model of the last forty years has BOTH run out of gas AND the Chinese government is now on a very different policy path that might best be characterized as neo-Maoist.

After unprecedented decades of growth that have pulled hundreds of millions of Chinese out of extreme poverty and made China the world’s second largest economy, China’s rate of growth has been slowing, and perhaps dramatically so in the last year or two (official numbers are increasingly suspect). This slowdown predates the arrival of President Trump with his sixteenth century mercantilist mindset, though the trade confrontation is probably exacerbating China’s slowdown and amping uncertainty.

Apple is the most successful Western company in China and it is now paying the price for that success. Some of its China travails are company and market-specific (particularly premium pricing and stronger Chinese competition). “Buy China” sentiment may be a consequence of Trump’s trade war. But Apple fundamentally is still at the mercy of a faltering economy (and, some argue, with a diminished moat in China relative to the rest of the world). The slowdown in Chinese consumption goes beyond just smartphones, and appears to have taken a sharp downward turn in the last months of 2018. Car sales in China, for example, plummeted 16% in November. Others point to a precipitous drop in Chinese luxury tax revenues late in the year.

My thinking about China has been greatly influenced by George MagnusRed Flags: Why Xi’s China is in Jeopardy. He lays out why what got China here economically won’t get them to the next level as they struggle with debt, demographics, and despotism under newly-minted President-for-life Xi Jinping (or Xi JinPooh as I like to call him). China’s investment-led and stimulus-happy economy badly requires fundamental rebalancing and reform, but China’s leadership finds itself trapped between conflicting needs to deleverage an excessive debt load, boost productivity to offset the world’s most rapidly ageing workforce, prop up a fixed exchange rate, and reform governance if they want to be an advanced economy. Xi’s orthodox economic instincts only increase the degree of difficulty here.

Magnus also underscores what a break Xi represents from Chinese leaders since Deng Xiaoping. Authoritarian, a Chinese Communist Party superfan, a centralizer of power facing few checks or balances: he’s bringing Maoism back. Domestically, he’s reasserting Party control everywhere while building a cutting-edge police state that melds Facebook-style all-seeing digital eye with old school Maoist re-education camps. Abroad, he’s abandoned Deng’s strategy to keep a low profile and is exercising a much more aggressive, nationalist, and jingoistic foreign policy. His promise to “unify” Taiwan during his reign as emperor might be the foremost threat to global stability today. And as the Chinese economy flags (a problem likely to be aggravated by his policies), the risk increases of doubling down on nationalist distractions.

These economic and political changes fly in the face of the last two decades of conventional wisdom about China. Its future economic trajectory is unlikely to resemble that of the recent past (as Herb Stein says: “If something cannot go on forever, it will stop.”). And China’s quiet rise is being replaced by a much more aggressive and confrontational external posture. The implicit Western assumption of ever greater liberalization accompanying growth has been crushed under Xi’s boot.

The technology industry seems late to recognizing and digesting this inflection point. Even in Washington DC, where bipartisan consensus is rarer than an IBM product living up to its hype, there is broad agreement that China is now a strategic rival. Xi’s commitment to “Make China Great Again” has galvanized America’s political attention. And as is often the case, the populist diagnosis is largely accurate even if they get the prescription wrong. The confrontation with the West is much deeper than trade. Superficial declarations of victory on trifling trade issues after making so much noise (the Trump modus operandi) will not dampen or counter China’s aggressive “Made in China 2025” industrial policy ambitions to seize leadership across a range of high tech industries and desire to become the dominant power in Asia.

Tech needs to set aside the intoxicating dream of selling just one product to each and every person in China and parse the new reality. The Great Firewall of China model of parallel ecosystems may expand to other domains, as today’s deeply intertwined global economy starts to disentangle. Supply chains will have to factor in tariffs and national security concerns. Cross-border flows of capital and acquisitions will flow less freely. Non-tariff barriers and even further technology executive hostage taking may escalate. The prospects for China’s private tech titans (i.e. BAT) are in flux as they find themselves with a Party invitation they can’t refuse.

I don’t believe Mr. Xi’s China’s model will prevail. Despotism undermines itself and it is hard to think of a worse way to build an advanced economy than looking to Mao for guidance. And the Chinese people have seen and tasted too much of the rest of the world to go along with Xi. But the happy path for China’s rise over the last two decades is over and we need to recognize that. It will be bumpy, and not just for Apple.

If I Had $34000000000

(With apologies to Barenaked Ladies and all the beleaguered IBM shareholders)

If I had 34 billion dollars
(If I had 34 billion dollars)
Well, I’d buy you a hybrid cloud
(I would buy you a hybrid cloud)
And if I had 34 billion dollars
(If I had 34 billion dollars)
I’d buy you a container platform for your hybrid cloud
(And maybe a HashiCorp for automation)
And if I had 34 billion dollars
(If I had 34 billion dollars)
Well, I’d buy you a Kube distro
(A nice reliant orchestrator)

And if I had 34 billion dollars I’d buy your linux
If I had 34 billion dollars
(I’d build a data center in our yard)
If I had 34 billion dollars
(You could help it wouldn’t be that hard)
If I had 34 billion dollars
(Maybe we could put a little tiny blockchain in there somewhere)

You know we could just go in there and hang out
(Like access the blockchain and stuff)
And there’d all be foods laid out for us
Like little permissioned tomatoes and things
They have permissioned tomatoes
But they don’t have permissioned DB2?
Well, can you blame them?

