[This fragment is available in an audio version.]

This is the week that AWS owns the tech news cycle because anyone who cares about it is watching the stories coming out of the re:Invent conference. I watched Adam Selipsky’s keynote this morning and it looks like AWS is moving in a very traditional direction for tech companies: Away from tech-first, towards business-first. Which is inevitable but also dangerous.

Messages · The “S” in AWS stands for “Services”. So presumably, the biggest imaginable announcement would be of new Web Services. There was only one in among the avalanche of biz-talk: managed 5G, nice but not terribly exciting. Powerful message!

What were the companies either heavily name-checked or invited on-stage? DISH, NASDAQ, United, Goldman Sachs, and 3M. None of them West-Coast. None seen as tech players. Powerful message!

The number-one problem I hear about from every single tech leader I talk to is the extreme difficulty in hiring and team-building. So I’d think that providing developer force multiplication by automating away server management would be a big deal. None of that; the focus was almost all (except for a bit from 3M) on hosting existing applications. Powerful message.

There’s also the fact that a high proportion of developers and technologists viscerally loathe at least two of those highlighted companies; so impressing those demographics was not on the priority list. Powerful message!

History · Let’s look way back, to the year I graduated from University: 1981 (no, really). My first job was with DEC. At that point it was the second-largest computer company in the world. The largest was IBM, which was ten times our size and, as our leadership said, “growing at one DEC per year”.

DEC at that time had the coolest computers you could buy and nobody working where tech mattered wanted anything else. Leadership noticed that despite that fact, IBM was kicking our asses. And IBM people never really talked about technology, it was all business all the time. So the word came down from on high that we would too.

I remember going to the announcement of some new computer which was a one-hour presentation, and the first 45 minutes were how DEC cared above all else about the Modern Enterprise and Leadership Agility and Profit-oriented Innovation. This wasn’t a computer, it was a Business Solution! Meanwhile the audience got angrier and angrier because they wanted to hear about memory and form factors and cycle time and benchmarking and so on. Those things were eventually crammed into the last ten minutes of the session, and the computer actually did pretty well.

But the company didn’t. It was skidding downhill by 1990 and didn’t make it to Y2K. My own geeky tribe was pretty convinced that it was because if you wanted to sell computers, you should find people who needed to buy computers and talk about computers to them.

I saw a similar story when I worked for Sun, which for quite a while sold the coolest computers but eventually fell on hard times and was snapped up by Oracle. Oracle talks about business and only about business. Once again, the tech-nerd demographic sneered at Oracle’s lack of technical acumen.

I’m pretty sure my tribe was wrong both times. Here’s why.

Growth space · People think the cloud business is pretty big, and it is. AWS is running at $60B/year and by my guess represents somewhere between a quarter and a half of public-cloud spend, so let’s say the whole Public-cloud biz is in the range of $200B or so. Wow, that sounds big! But estimates of global IT spending are north of $4T.

That math says that 95% of IT isn’t on the cloud yet. Will it all move there? No, but it feels inevitable that the cloud revenue potential is at least 10× today. And where is that 10×? I’ll tell you where it isn’t: In the kind of startup and cloud-native scenarios that led the charge onto the cloud over the last fifteen years. It’s in Establishment IT, where by “Establishment” I mean “big organization with significant systems in production”.

Establishment-IT organizations are cost centers, mostly. They report to people who neither understand nor care about technology, and are incented to find cheap decent solutions to their business problems. So it seems pretty obvious that whoever figures out how to talk to that leadership is in a position to win the 90% of the potential cloud market that’s still up for grabs. And that’s not by enthusing about leading-edge Serverless Event-driven Container Orchestration or whatever.

Horizontal Selling · Here’s another problem that the public Cloud faces as it tries to expand into the enterprise, one I’m bitterly familiar with because I’ve spent my whole life working on general-purpose infrastructural software. Which inevitably leads to the following conversation:

Tim: “Check out this slick new optimized event router we cooked up!”
Product Person: “Great! What can I do with it?”
T: “Well, anything! You know, make distributed apps faster and better!”
PP: “You don’t get it. What actual specific job, like manufacturing cars or processing insurance claims, will this help with? So I can go and talk to people who do those jobs.”
T: “But your apps will be more flexible and robust, too!”

At the moment, pretty well 100% of AWS revenue comes from totally horizontal technology, so you should pity the Product and Sales people who have to build a bridge between, for example, DynamoDB latency and Fast-Fashion inventory management. The fact that they’ve gotten this far is a tribute to them.

I couldn’t help but notice that in Adam Selipsky’s keynote this morning, there was a specifically-called-out emphasis on vertical offerings, for example the “Fleetwise” automotive-telemetry thing (which however launched without a charter customer). I think it’s pretty obvious why AWS is doing this. But the absence of a track record is worrying.

Which way forward? · I’m not gonna offer any recommendations about how AWS should position horizontal tech to sell to Establishment IT. If I did, they should be ignored. [You know who’s really good at it? Andy Jassy.]

But I do have three pieces of advice for AWS on their keynote presentations:

  1. The all-biz-all-the-time messaging implies a trade-off. You should do better selling into that huge 90%-of-trillions IT space that isn’t in the cloud yet. But you’re less likely to land the business of the next Netflix or Intuit or, well, Amazon.com. Maybe that’s OK, but also consider at least some messaging aimed at technology pioneers.

  2. I think that current technology leaders, even in the fustiest corners of Establishment IT, are noticing that they’re having a big recruiting and staffing problem. Modern Cloud-Native tech lets you deliver more business-app goodness with fewer developer-hours. I think that angle is becoming increasingly strategic, and deserves keynote time.

