Apologies for not publishing a newsletter yesterday. As some readers might know, I live in Las Vegas, so my mind was elsewhere.
But while all that was going on here, Oracle was in San Francisco announcing a product for seemingly every cloud capability under the sun—most of which are also powered by Oracle’s best-in-the-world artificial intelligence. Also, Amazon Web Services doesn’t know anything about elastic computing, but the new Oracle Database 18c will be a fully automated, fully elastic marvel of information technology.
Larry Ellison’s particular brand of keynoting and marketing is commendable, but it’s also a relic of an IT world that largely doesn’t exist anymore. At least not in the cloud computing world where Oracle is staking its future. In that world, you win by pleasing users, mostly by doing rather than by talking.
When Google talks about cloud infrastructure, people listen because Google is cloud infrastructure. When the new Microsoft talks about open source, people listen because the new Microsoft is a bastion of open source. You could swap Google for Microsoft, and vice versa, and both statements still would be more or less true. The same goes for AI research.
When AWS talks about, well, anything related to cloud computing, people listen because AWS became the giant it is by pioneering the cloud and then methodically responding to customers’ desires in the form of product releases and price cuts.
Oracle has built an empire selling database (and various other) software products, but as a cloud provider it hasn’t done a whole lot. So when Ellison takes the stage and starts bragging up his tech and bashing the competition, people rightfully shrug it off. They’ve used AWS, they’ve seen the teams Google and Microsoft
are assembling, they know the score.
Oracle co-CEO Mark Hurd appears to have delivered a much more prudent and practical vision
of Oracle’s cloud and AI roadmap, but I didn’t see too many people talking about that. Which is unfortunate—because while it’s not a whole lot of fun to talk about getting small things right before moving onto bigger things, it’s certainly realistic. And as counterintuitive as it might seem, a little humility might go a long way in helping give customers the impression that Oracle actually gets how the cloud works and is doing what it takes to get there.
In case you missed it, here’s a rundown of most (I hope) of what Oracle announced yesterday (sometimes from a couple of sources for added flavor):
In other news, I’m skimping the commentary on the news items below because there was a lot of stuff between Friday and Monday, but make sure you read these pieces:
To compete with new rivals, chipmaker Nvidia shares its secrets
(WIRED): This is a smart decision, assuming Nvidia can actually help foster a community of people/companies building on its open source designs. I’m guessing the business rationale is to blunt the increasing R&D in embedded AI chips by commoditizing the market and making Nvidia’s data center bread-and-butter the market that matters.
The hippocampus as a 'predictive map’ (DeepMind):
I do believe this new way of thinking about AI and the brain will inspire some big advances, but something strikes me as off about reducing the brain’s functionality to something that can be modeled in software.
Is AI riding a one-trick pony? (MIT Tech Review):
Yes, but it’s a really useful one for the foreseeable future. Otherwise, this is a good interview with Geoff Hinton of Google/Vector Institute/University of Toronto, who is also looking to the brain as the basis for his next big advance in AI.
Alibaba is leading a $27M investment in open source database startup MariaDB (TechCrunch):
This is really interesting, mostly because the traditional move by cloud providers—including Alibaba Cloud—is to offer MySQL or built their own product. I wonder if this is strategic (e.g., trying to seed adoption of MariaDB as the de facto option overseas) or perhaps a reaction to seeing lots of MariaDB adoption in China already.
Google’s Firebase releases Cloud Firestore, a NoSQL database for app developers (VentureBeat):
A few years ago, this would have looked like a shot across the bow of MongoDB. Today, it looks like Google is trying to capture the developers MongoDB needs to leave behind as it attempts to make its revenue match the size of its user base.