If I had 34 billion dollars
(If I had 34 billion dollars)
Well, I’d buy you a cloud
(But not a real cloud, that’s cruel)
And if I had 34 billion dollars
(If I had 34 billion dollars)
Well, I’d buy you some modern tech
(Yep, like lambda or a CPU)
And if I had 34 billion dollars
(If I had 34 billion dollars)
Well I’d buy you IBM Watson’s remains
(Ooh all them crazy exaggerated boasts)

And if I had 34 billion dollars I’d buy your linux
If I had 34 billion dollars
(We wouldn’t have to walk to the store)
If I had 34 billion dollars
(We’d take a scooter ’cause it costs more)
If I had 34 billion dollars
(We wouldn’t have to do consulting)

But we would do consulting
(Of course we would, we’d just charge even more)
And sell really expensive mainframes with it
(That’s right, all the fanciest — mainframes)

If I had 34 billion dollars
(If I had 34 billion dollars)
Well I’d buy you a cloud
(But not a real cloud, that’s cruel)
And if I had 34 billion dollars
(If I had 34 billion dollars)
Well I’d buy you a better start
(A Heptio or a Docker)
If I had 34 billion dollars
(If I had 34 billion dollars)
Well, I’d buy you a chaos monkey
(Haven’t you always wanted a monkey?)

If I had 34 billion dollars I’d buy your linux
If I had 34 billion dollars
(If I had 34 billion dollars)
If I had 34 billion dollars
(If I had 34 billion dollars)
If I had 34 billion dollars
I’d be 20 billion in debt


(I may yet finish a longer and less lyrical analysis of IBM’s proposed $34 billion acquisition of Red Hat).

Press Releases We’d Like to See: Larry Page and Mark Zuckerberg to Attend Cannes Advertising Festival

Larry Page and Mark Zuckerberg to Attend
Cannes Lions 2018 International Advertising Festival
Advertising Industry’s Most Powerful People to Attend
Industry’s Biggest Event for First Time

CANNES, France – June 10, 2018 – The Cannes Lions International Festival of Creativity today announced the addition of two very special guests to next week’s Festival schedule. Google co-founder and Alphabet CEO Larry Page and Facebook founder and CEO Mark Zuckerberg both will be attending advertising’s premier event for the very first time this year.

“After almost two decades of selling ads organizing the world’s information and hundreds of billions of dollars in advertising organizing revenues, I realized I’ve never actually met anyone from the advertising industry,” said Mr. Page. “I was told Cannes was a perfect place for such an encounter and that there is pretty good parking for my yacht.”

“The Cannes Advertising Festival is a great opportunity to advance the Facebook mission of apologizing for bringing the world closer together,” said Mr. Zuckerberg. “I’m particularly looking forward to meeting with the Russian delegation to talk about our shared interests in building community, collecting data, and undermining democracy.”

The two advertising titans are expected to let the rest of ad industry know what if any crumbs will be left for them to subside upon and also to thank European regulators for all 260 pages of GDPR.

Since its first outing in 1954, the Cannes Lion International Festival of Creativity has been bringing the creative communications industry together every year at its one-of-a-kind event in Cannes to learn, network, and celebrate.

Follow the CAPEX: Separating the Clowns from the Clouds

In previous installments (2017, 2018) of Follow the CAPEX (merch store opening soon!), we’ve looked at the capital expenditures behind the three hyper-scale public clouds. Now, we turn our attention to two vendors who often tell us they also belong in that group: IBM and Oracle. While I have written about their CAPEX spending before (IBM, Oracle), if derisively due to the paltry sums involved, in this post we will look at their investments systematically and compare them to the hyper-scale cloud players. Because, unlike marketing, CAPEX doesn’t lie.

CAPEX spending on cloud infrastructure is both a leading indicator of the ability to compete at hyper-scale and also confirmation of success with customers. In cloud computing, CAPEX is the ultimate form of putting your money where your mouth is, because no amount of jawboning alone will conjure up data centers or pack them with millions of servers.

As with previous analysis, the goal is identify an inflection point when a company started investing materially in cloud infrastructure and get a sense of their trajectory and cumulative investment.



Alas, IBM’s CAPEX spending is not the picture we expect from a rapidly (or even modestly) growing hyper-scale public cloud. There is no (positive anyway) inflection point, just protracted decline. Where on IBM’s cash flow statement are the vast data centers, much less transoceanic cables? In the cloud era, IBM’s CAPEX is shrinking in both absolute and relative terms. Even as IBM’s topline has plummeted by almost $28 billion since 2011 (equivalent roughly to today’s combined annual revenues of Adobe, NVIDIA and Salesforce), their CAPEX spending has plunged even faster.

In the absence of any clear upward inflection point, it is hard to make any estimates of IBM’s cumulative CAPEX spending on cloud, which is tantamount to saying they haven’t even started to compete. The multi-year, $1.2 billion CAPEX announcement from 2014 might actually be a real number, despite IBM’s long history of vacuous “billion dollar” announcements. To update Number Two from Austin Powers for the cloud: “A billion dollars isn’t exactly a lot of money these days.”


