Amazon Web Services is the undisputed leader in the public cloud. Azure and Google Cloud are competitive offerings on a feature set level, but Amazon has won the critical battle of being seen the the de facto choice in the category. AWS rivals Kleenex in brand ubiquity.
Whenever a public cloud rival launches a new feature, it’s always put into the relief of comparison to AWS. That status of a benchmark, is incredibly valuable, both in terms of market perception, and the competitive pressure it puts on all other players. And the 800lb public cloud gorilla shows no signs of slowing down. They continually lead in capital expenditures, to extend the infrastructure lead they already have in the space.
But as the saying goes, it gets lonely at the top. The cloud is still an emergent category, there’s a lot of industries and use cases to still expand into. It’s not a settled market in any sense. But still, Amazon has some leeway to get, what I can only describe as, weird.
The entire idea behind Amazon Voice Services seems to come out of the embarrassment of compute resources available to AWS at any given moment. The original Echo seemed to come out of left field when it hit the market in 2014. But with rock solid basic functionality at launch, and a steady build of partnerships and robust developer tools, the Echo has been a success, with Alexa rivaling Siri for name recognition. So how does Amazon build out the product family?
I think everyone saw adding a camera to an Echo as a natural evolution of the product. What they didn’t see was that it would be marketed as a way to digitally check your wardrobe. I’m sure developers and/or Amazon will eventually expand its capabilities for home monitoring, but the initial market focus is baffling. I’m willing to be there’s quite a computing backhaul for this to be useful, from image recognition on particular outfits, to using AI to analyze the “fashionability” compared to other options. It’s the kind of product you create when backend cloud compute is essentially free and infinite…
Amazon Polly is an example of a really cool service that is also weirdly creepy. It’s a speech-to-text service that lets you customize the output with 47 specific voices. But because Weird Amazon, that’s not enough. Nope, now there’s a separate SSML tag that allows for a “whispered” voice as well. I vaguely see a use case for this, it allows for a little more subtlety in your voice output. But this is such a tiny refinement, it screams of boredom. I do enjoy the idea of an engineer having to explain why whispered speech is fundamentally different than simply lowering the volume. Also, enjoy the announcement video featuring a man who is really good at doing exactly one thing with his hands.
I used to think snowmobiles were simply the preferred conveyances of Bond villains at a ski chalet. For Amazon, it’s a ton of storage on wheels. If Imperator Furiosa needed to move 100PB of data across the scorched wastes, she’d saddle up and ride an AWS Snowmobile. I’ve written about this rolling SAN behemoth before, but if this doesn’t scream Weird Amazon, I don’t know what does. I can’t help but think the AWS Snowball was initially developed as a dare, and then everyone just went along with it because it was kind of awesome. Competitors are trying to build new regions where their customers are, Amazon is building mobile storage to physically bring your data center to their cloud.
Here’s the thing, weird doesn’t mean bad. In some ways Amazon is analogous to Google’s market position with search. Having such a commanding lead in a space gives you the luxury to try weird. I’d rather see a lot more Weird Amazon than for them to shift into the malaise of late empire. Maybe that will occur once the public cloud market solidifies in the next decade, when they’ve built an infrastructure so vast with such deep economies of scale as to be untouchable.
But for now there are enough players still gunning for at least a substantial portion of the public cloud market. It’s still Amazon’s game to lose for a large extent, and Weird Amazon may at some point bleed over into sheer indulgence. But for now, sitting at the top, they’ve bought themselves time. Time enough to get a little lonely. Time enough to get a little weird.
- Gestalt IT Comment Policy - September 15, 2017
- Intent-Based Network Automation with Ansible - September 15, 2017
- Network Longevity – Think Car, Not Smartphone - September 14, 2017
- Packet Works on ARM - September 13, 2017
- Storage Myth Making, New Distributed Storage, and Brave New World of NVMe inGestalt Storage News 17.4 - September 13, 2017
- Mesosphere, Kubernetes and the coming container orchestration consensus - September 12, 2017
- Managed Services from the 90s to Now – The On-Premise IT Roundtable - September 12, 2017
- Choice: Illusion or Tradeoff? - September 6, 2017
- Deep Thoughts: VMware on AWS - September 5, 2017
- Reconsidering DH2i with DxEnterprise 17 - September 1, 2017