  3. Pick up the damn pace! That two-hour keynote was grindingly slow and contained maybe a half-hour of actionable information. Your audience mostly comprises really smart people and they appreciate high-intensity high-density discourse. (This is another space where you might want to learn from the Jassy style.)

More to come · I understand that we systems Morlocks aren’t the target audience for the big glamorous launch keynote from the CEO. Do I feel slightly disrespected and ignored? Yes. Am I going to go back and listen for actual technology announcements from one of the world’s most interesting technology companies? Sure.


Comment feed for ongoing:Comments feed

From: Paul Clapham (Nov 30 2021, at 17:02)

It was about 1981 when the consulting company I was working for sent me out to Lloydminster to write and install a business application on a DEC computer.

Of course DEC knew nothing about canola processing but our company knew how to talk to businesses and produce working software. So yes, intermediaries like that are what AWS needs.


From: André P (Dec 01 2021, at 13:06)

The problem with vertical offerings is that they risk stepping on their current clients. Nobody wants their 800 pound supplier muscling in on their turf.


From: Paul Marsh (Dec 01 2021, at 14:08)

I wonder if there's an equivalent tech sector metric that would be analogous to the sitcom metric of "jump the shark"?

I've been in a few meetings speaking engineer and it has not translated particularly well with 'client in charge of purse strings'. Is there a moment where the divergent path forward is just totally apparent to anyone actually looking? Once that is hit, it's just a matter of time till the ship sinks. I'm not suggesting for a second this is AWS, they seem pretty focused to miss but lots of other examples do come to mind and I don't see a "jump the shark" type of thing, (or I'd be a much better short-seller).


From: Tim (but not THE Tim) (Dec 02 2021, at 22:23)

I admit to being a Morlock (mainframe systems programmer - now known as system administrator more or less), but I really don't know who the Eloi were


From: Paul Boddie (Dec 04 2021, at 07:56)

I find the whole cloud bandwagon to be rather depressing. Sorry if that is a strong and inappropriate word to use, but it seems to me that as technology gets more sophisticated and complicated, the industry's reaction is to add more technology and complexity. All of this leaves public and private infrastructure vulnerable to catastrophic failure, despite the best efforts to make the technological foundations reliable and continuously available, not to mention completely reliant on a variety of external conditions including the benevolence of various corporations.

Now, I don't really pay any attention to Amazon, either commercially or technologically, unless the company somehow gets covered in something I'm reading, like these blog posts or in various news articles. I stopped buying products from Amazon when the company was trying to patent user interface mechanisms and when it was clear that the corporate strategy was to drive everyone else out of business and make itself the only choice if you wanted to buy anything. This coupling of unhinged consumerism and predatory business practices is slowly destroying our societies: reducing quality employment to "gig economy" jobs, incentivising needless consumption, undermining the funding of our democracies, sabotaging public policy and the relationship between people and government.

Of course, big business is very interested in what was once called "utility computing" and presumably there will be big money to be made. Twenty years ago, this just wasn't feasible with the available bandwidth or technology, but now any organisation can tell its employees and customers that Company X will "take care of things", never mind that only computing products used in the designated way can satisfy such a transactional mentality towards procurement: computing *solutions* actually require the needs of the users to be understood and met. Such lessons seem to need learning by every human generation and with every technological generation.

That brings me to the observations about DEC. Obviously, you and others were there at the time, whereas I only encountered DEC's products in the early to mid-1990s, although that gives me my entry point into the topic. If you look at industry commentary from the era, it is clear that DEC was a company with very proprietary instincts (rather like IBM) which did deliver platforms that enabled their customers to actually address their computing needs, and indeed many emerging technology companies were empowered by the availability of the VAX platform, and yet those instincts ultimately made the company's strategy incoherent and even harmful to its customers.

One occasion where this manifested itself quite clearly, also being precisely where I first encountered DEC's products, was the way that the company rolled out a line of Unix-based workstations - the MIPS-based DECstation range running Ultrix - and then, after having persuaded its customers that DEC finally "gets" the significance of Unix and "open systems", proceeded to leave its customers without either a hardware roadmap (throwing MIPS overboard) or a software roadmap (after having promoted OSF/1, leaving all those customers running Ultrix). You can read the customer reactions to all of this: they aren't kind.

The conclusion to all of this is that DEC acted in a strategically incoherent and inconsiderate way, but it also harmed the company's own prospects in the end. By depriving its then-current customers of OSF/1 and driving them and developers away, no-one was tooling up to deliver stuff for Alpha, and thus everything got off to a far slower start than it should have. DEC was also being bamboozled by Microsoft at the time, which has never been a good place to be for anyone. Sun, where you also worked, and DEC were companies that seem to have wasted far too much energy on internal turf wars that harmed customers and ultimately the companies themselves. It is no real surprise that both Sun and DEC got acquired.

As for Amazon, whilst the company entertains the involvement of others on its technological and commercial platforms, its growth always appears to be at the expense of both its competitors, which would be natural, but also its customers, which is rather more predatory. The one thing you could say about the traditional technology companies was that they were generally just looking out for themselves and taking care of their own business. With Amazon, the impression is rather more sinister and, as I noted to begin with in the context of industry trends, rather more depressing.


author · Dad · software · colophon · rights
picture of the day
November 30, 2021
· Business (123 fragments)
· · Software (4 more)

By .

The opinions expressed here
are my own, and no other party
necessarily agrees with them.

A full disclosure of my
professional interests is
on the author page.