Oracle’s fiscal year ends in May, so their annual spending has been normalized to a March through February year. When compared to the other cloud companies in this analysis, they get to pull in two months of spend from the following year. Stay tuned to see how this little advantage pays off.


Oracle’s chart is much better than IBM’s. It is up-and-to-the-right as we expect from cloud providers and there is a discernable, if late for the overall cloud market, inflection point in 2014. This is a much more precise way to figure out when Oracle started to take cloud seriously versus trying to figure out which of their annual OpenWorld proclamations over the last decade to take seriously.

While there is some CAPEX in here for various SaaS acquisitions, Oracle’s cumulative spend on cloud CAPEX looks to be about $3.5 billion. However, that’s about what each of the big three public clouds spend in a quarter.

Stepping into the Big Leagues

Now let’s examine how our two challengers stack up with the hyper-scale clouds (trigger warning: it isn’t pretty):


We see IBM’s CAPEX slowly trailing off, like the company itself. IBM has always spent a lot on CAPEX (as high as $7 billion a year in their more glorious past), from well before the cloud era, so we can’t assume the absolute magnitude of spend is going towards the cloud. The big three all surpassed IBM’s CAPEX spend in 2012/13. In resisting the upward pull on CAPEX we see from all the other cloud vendors, IBM simply isn’t playing the hyper-scale cloud game.

That red line you may mistake for the x-axis is Oracle’s CAPEX spending. While they have finally separated themselves from the axis in the last couple years, they’re still by far the smallest spender of the bunch. Amazon, Google, and Microsoft each spent more on CAPEX in 2017 than Oracle has in its entire history.


While the big three have converged to spending about 12% of revenue on CAPEX, IBM and Oracle are eerily similar at 4.8%. It is interesting to compare Oracle to Microsoft, as both were asset-light software companies at the dawn of the century. They had similar CAPEX/Revenue ratios until 2005, when Microsoft started putting its software in data center-sized boxes.

On Magical Thinking

One explanation for lower CAPEX spending, and one advanced by Oracle in particular, is that their technology stack is just so much better than anyone else’s that they don’t have to spend nearly as much on CAPEX. Oracle Co-CEO Mark Hurd:

“We try not to get into this capital expenditure discussion. It’s an interesting thesis that whoever has the most capex wins. If I have two-times faster computers, I don’t need as many data centers. If I can speed up the database, maybe I need one fourth as may data centers. I can go on and on about how tech drives this.”

James Hamilton at AWS debunks this completely and the Register recaps the smackdown with even more snark than I can muster.

But more alarming for Oracle is they don’t seem to know that when they think they’re bragging about their cloud scale, they’re actually demonstrating how hopelessly far behind they are. Deploying dozens of racks a week may sound good if you don’t realize the hyper-scale clouds do orders of magnitude more every week. These kind of data points just reinforce that Oracle is falling ever further behind with every passing week.

End Game

As I keep repeating, CAPEX is both a prerequisite to play in the big boy cloud and confirmation of customer success. Both IBM and Oracle are tens of billions of dollars in cloud infrastructure CAPEX behind Amazon, Google, and Microsoft. Oracle’s spending has at least ticked up, but their spending is not enough to keep pace, much less to have any hope of catching up to the infrastructure of the big three.

Both IBM and Oracle (like many vendors) make it difficult to do apples-to-apples comparisons across different parts of their cloud businesses, but we can draw some conclusions from their reported growth rates. For the first quarter in 2018, IBM claimed their cloud business at the broadest level (which includes both SaaS revenue and – literally hardware — the proverbial kitchen sink) was up 22%, while their “as-a-service” cloud business was up 25% (which is less than half the size of their “not-cloud cloud business”, whatever that is). Oracle claimed their overall cloud business was up 32% (again, a lot of SaaS) while IaaS and PaaS lagged slightly at 28%. Contrast this to first calendar quarter 2018 reported growth for AWS of 48% and 89% for Microsoft Azure (Google doesn’t report growth or break out Google Cloud Platform revenue). Both Amazon and Microsoft are growing much faster than IBM or Oracle on dramatically larger revenue bases. Given vastly smaller CAPEX investment and a fraction of the revenue growth on an at least one order of magnitude smaller base, it is next to impossible to see how IBM or Oracle ever really are competitive with the hyper-scale cloud vendors.

Maybe they are reconciled to sitting at the children’s table of cloud, but the problem for both IBM and Oracle is cloud is eating their existing businesses. It has eaten the server business and now starting to feast in earnest on software infrastructure, including the database, which is the profitable heart of these companies. IBM’s revenue implosion has been on display quarterly for years. Oracle likes to point out Amazon runs their e-commerce business on Oracle database, and there are a lot of other legacy Oracle customers out there. But there just aren’t very many new customers for Oracle’s database. New software license revenue is down to 14% of total revenue in the most recent quarter and that number continues to decline. And be sure that when gets off Oracle (and they’re hard at work on this), they’re going to show the whole world how to do it. AWS already has a booming migration service for Oracle (and other) databases. The need to replace the database business and/or move it to the Oracle cloud will only become more acute.

