Don't forget that Intel still owns the data center

Intel released its Xeon Skylake processor family on Tuesday, and along with the announcement come the
Don't forget that Intel still owns the data center
By ARCHITECHT • Issue #110
Intel released its Xeon Skylake processor family on Tuesday, and along with the announcement come the now-common claims about how Intel is struggling against growing competition and how it’s late to the artificial intelligence party. Much like with claims about Apple’s impending demise, I would argue that doom is in the eye of the beholder. Profits and market share might slip—maybe temporarily, maybe permanently—but there is still a whole lot of money to be made.
There are a lot of reasons for this, some of which, with regard to AI, I’ve laid out here before. But the boiled down version is that we’re still very early in terms of AI adoption and applications. Even if Nvidia forever owns the training of AI models on its GPUs, Intel has plenty of space to operate in cars, industrial devices, wearables and everyplace else that experts predict we will see embedded chips specialized for AI tasks. Of course, Intel is also investing heavily in its own AI and high-performance architectures, and isn’t ready to cede those spaces to Nvidia just yet.
Here are a handful of previous newsletter posts on this topic:
It’s also worth noting that SaaS services and other applications that utilize AI algorithms will run, as well as big-data processing workloads, will likely continue to run on data center CPUs for the foreseeable future. AI is through-the-roof hot right now, but AI workloads that presently benefit from (but don’t technically require) GPUs are only a blip on the radar of total data center workloads.
Intel presently owns an estimated 96 percent of that market. While Nvidia’s data center revenue more than doubled to $409 million in Q1 2017, Intel grew only 6 percent but did reach $4.2 billion
Cloud computing is only growing, too, and cloud providers including AWS, Google and DigitalOcean have all publicly announced the availability of the new Intel processors on their clouds. There’s really no doubt that other cloud providers will also offer them in due time, if they’re not already. Selling at mega-scale to cloud providers is a lower-margin business for Intel than selling servers to every company under the sun, but it’s a big business that is not going away.
For more details on Intel’s new chips (which it claims offer 1.6 times more performance than the previous generation) and more takes on Intel’s business, check out the following items. And please try not to shed too many tears (at least not just yet) for the company currently ranked No. 47 on the Fortune 500:
One other quick note …
HyTrust not previously a company I knew much about, but I knew of DataGravity because its co-founder and CEO Paula Long (who also co-founded EqualLogic back in the day) was a veteran of multiple Structure conferences, as well as the old podcast that Barb Darrow and I did at Gigaom. DataGravity’s software scours enterprise data stores, identifying potentially sensitive information and analyzing user behavior (e.g., who’s accessing it).
Well, DataGravity was apparently struggling and was acquired by HyTrust, which focuses on enterprise compliance and data security, and also announced $36 million in funding on Tuesday
I spoke with HyTrust co-founder and president Eric Chiu, who did a good job explaining the value of DataGravity’s technology (which HyTrust intends to keep selling) in terms of recent and future data privacy laws that many countries are enacting. Essentially, he said, companies need to know where user data is being stored so that (a) it doesn’t cross borders and (b) they can delete it if required to do so. The problem isn’t just that cloud platforms are often distributed globally and somewhat abstracted, but also that employees will often share data and use data in ways that might open their companies up to risk. So they need to be able to track it down wherever it now resides.
Whether or not you’re interested in this acquisition, the topic of data privacy should be on everybody’s mind right now. Yes, it’s an issue that raises lots of thorny ethical questions, but it’s also an issue that is could have a major influence over the business of buying cloud resources in Europe, China and elsewhere.

Sponsor: DigitalOcean
Sponsor: DigitalOcean
Sponsor: Linux Foundation
Sponsor: Linux Foundation
Artificial intelligence
Admittedly, this research is focused on a specific application (object recognition) and a specific model type, but that is where a lot of deep learning energy still goes. That said, the big money will be in AI models that learn from less data.
And its first official investment is in a startup called Intuition Robotics, which wants to help senior citizens stay in their own homes. It’s easy to think about Toyota as just a car maker, but the company has a much broader range of businesses that will benefit from AI.
This was first reported back in May, but now Google has officially confirmed it has an engineer-led AI investment arm called Gradient Ventures. You can check out its first four investment here.
Darktrace pushes the AI-for-cybersecurity marketing pretty hard, but it also has some staunch backers, including the CIO of Las Vegas. The company is pushing unicorn status, so the pressure is now on to really deliver.
The company is called Innoetics, and TechCrunch is reporting the purchase price was less than $50 million. Samsung appears to be struggling with its digital assistant products, so any tech that can help with that is a positive.
This really cool (it will full-on drive for you at less than 60 mph), kinda creepy (especially the driver-monitoring camera part) and the kind of baby steps we need before going fully autonomous.
Enterprise information management is not the world’s most exciting space, but it’s relatively important. Really, when we talk about how AI will remake the economy, that will largely be by remaking lots of mundane tasks.
Sponsor: CircleCI
Sponsor: CircleCI
Cloud and infrastructure
You might not have heard of Buoyant, but its linkerd technology is part of the Cloud Native Computing Foundation and its founders have lots of microservices experience at Twitter and elsewhere. 
I referenced this up top, but wanted to highlight it here, as well. Watching DigitalOcean mature is interesting, especially in a cloud ecosystem dominated by huge companies. Check out CTO Julia Austin on the podcast back in March.
The race to zero apparently stops, or seriously slows down, at about 3 cents per gigabyte.
This is a good read for application developers and architects looking for a way to make the process that much more agile. It’s also a good reminder that “serverless” doesn’t have to mean “Amazon Lambda” or another managed service.  •  Share
It uses Lambda, and complements it with a bunch of other AWS services. You gotta pick your battles and, sometimes, your cloud providers.  •  Share
Sponsor: Bonsai
Sponsor: Bonsai
All things data
I feel like we’ve seen this before—including from Salesforce—with mixed results. But analytics still isn’t a solved problem, so why not try it again?
This is good advice from the folks at Silicon Valley Data Science, but it seems many companies are still far away from any sort of workable cross-company data plan.  •  Share
It’s based around Presto and HDFS, with appearances by Flink, MemSQL, MySQL, Postgres, Kafka and more. My question is when we’ll start seeing the next-big-thing companies that run next-gen, “smaller data” systems and what those will be.
On the one hand, this seems pretty smart because IoT is, well, IoT. On the other hand, there’s a lot of coopetition happening between SAP and its cloud partners, who also are trying to own IoT data management.
Sponsor: Cloudera
Sponsor: Cloudera
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.