ARCHITECHT Daily: What's hot right now: Driverless cars and GPUs

It's another brief intro today because I'm on an early morning flight to TiEcon in San Jose, where I'
ARCHITECHT Daily: What's hot right now: Driverless cars and GPUs
By ARCHITECHT • Issue #70
It’s another brief intro today because I’m on an early morning flight to TiEcon in San Jose, where I’ll be moderating a panel on software-defined infrastructure, containers, microservices, etc. If you’re at the show and want to chat, drop me a line or just come find me.
But here are three items from yesterday that you probably don’t want to miss:
DeepMap raised $32 million for its new take on mapping for autonomous cars (DeepMap): These guys are taking what they’re calling a “full stack” approach to mapping for autonomous vehicles, including utilizing data from fleet vehicles and providing regular, localized updates. The founding team’s resumes include Google, Apple, Carnegie Mellon, Baidu and more, so they have some chops in this space. Accel led DeepMap’s series A round, along with Andreessen Horowitz and GSR Ventures out of China.
Fastdata raised $1.5 million for GPU-powered stream processing (InsideHPC): Not surprisingly, Nvidia—which has been investing pretty heavily lately—led this seed round. It’s smart for Nvidia to seed the industry with as many companies as possible to make GPU processing as ubiquitous as possible. Advances in chips optimized for AI could limit the utility of GPUs on the edge, but there’s still a lot of opportunity for someone to own stream processing inside the data center. (GPU-powered database startup MapD has making some noise lately, too; I touch on it in this issue.)
Shazam moved its GPU farm to Google’s cloud (Shazam): This is a good explanation from Shazam on how it uses GPUs to power its music-recognition service, and why and how it moved its seemingly sizable GPU infrastructure to Google Cloud. The popularity of cloud-hosted GPUs, even for non-AI workloads, has to have Nvidia feeling pretty optimistic (although, like I said above, it shouldn’t take the opportunity to rest on its laurels). And although Google has a long road to catch AWS, its seeming ability to win greenfield cloud deployments (of which there are many) has to have it feeling pretty confident. 

Sponsor: Cloudera
Sponsor: Cloudera
Artificial intelligence
For a company like CrowdFlower, the popularity of AI has to be a boon for business. All of a sudden, you have companies being told the next big thing requires lots of high-quality data, which likely means paying someone to get yours in order.
This is partially specific to TensorFlow, but mostly provides some best practices that anyone managing an open source community might consider. Of course, it helps to have Google’s resources behind your project, too ;-)
Microsoft’s prowess in AI research tends to get overlooked, but don’t be fooled. The company has teams of smart researchers spread across the globe. You have to wonder where it will find its killer app though, which could be harder than for some of its competitors with stronger consumer profiles.
This is a cool experiment to make dairy farms smarter by utilizing all the data they’re collecting. It also raises, to me, the possibility that AI-washing (which this probably is) is actually a good thing because it gets people’s gears turning more than just the idea of being “data-driven” ever did.
Listen the the ARCHITECHT Show podcast. New episodes every Thursday!
Listen the the ARCHITECHT Show podcast. New episodes every Thursday!
Cloud and infrastructure
Hosting petabytes worth of AT&T databases a big win for Oracle’s cloud, but AT&T is an existing customer. Oracle will need to add a lot of new customers if it wants to join the discussion with AWS et al.
I don’t understand Salesforce’s stream of public commitments to infrastructure providers (last year, it was AWS) or what it gains from them. Especially when building your own gear is the new black.
For the 9,000th time: Companies are looking at containers as a unit of resources they can move from cloud to cloud, if need be. However, the pull to use those managed, optimized cloud-provider services is strong.
This isn’t the clearest explanation of why they are, but the author is onto something. They should feed off each other for quite a while as developers seek lighter-weight infrastructure for lighter-weight, distributed, disposable endpoints.
This quote pretty much sums up my feelings: “SAP investors may welcome this litigation, but it’s yet one more reason why new customers are fleeing to the hills, choosing Amazon Web Services or Google, where they can, for their future software needs.”
Media partner: GeekWire
Media partner: GeekWire
All things data
Revenue up 35 percent, losses down about 20 percent. If Hortonworks can get everything rolling in the right direction, we might see the Hadoop space get very interesting again. And Hortonworks is making some smart moves, especially around IoT and edge infrastructure.
I really don’t know what Saggezza is or does, but I do think Cisco is smart to invest as heavily as it can in network analytics as customers’ architectures start incorporating more clouds, microservices and devices.
Did you enjoy this issue?
The most interesting news, analysis, blog posts and research in cloud computing, artificial intelligence and software engineering. Delivered daily to your inbox. Curated by Derrick Harris. Check out the Architecht site at
Carefully curated by ARCHITECHT with Revue. If you were forwarded this newsletter and you like it, you can subscribe here. If you don't want these updates anymore, please unsubscribe here.