The problem for both IBM and Oracle is their sticky, high margin platform businesses are being eroded, and their customers are migrating to other cloud platforms. Both companies have acquired a number of SaaS applications, which will bolster their sense of self-worth and belonging in the cloud,  but there is little to no platform leverage associated with these apps (and platform leverage = profits!!!). Meanwhile, the big clouds will leverage their platforms and move up from the infrastructure layer to take an ever larger bite out of overall IT spend, i.e. IBM and Oracle’s customer bases.

IBM at this point has greater self-awareness about their predicament than Oracle (and perhaps that is why they’re not making any incremental CAPEX investment because they know they can’t win this game). I’ve expressed some skepticism about IBM’s prospects in the cloud era for a while. After a cringe-inducing campaign claiming to be bigger than AWS (which reinforced IBM doesn’t really understand the difference between hosting and cloud), they have mostly moved on to making preposterous claims in other areas (Watson which is in serious contention to be the biggest “overpromise and underdeliver” in tech industry history, now blockchain as they try to save humanity from our looming existential tomato provenance crisis, and as that founders in its absurdity,  they are moving onto quantum computing). IBM has been better at storytelling than delivering technology in the 21st century. They still tell Wall Street ridiculous things with Watson-esque bravado about their “position as the leading enterprise cloud provider”, but it seems half-hearted at best. IBM’s biggest investment in cloud seems to be “convincing” (perhaps using monetary inducements) a boutique analyst firm to define cloud as cloud plus whatever it is that IBM does (“private hosted cloud” ¯\_(ツ)_/¯ ) in a way that lets them claim to be a “Big Four” cloud provider (Synergy’s numbers are ridiculous, but that is a post for another day).

At the most basic level, IBM has a product problem and a customer problem. The product problem is they can’t build competitive technology any more. As Gartner delicately puts it: “IBM has, throughout its history in the cloud IaaS business, repeatedly encountered engineering challenges that have negatively impacted its time to market.” And their customer problem is who their customers are at this point: the disrupted. You may not get fired for buying IBM, as the old saying goes, but it is increasingly likely your employer will go out of business if you’re in an industry where technology matters.

Oracle finally understands the threat cloud poses, and have aggressively responded, but don’t yet seem to have a realistic view of their prospects. Their traditional strategy of picking a verbal fight with the leader in a new market isn’t going to work when they haven’t actually spent the money to build out a cloud to back up the rhetoric. Oracle doesn’t suffer from IBM’s product development problem. They’re throwing tons of money at people in Seattle and, like a bank, making every employee a vice president. But they’re way behind, need to spend tens of billions to have a competitive global cloud infrastructure, and have a much more severe customer problem than IBM: their customers hate them.

No one is looking to electively increase their dependency on Oracle (and embracing cloud takes vendor dependency to a whole new level). Oracle has always been very aggressive on the sales side, and their desperation around cloud appears to be taking it to new heights. Gartner, who are otherwise inexplicitly inexplicably upbeat about Oracle’s cloud efforts, says, “Oracle sometimes uses high-pressure sales tactics to sell its cloud IaaS offerings, including software audits or threatening to dramatically raise the cost of database licenses if the customer chooses another cloud provider.” The risk for all existing Oracle customers, as the company’s cloud ambitions collide with reality, is the company will get even more aggressive in monetizing its non-cloud installed base in order to sustain its revenue. If you think their maintenance fees and audits are bad now, just wait until they start waterboarding customers (or maybe they already do?).

TL;DR: Thanks for playing IBM and Oracle.

Rumor has it IBM's hat seen here was designed by Watson
Clown-1-1” (CC BY-SA 2.0) by americanbulldogbully007

Follow the CAPEX: Cloud Table Stakes 2017 Edition

Death Star II

In our last installment of Follow the CAPEX™ (motion picture rights are still available), we looked at the gargantuan CAPEX spending by the three hyper-scale public cloud companies: Amazon, Google, and Microsoft. That previous post compiled CAPEX spending for the cloud leaders back to 2001 and let us speculate about when they got serious about cloud and what might be their cumulative investments in infrastructure. This post updates that data series with the nearly $45 billion they collectively spent on CAPEX in 2017. See that previous post for the methodological details.

It is mighty expensive to be a software company in the cloud age and the hyper-scale clouds show no signs of slowing down their infrastructure investments. All three companies registered all time highs for CAPEX spending in 2017. It is important to note that none of these companies break out their specific cloud infrastructure investments and all of them spend significant CAPEX on things beyond cloud infrastructure, particularly Amazon where the list is almost endless, so these are gross but directionally interesting numbers.

In 2017, Amazon spent $19.7 billion, Google $13.2 billion, and Microsoft $11.4 billion on CAPEX (including capital leases for both Amazon and Microsoft). Those are year-over-year increases of 58%, 29%, and 12% respectively.


While Amazon has been the biggest annual CAPEX spender since 2016, Google still has the most cumulative spend (assuming 2001 as the starting date), but on current course Amazon will take the lead next year (half their cumulative spend has happened in the last two years!). Collectively, these three companies have spent over $195 billion on CAPEX in the last 17 years.


On a percentage of revenue basis, we’re seeing an eerie convergence by all three companies around 12%.



