ArchiTECHt Daily: Intel's new AI group and the hardware renaissance

Intel announced a new AI product group and research lab on Thursday, doubling down on its investments
ARCHITECHT
ArchiTECHt Daily: Intel's new AI group and the hardware renaissance
By ARCHITECHT • Issue #45
Intel announced a new AI product group and research lab on Thursday, doubling down on its investments in AI over the past year or so:
The new product group will be ran by Nervana co-founder Naveen Rao, who himself will report directly to CEO Brian Krzanich.
One way to look at this news is to focus on the hiring aspect—that Intel’s deep pockets will result in yet another big company sucking up all available talent. Hiring is certainly on the top of mind right now, especially with Tuesday’s news that AI expert Andrew Ng is resigning from his leadership role at Baidu. Many people missed another important move in China on Thursday, with Zhang Tong, former head of Baidu’s big data lab, heading to rival Tencent to lead it 250-person AI lab. 
To paraphrase Heidi Klum’s Project Runway catchphrase: “In AI, one day you have all the talent, the next day you don’t.”
However, I tend to look at all of Intel’s investments as evidence that, with the algorithmic flywheel now running at top speed, it’s time to focus on hardware. GPUs might be a fine choice for training models and running workloads inside data centers (although there’s evidence that FPGAs will actually deliver better performance), but they aren’t such a sensical option when you’re talking about a wearable device or an IP camera
Apart from the companies Intel has acquired, there is lots of academic research happening into mobile-scale AI chips and, of course, IBM’s much-hyped TrueNorth technology. Just yesterday, in fact, a startup called Mythic raised $9 million for its button-sized AI processors that it claims can deliver processing power equivalent to 10 mobile GPUs while using remarkably little power (it’s not clear from what I’ve read whether Mythic’s platform actually includes GPUs or not).
As I’ve said before: Intel missed mobile, is seeing declining server sales, and lost out on the early AI workloads to Nvidia. As AI becomes embedded into the consumer experience, in what some are calling the next big era of computing, Intel is determined to dominate once again.

What's new on ArchiTECHt
Around the web: Artificial intelligence
Canada is home to some of the leading AI research universities, and $125 million is a small investment if it can capitalize on its opportunity. Frankly, Canada probably looks like a better draw to a lot smart folks right now.
Google breaks down the the process, and the models, that made the new YouTube feature possible. More impactful yet will be when Google inevitably releases this labeled dataset.
Looking past the fact that Ripcord uses robots to handle much of the labor involved in scanning documents, digitizing them is the first step in actually treating them as analyzable data for other AI tools.
This story is a bit convoluted, to be honest. But the part about having train the bot to give correct answers struck me as a more valuable learning experience than just having a machine TA.
Sponsor: Datos IO
Sponsor: Datos IO
Around the web: All things data
There have been countless attempts at democratizing machine learning and predictive analytics over the past several years. I’m going to take DataRobot’s $111 million in funding as a sign that it’s onto something.
And its first investment, Paradata, does predictive analytics for supply chain optimization. Expect to see lots data-driven, useful and decidedly non-sexy investments from SAP.
This is a good assessment of Cloudera’s strategy, based in part on its recent analyst conference (which I attended, too). The biggest challenges for anybody selling Hadoop will involve cloud services and lighter-weight data platforms.
MIT researchers developed a system called Splinter to spread database queries across multiple copies. It could be a good approach, provided users are willing to make some sacrifices in site selection and performance.
Sponsor: Marshal.io
Sponsor: Marshal.io
Around the web: Cloud and infrastructure
Maybe you’ve heard about the importance of microservices and APIs? Mashape has a lot of momentum right now, and business is likely just picking up as microservices go mainstream.
Here’s another look at what’s actually behind the bevy of announcements at IBM’s cloud event earlier this week, and who’s buying what IBM is selling.
CoreOS is often overlooked, but it’s doing some interesting things with Kubernetes. The big news here its Kubernetes application registry, which automates the process of deploying entire apps, including container images and configuration.
coreos.com  •  Share
Did you enjoy this issue?
ARCHITECHT
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 https://architecht.io
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.