Intel’s new companion chip for cloud providers has Arm inside

The newest member of Intel's IPU data-center strategy is a chip with 16 cores designed by its longtime rival.

Intel's Mount Evans IPU

The new Mount Evans IPU, designed to help cloud providers manage their internal computing needs alongside those of their customers, will come with 16 Arm Neoverse N1 cores.

Image: Intel

Intel turned to an unlikely source for the newest version of its infrastructure processing unit strategy: longtime rival Arm.

The new Mount Evans IPU, designed to help cloud providers manage their internal computing needs alongside those of their customers, will come with 16 Arm Neoverse N1 cores. That's the same core that's at the heart of AWS's Graviton2 processor, one of the greatest threats to Intel's decades-long dominance of the data center market.

But Mount Evans isn't designed to run cloud customer applications like Graviton2. Instead, it's the latest iteration of Intel's attempt to take a page from modern cloud-server designs and build its own companion processors to help cloud providers run their data centers more efficiently.

"We're looking at this in a very pragmatic way," Guido Appenzeller, chief technology officer for Intel's Data Platforms Group, told Protocol. "We make design decisions based on a number of different factors, and in this case, these Arm cores met the design target performance we were looking for."

Still, it's yet another sign that Intel is changing more than six months after Pat Gelsinger returned to the company where he played an integral role in shaping the PC era of the tech industry around Intel's x86 instruction set. Almost all modern PC and server software has been designed with those chips in mind, and over the last two decades Intel executives have been reluctant to even acknowledge the existence of alternative instruction sets, let alone discuss their benefits.

Data centers as hotels

Companies that manage their own servers run everything on the processor at the heart of that server. In the very early days of cloud computing companies like AWS and Microsoft followed a similar strategy, but there is a lot of additional overhead required to manage huge data centers and support modern application designs, which began to overwhelm those processors.

This led to a number of cloud providers piecing together sophisticated networking chips and other co-processors to handle some of that administrative load. Nvidia found a lot of success with chips for this market over the last few years, and Intel committed itself to this design strategy earlier this year with the introduction of its IPUs.

Appenzeller compared Intel's IPU strategy to the way living spaces are designed depending on who owns the space. In your own home, you move between different rooms as you like. When you stay in a hotel, you have freedom to use your own room however you want, but you can't get into your neighbor's room and facilities such as the dining area and lobby are available for your use but are controlled by the owner.

In this analogy, an IPU is the dining area and lobby; facilities you want and expect, but have no expectation or desire of controlling. Intel's Xeon processors or AWS's Graviton2 processors would therefore be like individual hotel rooms, where customers (guests) can access and control the activity in their own space.

"We think of this not so much as an offload, but as a dedicated place to run the infrastructure functions and be under the control of the infrastructure operator," Appenzeller said.

Coming down the mountain

Mount Evans will be Intel's first ASIC (application-specific integrated processor) design for its IPU strategy. Earlier versions, as well as two other new IPUs scheduled to be unveiled Thursday, were based around FPGA chips that can be configured by the customer to suit a variety of needs, but Appenzeller said the ASIC design offers better performance.

It will be able to support up to four Xeon processors that are running customer applications inside cloud data centers, helping move data into and out of those chips with Intel's networking technology while offering additional computing resources with the Arm Neoverse cores. Intel has made several programmable chips for embedded systems that use Arm cores, but Mount Evans will be one of the company's most prominent endorsements of its technology.

Intel is unwilling to say much more about Mount Evans than it plans to reveal Thursday, including when it expects to start shipping the new IPU. Mount Evans was "designed in collaboration with a large CSP," or cloud-service provider, Intel said, but it declined to confirm which one.

Microsoft has a long history of collaboration on both PCs and servers with Intel, but the chip company said it was working with "a different partner" on Mount Evans. AWS built something similar to Mount Evans called the Nitro system for its own internal use a few years ago.


A pro-China disinformation campaign is targeting rare earth miners

It’s uncommon for cyber criminals to target private industry. But a new operation has cast doubt on miners looking to gain a foothold in the West in an apparent attempt to protect China’s upper hand in a market that has become increasingly vital.

It is very uncommon for coordinated disinformation operations to target private industry, rather than governments or civil society, a cybersecurity expert says.

Photo: Goh Seng Chong/Bloomberg via Getty Images

Just when we thought the renewable energy supply chains couldn’t get more fraught, a sophisticated disinformation campaign has taken to social media to further complicate things.

Known as Dragonbridge, the campaign has existed for at least three years, but in the last few months it has shifted its focus to target several mining companies “with negative messaging in response to potential or planned rare earths production activities.” It was initially uncovered by cybersecurity firm Mandiant and peddles narratives in the Chinese interest via its network of thousands of fake social media accounts.

Keep Reading Show less
Lisa Martine Jenkins

Lisa Martine Jenkins is a senior reporter at Protocol covering climate. Lisa previously wrote for Morning Consult, Chemical Watch and the Associated Press. Lisa is currently based in Brooklyn, and is originally from the Bay Area. Find her on Twitter ( @l_m_j_) or reach out via email (ljenkins@protocol.com).