The top line Amazon CAPEX number is pretty useless at this point for insights about AWS infrastructure spending. There are just so many other things consuming CAPEX – distribution centers, warehouse robots, grocery stores, cargo jets, office buildings, biospheres, a hyper-segmented line of talking alarm clocks, etc. – that AWS infrastructure is almost certainly a minority of the total company CAPEX. But at a company level, the spend is staggering and is now bigger than any US company I can find except AT&T (and Amazon should pass them this year assuming very low double digit CAPEX growth). This is where Amazon’s cash flow is going.


Given the relative maturity and scale of the AWS business and its fairly consistent growth rates at scale, I suspect AWS CAPEX investment is largely tracking revenue at this point. The capital leases are probably a pretty good proxy for AWS CAPEX ($9.6 billion in 2017).


Google’s CAPEX spiked materially in 2017. Partly this is because they seem to have a three to four year cycle and also I suspect because they have finally accepted the idea of data sovereignty. Google resisted the idea of smaller datacenters in more places for architectural reasons but seems to have finally bowed to customer and government demand.

The other observation for Google is CAPEX associated with their Other Bets declined precipitously by almost two-thirds. Alas, it doesn’t look like we’ll get a space elevator from Google. Bequeathing a space elevator to humanity remains the only acceptable act of Google atonement I can imagine for their role in pioneering the surveillance economy.



Of the three cloud players, Microsoft’s numbers most closely approximate their actual cloud infrastructure spend because they have the fewest CAPEX-consuming extracurricular activities (though a forthcoming campus redevelopment may impact that). Microsoft took an early lead in establishing a broad global footprint and have more regions than Amazon and Google combined, with at least ten more on the way. LinkedIn evidently is continuing to run in its own datacenters, which raises interesting issues for how they will integrate with Dynamics and other products.


For a future installment, I may yet go try to tease Apple’s cloud infrastructure investment out of their total CAPEX spending of $12.3 billion in 2017 (traditionally they spend lots on manufacturing tooling). Or try to make sense of Alibaba Group financial reports to get a handle on their infrastructure investments. Or perhaps look at whether the CAPEX backs up the chatter from cloud pretenders IBM and Oracle.

The Cloud is Not Flat

I am late to updating the geographic view of Gartner’s Cloud Infrastructure as a Service Magic Quadrant. Here is the update for 2017 (see 2016, 2015):



1\ The Seattle region (aka the Leaders quadrant) is unchanged. This supermassive region may turn into a black hole, warp the space-time continuum, and consume most of the world’s $2.5 trillion in IT spending. A hat tip to Seattle companies CenturyLink and Skytap who presumably by virtue of breathing the local air also make the magic quadrant, albeit in inner Nichelandia.

2\ The biggest change this year is the introduction of the Lesser Seattle region (and only truly old school Seattleites will fully appreciate that label), which captures companies that may physically do their cloud work in Seattle, but are not wholly of Seattle or the cloud.

Previously we have included Google in the Seattle region, but the company’s cloud efforts have undergone a major transition. What had been a largely unsupervised outpost in Seattle that couldn’t help but marinate in all things cloud has become “strategic” and returned to the corporate yoke in Mountain View. This shift in center of gravity and management accretion has resulted in key personnel departures and left Google Cloud as perhaps the company’s number six or seven priority in the fight for corporate attention and resources.

Joining Google as an inaugural member of the Lesser Seattle region is Oracle, who have been busily hiring the most mercenary and/or aggrieved of AWS and Azure employees to staff their latest attempt to do cloud (I regret I have lost count of what attempt number this is). They are at least enthusiastic about it, to quote one delusional Oracle recruiter: “We’re the only company delivering the most compelling services at every layer of the cloud.”

3\ Denizens of Nichelandia

Alibaba’s Alicloud makes its MQ debut sited on the finest property in all of Nichelandia. It will be very interesting to see if they can get customer traction beyond China.

Gartner seems to have decided not to fight the battle this year with IBM about their position on the MQ (IBM last year delayed the MQ release by over two months with massive executive escalations, because at this stage they’re pretty much down to browbeating analysts, exploiting tax loopholes, and hyping perpetual motion machine Watson). Instead, Gartner finessed the issue by optimistically positioning them in eastern Nichelandia based on a product that doesn’t actually exist while highlighting their actual cloud offering is a clown show perhaps less than you might expect from a company that proclaims with Watson-esque bravado that it is “the enterprise cloud leader”:

The current offering is SoftLayer infrastructure, not NGI (Next Generation Infrastructure). Other than an early 2015 introduction of new storage options, SoftLayer’s feature set has not improved significantly since the IBM acquisition in mid-2013; it is SMB-centric, hosting-oriented and missing many cloud IaaS capabilities required by midmarket and enterprise customers. The details of the future NGI-based cloud IaaS offerings have not been announced. IBM has, throughout its history in the cloud IaaS business, repeatedly encountered engineering challenges that have negatively impacted its time to market. It has discontinued a previous attempt at a new cloud IaaS offering, an OpenStack-based infrastructure that was offered via the Bluemix portal in a 2016 beta. Customers must thus absorb the risk of an uncertain roadmap. This uncertainty also impacts partners, and therefore the potential ecosystem.

