AWS has been relatively quiet until lately on its machine and deep learning story. Has this given Google a beachhead?
Google has been running a distant third in the cloud computing wars, with first-mover Amazon Web Services and enterprise darling Microsoft Azure outpacing it in the infrastructure-as-a-service market. But what if, as Stratechery analyst Ben Thompson has posited, infrastructure (or platform) isn't really the big deal, but rather is democratizing the ability to make sense of big data? In that battle, Google has more than a fighting chance, but new moves from AWS to democratize machine learning will make it hard to gain ground.
Let us teach your machines for you
Machine learning is a hot topic, but it's also a particularly thorny one to navigate. Put simply, most enterprises lack the technical chops to be able to master machine learning. For these mainstream enterprises, Thompson has argued, a select few cloud behemoths can credibly offer an answer:
It seems certain that machine learning will be increasingly dominated by cloud services: Both are about processing scale and massive amounts of data, and only a select few behemoths will have the financial capability to not only build out the infrastructure required but also have the wherewithal to employ the best machine learning engineers in the world. That, by extension, means that for most enterprises the differentiation arising from machine learning will derive first and foremost from whether or not their data is in the cloud (there will be on-premise solutions, but I expect them to fall more and more behind over time), but secondly from which cloud provider they choose.
I've argued that Google's master cloud plan is to open source the magic that runs its own operations, with Kubernetes a classic example. Google can then attempt to offer the best cloud for running that software at scale.
SEE: Why Kubernetes may be a bigger threat to Amazon than Google's cloud (TechRepublic)
Interestingly, in such a world the competition shifts from offering the best infrastructure services and rather moves toward those vendors that can essentially "appify" the complexities inherent in machine learning and AI. In such a world, Google, given its deep experience with big data, has a competitive advantage, as Thompson argued:
[S]uperior machine learning offerings can not only be a differentiator but a sustainable one: being better will attract more customers and thus more data, and data is the fuel by which machine learning improvement comes about. And it is because of data that Google is AWS' biggest threat in the cloud.
Not that AWS is sitting still, waiting to be big data'ed out of existence.
Giving it away, making it easy
At AWS re:Invent this week, AWS released a slew of new features and products to help enterprises harness their data and accelerate machine learning initiatives. For those that want to build their own, AWS announced powerful ways to take advantage of GPUs and FPGAs.
For those that don't, there's Rekognition, Polly, and Lex. Each is somewhat limited in and of itself, but those same limitations are also what makes them powerful: They make embracing machine learning and AI relatively simple for enterprises that would otherwise be completely priced (by cost of hiring talent) out of the market. The goal, said Swaminathan Sivasubramanian, general manager for AWS, "is to bring machine learning to every AWS developer."
SEE: How to Implement AI and Machine Learning (ZDNet/TechRepublic special feature)
This is where the analysts seem adrift from reality. Like before, they're positioning AWS behind the machine learning pack, arguing it's "nowhere on the agenda" for some cutting-edge enterprises, according to Gartner research vice president Alexander Linden, with Google and Microsoft way out in front. This is the same sort of analysis that originally pooh-poohed AWS as merely a toy for test-and-dev workloads.
What Linden and others like him don't seem to realize is that the very arguments they're raising against AWS are the reason it could well succeed. Linden told ZDNet that AWS is "definitely good enough to solve many different problems. But if you really want to be cutting edge, then I have never seen anybody actually doing anything with Amazon." Here's the thing: Most companies aren't "cutting edge"—and "good enough" and "easy/cheap/convenient" win every time.
Could Amazon keep on being "cutting edge"? Of course. As AWS general manager of product strategy Matt Wood told me in an interview, "Amazon as a whole has thousands of engineers working on machine learning and deep learning....We've been doing this for decades." But he continued, Amazon AI services are "all designed to put sophisticated, high-quality deep learning, which is easy to use and priced aggressively, in the hands of as many people as possible." The focus isn't on making hard things harder, but rather about making them approachable.
Let's be clear: The winner in the cloud isn't going to come down to who has the most advanced ML/AI capabilities. It will derive from the same winning formula as before. Whoever makes it easiest for the broadest swath of users will win. In this area Google is competitive—but so, increasingly, is AWS. Game on.
- Google wants to commercialize database Spanner, but MongoDB or Cassandra could be safer bets (TechRepublic)
- Google's master cloud plan: Buy more infrastructure, charge less for it (TechRepublic)
- Google finally gives developers access to its cloudy secret sauce (TechRepublic)
- Why Kubernetes may be a bigger threat to Amazon than Google's cloud (TechRepublic)