Some of the most astounding tech-enabled advances of the next decade, from cutting-edge medical research to urban traffic control and factory floor optimization, will be enabled by a device often smaller than a thumbnail: the memory chip.

While vast amounts of data are created, stored and processed every moment — by some estimates, 2.5 quintillion bytes daily — the insights in that code are unlocked by the memory chips that hold it and transfer it. “Memory will propel the next 10 years into the most transformative years in human history,” said Sanjay Mehrotra, president and CEO of Micron Technology.

Keep Reading Show less
James Daly
James Daly has a deep knowledge of creating brand voice identity, including understanding various audiences and targeting messaging accordingly. He enjoys commissioning, editing, writing, and business development, particularly in launching new ventures and building passionate audiences. Daly has led teams large and small to multiple awards and quantifiable success through a strategy built on teamwork, passion, fact-checking, intelligence, analytics, and audience growth while meeting budget goals and production deadlines in fast-paced environments. Daly is the Editorial Director of 2030 Media and a contributor at Wired.

Ripple’s CEO threatens to leave the US if it loses SEC case

CEO Brad Garlinghouse said a few countries have reached out to Ripple about relocating.

"There's no doubt that if the SEC doesn't win their case against us that that is good for crypto in the United States,” Brad Garlinghouse told Protocol.

Photo: Stephen McCarthy/Sportsfile for Collision via Getty Images

Ripple CEO Brad Garlinghouse said the crypto company will move to another country if it loses in its legal battle with the SEC.

Garlinghouse said he’s confident that Ripple will prevail against the federal regulator, which accused the company of failing to register roughly $1.4 billion in XRP tokens as securities.

Keep Reading Show less
Benjamin Pimentel

Benjamin Pimentel ( @benpimentel) covers crypto and fintech from San Francisco. He has reported on many of the biggest tech stories over the past 20 years for the San Francisco Chronicle, Dow Jones MarketWatch and Business Insider, from the dot-com crash, the rise of cloud computing, social networking and AI to the impact of the Great Recession and the COVID crisis on Silicon Valley and beyond. He can be reached at bpimentel@protocol.com or via Google Voice at (925) 307-9342.


The Supreme Court’s EPA ruling is bad news for tech regulation, too

The justices just gave themselves a lot of discretion to smack down agency rules.

The ruling could also endanger work on competition issues by the FTC and net neutrality by the FCC.

Photo: Geoff Livingston/Getty Images

The Supreme Court’s decision last week gutting the Environmental Protection Agency’s ability to regulate greenhouse gas emissions didn’t just signal the conservative justices’ dislike of the Clean Air Act at a moment of climate crisis. It also served as a warning for anyone that would like to see more regulation of Big Tech.

At the heart of Chief Justice John Roberts’ decision in West Virginia v. EPA was a codification of the “major questions doctrine,” which, he wrote, requires “clear congressional authorization” when agencies want to regulate on areas of great “economic and political significance.”

Keep Reading Show less
Ben Brody

Ben Brody (@ BenBrodyDC) is a senior reporter at Protocol focusing on how Congress, courts and agencies affect the online world we live in. He formerly covered tech policy and lobbying (including antitrust, Section 230 and privacy) at Bloomberg News, where he previously reported on the influence industry, government ethics and the 2016 presidential election. Before that, Ben covered business news at CNNMoney and AdAge, and all manner of stories in and around New York. He still loves appearing on the New York news radio he grew up with.


Microsoft and Google are still using emotion AI, but with limits

Microsoft said accessibility goals overrode problems with emotion recognition and Google offers off-the-shelf emotion recognition technology amid growing concern over the controversial AI.

Emotion recognition is a well-established field of computer vision research; however, AI-based technologies used in an attempt to assess people’s emotional states have moved beyond the research phase.

Photo: Microsoft

Microsoft said last month it would no longer provide general use of an AI-based cloud software feature used to infer people’s emotions. However, despite its own admission that emotion recognition technology creates “risks,” it turns out the company will retain its emotion recognition capability in an app used by people with vision loss.

In fact, amid growing concerns over development and use of controversial emotion recognition in everyday software, both Microsoft and Google continue to incorporate the AI-based features in their products.

“The Seeing AI person channel enables you to recognize people and to get a description of them, including an estimate of their age and also their emotion,” said Saqib Shaikh, a software engineering manager and project lead for Seeing AI at Microsoft who helped build the app, in a tutorial about the product in a 2017 Microsoft video.

Keep Reading Show less
Kate Kaye

Kate Kaye is an award-winning multimedia reporter digging deep and telling print, digital and audio stories. She covers AI and data for Protocol. Her reporting on AI and tech ethics issues has been published in OneZero, Fast Company, MIT Technology Review, CityLab, Ad Age and Digiday and heard on NPR. Kate is the creator of RedTailMedia.org and is the author of "Campaign '08: A Turning Point for Digital Media," a book about how the 2008 presidential campaigns used digital media and data.

Latest Stories