The IBM Cloud experience is currently disjointed. Some compute capabilities, such as the IBM Bluemix Container Service and OpenWhisk, reside in Bluemix, but Bluemix is hosted in just three SoftLayer data centers, and is thus not local to most SoftLayer infrastructure. Some SoftLayer infrastructure can be provisioned through the Bluemix portal, but this is not currently an integrated IaaS+PaaS offering, because Bluemix and SoftLayer do not share a single self-service portal and catalog with a consistent CLI and API; do not provide customers with a single integrated low-latency network context; and do not offer a unified security context that allows the customer self-service visibility and control across the entire environment.

Finally, even in Nichelandia, self-proclaimed technology powerhouses Los Angeles and New York City are inexplicably unrepresented. Instead they will pay rent to the big cloud providers who will also pluck their tenants’ juiciest morsels as platform vendors inevitably do (and Amazon will probably even pluck less juicy morsels like the putting-stuff-in-a-box-and-shipping-it “technology” companies so popular in these ecosystems).

Follow the CAPEX: Cloud Table Stakes

The capital expenditures (CAPEX) going into the cloud are Sagan-esque, with billions upon billions being spent on sprawling complexes of interconnected datacenters scattered across the planet. The hyper-scale public cloud operators (Amazon Web Services, Google, Microsoft) operate BFGCs (big, uh, freakin’ global computers) at immense industrial scale, with townships of well-ventilated warehouses that collectively hold millions of servers, connected by their own transoceanic cables.

Just to give a feel, here is a Microsoft datacenter complex in the eastern United States:


It seems pretty impressive until you see the expansion under way, which dwarfs the original complex (in the distance at the top of the following picture) and will result in a facility a mile long:


Beyond being mind-blowing to people like myself who come from the traditionally asset-light-to-the-point-of-no-assets-beyond-a-couple-laptops software business, the vast sums being spent to move dirt, pour concrete, bend metal, and sling electrons also help us handicap the race for the trillion dollar cloud “jackpot”.

CAPEX is not sufficient to win the cloud, but it is surely necessary. Not only is massive CAPEX outlay a prerequisite to offer cloud services globally at scale, but also a strong indicator of on-going success. Customer adoption and utilization necessitate further CAPEX spending.

And there are significant economies of scale operating at hyper-scale. You get efficiencies of operation, specify your own highly-optimized and cost-reduced servers and network equipment, get exclusivity on the latest CPUs (and probably soon will just design your own) and increasingly run a private global fiber network so you don’t have to pay telco retail (where there is no Moore’s Law). It is an unprecedented level of vertical integration.

And these accumulated CAPEX expenditures, given their sheer scale and the time required to translate balance sheet cash into a global footprint of fully operational and interconnected datacenters, constitute a significant competitive moat. This cumulative investment is a proxy for the size of those moats.

This post examines when the hyper-scale players had their “cloud inflection point”, i.e. when they began to devote CAPEX to their cloud, as well as the magnitudes of their cumulative and on-going CAPEX investment. From this analysis we get a sense of just how big a check new entrants will have to write if they want to play this game seriously.

Unhelpfully for our task, the cloud players don’t break out their cloud-specific CAPEX. It requires some work, speculation and reading tedious financial documents to glimpse the portion dedicated to cloud as opposed to other investments. Every company at this scale makes non-trivial investments on mundane things like office buildings and other facilities. But each also has CAPEX unique to their businesses. Amazon spends big on warehouses and the robots scurrying around inside. Google has (or at least had) investments in self-driving cars, a veritable air force of flying objects (balloons, drones, satellites), retail fiber networks and one still hopes the odd space elevator that all require some degree of CAPEX. All three companies build hardware products for which they may invest in manufacturing tooling which their outsourced manufacturing partners then operate.

Our analysis begins in 2001, as this is the first year for which we have Google CAPEX numbers (a whopping $13 million). We start with the investment section of the cash flow statements in their publicly reported financials. For Amazon and Microsoft we combine the Plant and Equipment line on the cash flow statement with their separately, contentiously and somewhat ambiguously reported capital leases (equipment that is financed and is paid for over time instead of up front, which maps very well to servers that have a limited useful life and are generating revenue over that time, plus we live in a zero interest rate world so why not). While bean counters and finance wonks can argue about the accounting impact of these leases, in real terms they constitute even more CAPEX. Including the capital leases makes the CAPEX numbers significantly higher. In 2016, Amazon did $5.7 billion in capital leases on top of $6.7 billion in CAPEX. We find an extra almost $1.1 billion in CAPEX for Microsoft in 2016 when we check their couch cushions for capital leases. Google does not appear to be using capital leases to fund their infrastructure. The Microsoft numbers have also been mapped from their July to June fiscal year to the calendar year, which the other two companies use as their fiscal year, for a true annual comparison.


All three companies show roughly similar trajectories, with annual CAPEX spending in 2016 between $10 and $12 billion. There is some variation from trend around the financial crisis, with Google pulling back sharply in 2009, Microsoft in 2010 and Amazon going parabolic out of the crisis.

The magnitude of these expenditures is even more impressive when compared to some of both the biggest companies on the planet and the biggest spenders on CAPEX. It is stunning that Microsoft now outspends Intel on CAPEX.


When we look at cumulative spend since 2001 through 2016, the curves are also similar. Google has spent $58 billion on CAPEX since their founding, while over the same time period Microsoft has spent over $48 billion, and Amazon almost $45 billion (though it is the most backloaded):


When we normalize against revenue, things are choppier:


Google is the biggest relative spender, though fluctuating dramatically from 4% in the aftermath of the financial crisis to as much as 18%, averaging 12% over the entire period.

The Microsoft numbers may be the most interesting as they illustrate the transition from an asset-light software company into one with a global cloud footprint, going from under 4% to 12% of revenue spent on CAPEX today. I’m also told these numbers understate the increase as the cloud build-out began while Microsoft was in the midst of a big office building spree, meaning the typical CAPEX for a pure software business is even lower (hmm, Oracle might be interesting to look at in this regard…).

Next we’ll drill down on each company.


Google pioneered the cloud computing model to support their search and ad business, so their cloud inflection point is basically coincident with the company’s inception. It is only much more recently that they have begun to try to dumb down and expose a tiny fraction of their infrastructure externally as a public cloud.

Given this data cover almost the entire lifespan of the company, it also includes things like office space for over 70,000 employees (with slides a de rigueur if extraordinary facilities expenditure).


Google’s cloud infrastructure supports search, YouTube and a billion some odd Android devices, amongst other things. Google Search is the biggest application in the world, entailing trillions of annual queries against the many copies of the entire Web they store (which they size at a modest 60 trillion URLs). The ad business relentlessly tracks every click by billions of users. YouTube, which Google bought in 2006, serves up millions of hours of video daily and no doubt has required totally insane CDN and network investments, contributing to its continuing lack of profitability. Google Cloud Platform is basically a rounding error compared to these other applications, but nevertheless gets to leverage the underlying infrastructure.

Google’s extracurricular “Other Bets”, aka Google “X-cess”, also consume CAPEX. Google started to break out the CAPEX associated with “Other Bets” in 2014. Despite the sheer amount of metal involved in many of these activities, the CAPEX involved is actually a smaller percentage of the overall spend than might have been expected (and this is probably bad news for hopes of a grand societal bargain wherein we accept the all-surveilling eye of Google tracking our every move in exchange for a space elevator). However, CAPEX for “Other Bets” has grown rapidly from $501 million in 2014 to $1.39 billion in 2016 (over 13% of total CAPEX). It will be interesting to see what effect jettisoning various satellite and drone programs as well as approaching “put-up-or-shut-up” time for self-driving cars will have on this spend going forward.

It isn’t crazy to think Google may have spent upwards of $50 billion on their infrastructure over the lifetime of the company. No small part of that represents multiple generations of servers and exabytes of disks that have been replaced due to obsolescence and failure, as opposed to purely incremental capacity (so with a three year useful life, Google could have deployed six generations of hardware for some of its capacity). Broadly, Google’s public cloud efforts get to piggyback on these efforts, although they are also making some incremental investments just for Google Cloud Platform. They have had to deviate from their historic and highly centralized “just do it our way” attitude and are investing in smaller datacenters located in more geographies to address data sovereignty demands.

While Google faces a variety of challenges in its public cloud efforts, it is safe to say they are much more likely to involve fickle humans, product-market overshoot, and finding the right software face to put on their infrastructure as opposed to infrastructure itself.


Microsoft’s cloud inflection point is pretty easy to discern. We see two big accelerations in Microsoft’s CAPEX spending. One starts with an perhaps overly optimistic build-out for search in 2005-2009 that got the company cloud expertise and a global infrastructure footprint. The second begins in 2012 and continues to accelerate through 2016, driven presumably by Azure and Office 365. The bulk of that growth is for cloud CAPEX though some may also support the Surface and Xbox hardware businesses. It is notable that Nokia came and went with little or no discernable impact on Microsoft’s CAPEX spend (presumably Nokia had outsourced their manufacturing by that point).


Microsoft also has a search business. While they have only a fraction of Google’s queries, it still requires ginormous global infrastructure. Microsoft has spent $43 billion in CAPEX since 2006. Taking a stab and saying 80% of that is for cloud infrastructure suggests a cumulative cloud CAPEX investments of on the order of $30 billion (similarly summing total CAPEX above 2% of revenue yields similar number).


As with all things Amazon, there is frenetic activity on multiple fronts. CAPEX spending barely registers before 2009 and in that year we see the knee of the proverbial hockey stick, with CAPEX growing 40-fold since. This CAPEX explosion includes AWS infrastructure (and Amazon isn’t leveraging an existing search-scale infrastructure, so this spend is driven by AWS), a massive expansion of e-commerce distribution centers to get ever closer to ever more of the buying public, and an office building frenzy that has turned Seattle into the “crane capital of America”.


Figuring out how much of this is for AWS is difficult. Amazon, not renowned for its transparency, actually provides a little more breakdown on its CAPEX spend than the other companies, breaking out both capital and build-to-suit leases (hat tip to The Next Platform for charting):

Unfortunately, like Amazon’s infamous charts with no units on the vertical axis, this additional information isn’t particularly useful. We can speculate that buildings (datacenters, distribution centers, office buildings, and/or biospheres) are constructed on build-to-suit leases and the computing stuff that goes into the AWS datacenters is purchased via capital leases. Amazon does say about their capital leases: “the increase reflecting investments in support of continued business growth primarily due to investments in technology infrastructure for AWS, which investments we expect to continue over time.” Amazon has spent about $11.4 billion via capital leases since 2009.

The challenge is teasing apart what has gone into AWS vs. the traditional Amazon e-commerce business (that AWS now subsidizes). We know AWS has seen explosive growth since its debut in 2006, but so too has the the rest of Amazon which has grown revenue by $115 billion (not including AWS) in the same period. And they have grown their distribution footprint at least five-fold in that time:

Amazon's warehouse footprint has grown rapidly in recent years. (Chart Via Institute for Local Self Reliance)

These warehouses are occupied by at least 30,000 Minion-esque Kiva robots. Even if these robots were funded by capital leases, at a cost on the order of $1000 per robot, they’re not likely to be material.

My guess is AWS’ infrastructure spend is on the order of $15 billion between servers and buildings. One area where AWS is behind both Google and Microsoft is owning their own network, which means they are paying a lot more to move bits. Presumably they are planning to rectify this and will add it to their CAPEX spending. We can expect Amazon’s CAPEX investment to continue to grow based on the three vectors above, plus they are opening a new vector as they expand their distribution business with 40 planned air freighters plus long haul and delivery truck fleets (FWIW, FedEx spends $5 billion a year on CAPEX, 10% of revenue). And maybe the Prime Air drones will be more than a PR stunt.

Cloud Table Stakes

This reading-between-the-lines analysis suggests the hyper-scale clouds collective CAPEX spend on their infrastructure could be approaching $100 billion ($50B for Google, $30B for Microsoft, $15B for Amazon). Given the cloud jackpot is a trillion plus dollar annual opportunity, that doesn’t seem crazy at all.

Are there are other estimates or disclosures of CAPEX spend on cloud infrastructure out there? Am I missing anything (e.g. I ignored amortization because I’m interested in the gross spend, not accounting values)?

In a future post, we’ll play follow the CAPEX for two companies that keep telling us they are leaders in the cloud contest: IBM and Oracle. The CAPEX will tell us if they have real clouds or are just clowns…

The Spectacle that is Dreamforce

My condolences to anyone in the Bay Area next week as Dreamforce descends like a khaki-clad, expense account-fuelled barbarian horde upon ancient Rome.

Salesforce may feel compelled this year to surpass its own superlative standards of spectacle to compensate for a mounting set of inadequacies, including slowing growth at a company (and stock) sustained by growth, confronting real competition that can actually write code, being reduced to enlisting the EU as a strategic partner and an increasingly Oracle-esque purchase experience (and pricing). In a cosmic twist of irony, Salesforce has become the Siebel of its generation, focused on defending its multiple at all costs to keep it out of Oracle’s clutches.

For your own safety, do not stand between Salesforce and trendy topics, due to the risk of extreme and hyperbolic buzzword triangulation. They will no doubt put the A and I into “marketing” and serve as the definitive example of the saying “AI is whatever hasn’t been done yet.”

If you’re going to miss this rapturous occasion, here are some excerpts to tide you over from the funniest chapter in Dan Lyons’ book Disrupted (my review), about HubSpot’s visit to Dreamforce:

“Imagine Joel Osteen pumped up on human growth hormone. Imagine there’s a secret government lab where scientists have blended the DNA of Tony Robbins with the DNA of Harold Hill, the aw-shucks shifty salesman from The Music Man. Imagine a grizzly bear in a pinstriped suit, standing on his hind legs and talking about changing the world through disruptive innovation and transformation. If you can imagine those things then you can almost imagine the horror of seeing Marc Benioff, the billionaire founder and CEO of, on stage at his company’s annual conference, Dreamforce.”

“The whole thing makes me depressed, in part because Benioff is a buffoon, a bullshit artist, and such an out-of-control egomaniac that it is painful to listen to him talk. He lives in Hawaii and signs his emails “Aloha.” He’s a Buddhist and hangs out with Zen monks from Japan, and he gave his golden retriever the title “chief love officer” at his company. He is the Ron Burgundy of tech. He and this conference are the essence of everything that has gone wrong in the industry.”

“There’s an art to this kind of horseshit, and Benioff is its Michelangelo.”

“Sure, Benioff is full of shit, but so are we, and Benioff is way better at being full of shit than we are.”

“Now, here in the Moscone Center, the P. T. Barnum of the tech industry is giving a master class in how the game is played.”

“The truth is that has little new to introduce. All of the stuff about hospitals and Haitians and Huey Lewis is meant to distract us from noticing that Benioff doesn’t really have much to talk about other than warmed-over versions of old products.”

“The final day features a speech by Deepak Chopra, noted charlatan and quack. He and Benioff are friends. Chopra rambles on about joy and meaning and interconnectedness and the importance of loving yourself. The old W. C. Fields line “If you can’t dazzle them with brilliance, baffle them with bullshit” seems like the motto not just for Chopra but for the entire conference. Benioff and his philanthropy, the dry ice and fog machines, the concerts and comedians: None of this has anything to do with software or technology. It’s a show, created to entertain people, boost sales, and fluff a stock price.”