As consumers face a chip shortage that is causing scarce availability of everything from cars to toaster ovens, silicon giant Intel is grappling not only with how to make more processors, but also how to make them better.
The company this week offered a peek at the next three years of its processor development plans. For starters, going forward, Intel will no longer refer to its processor node names by the size of their semiconductor fabrication process (usually measured in nanometers, or nm). That nanometer spec—10nm, 14nm, and so on—has long served as the most prominent high-level yardstick for how advanced a CPU is relative to others. Instead, future chips will use fabrication technology identified simply by a number decoupled from the process size, such as Intel 7, Intel 4, Intel 3, and Intel 20A.
It’s arguably a more appropriate way to gauge innovation than by simply referring to the size of the production process in nanometers. After all, with the latest chips built using 7nm and smaller processes, eventually they’ll require measurements in angstroms, the next smaller unit of measure after nanometers. And Intel hopes to use the new naming system as a marketing advantage in the face of stiff competition from AMD, Qualcomm, and other chip companies that are beating it at the moment in the nanometer-number wars.
By the end of 2023, chips based on Intel 7, Intel 4, and Intel 3 production processes should be available. By 2025, Intel says all of the stages of the roadmap it just laid out will be completed. This means that people in the market for a new laptop, desktop, or tablet over the next few years could have some exciting technology to look forward to. We’ve laid out four key ways Intel’s new roadmap could transform personal computing in the near future.
1. Faster PCs: Amping Up Performance Per Watt
The semiconductor industry has long been focused on cramming ever more transistors onto silicon wafers in order to speed up the time it takes for them to handle instructions from the software on your computer. In 1975, Intel’s co-founder Gordon Moore predicted that, given the trends of the time, transistor count was on a path to double likely every two years; industry followers popularized the term “Moore’s law” to reflect that cadence.
Intel plans to introduce new silicon fabrication technology that it says promises more performance per watt than each existing process.
But new production processes that enable more transistors don’t necessarily correlate to the things you look for in a CPU for your next computer—namely, how quickly it can crunch the formulas in your spreadsheet, or how many browser tabs you can leave open without a crash. On workflows that are CPU-intensive, a chip’s speed and number of cores mostly dictate how fast the computer can run. In other words, rendering a video on a dual-core processor will often be slower than rendering it on a quad-core chip, especially if each of its cores has a higher clock speed and is allowed to consume more electricity. (This, of course, hits practical limits, especially in laptops.)
So what really matters to power users is what a processor can do with the power it’s given, a measurement best described as performance per watt. Intel says that the first chips based on the new Intel 7 technology will deliver an approximately 10% to 15% performance-per-watt increase versus the current most advanced 10nm technology.
Following that, Intel 4—scheduled for production in late 2022—is forecast to offer another 20% performance-per-watt increase. And Intel 3 will add an 18% performance-per-watt increase over Intel 4. You don’t even need to do napkin math to tell that each of Intel’s new process technologies will allow for substantially more capable computer chips within the same power parameters.
2. Smarter PCs: Refining ‘Big-Little’ CPUs
But many consumers aren’t power users looking to cut down on video-rendering times or increase the frame rates of immersive video games. Instead of hulking desktop workstations and gaming rigs, you might be interested in futuristic devices that will allow you to work and play however you want. The Lenovo ThinkPad X1 Fold, the world’s first foldable-screen Windows computer, falls into this category. So does the Microsoft Surface Neo, a prototype dual-screen device that Microsoft teased a few years ago before eventually relegating it to a back burner.
Intel has been working for years to come up with processors suited for these types of trim, flexible devices, which are much bigger than a phone but more versatile than a clamshell laptop. So far, efforts have centered around a “big-little” approach, where a single chip consists of multiple processor cores, each with different capabilities, and a central brain that is smart enough to assign each core to the task that suits it best. (That’s in contrast to today’s typical PC designs, which you might define as “all big.”)
The first step in Intel’s new roadmap, Intel 7, could be a great leap forward in refining the company’s big-little design, an approach that rivals Apple, Qualcomm, and Samsung already use to great success. The Intel 7 node will be featured in products such as the upcoming 12th Generation “Alder Lake” family of processors; these chips are already slated for production later this year. Intel hasn’t shared details about the Alder Lake architecture, but rumors suggest it will have a combination of powerful cores with high clock speeds for “bursty” workflows, alongside more-efficient cores that run more slowly to handle background tasks.
A successful Alder Lake family based on Intel 7 processors could make the next generation of dual-screen and foldable-screen PCs more viable, following in the footsteps of successful foldable-screen Android phones from Samsung and others.
3. Longer-Lasting PCs: More Efficient With PowerVia
As with any new technology, behind-the-scenes innovations abound, with arcane acronyms and concepts that won’t ever show up in the marketing materials on the shelves of your local electronics shore. But the quest to increase performance per watt is studded with them—terms like signal routing, voltage droop, and gate-all-around transistors—and they point not only to more power, but also to greater efficiency. And to anyone who has ever glanced anxiously at their laptop’s battery meter, efficiency offers the welcome prospect of longer battery life.
For that, we may have to wait until the end of Intel’s latest roadmap, when Intel 20A comes into existence. It’s the first node to be measured in angstroms instead of nanometers.
Besides more transistors, one of the breakthrough innovations it promises is routing a chip’s power source around the back of a silicon wafer instead of the front, a concept that Intel is branding as “PowerVia.” The technique, which Intel claims is an industry first, seeks to address two arcane but important problems: voltage droop and inefficient signal noise. Both of these prevent a processor from running at its full potential, which could cause it to sap your laptop’s battery and leave you worrying about where to find a power outlet.
Recommended by Our Editors
4. Opening Fabs to Former Rivals
The silicon industry as a whole is an adherent of long-term planning. After all, it’s expensive and time-consuming to develop new production processes, and it takes years to bring a new fab, or chip-making facility, online.
But lately Intel hasn’t talked much about future plans, in part because the arrival of its current state-of-the-art 10nm process experienced significant delays during a period when rivals like AMD consistently introduced new 7nm products to widespread acclaim. One reason for Intel to announce a multi-year roadmap now, with a new CEO and a worldwide chip shortage that shows few signs of abating, is to signal to the world that the company intends to remain a leader in semiconductor fabrication.
Intel’s campus in Arizona
Good leadership involves plenty of cooperation and delegation, so part of Intel’s roadmap will involve partnering with other companies. While Intel forges ahead on its new node technologies, it will also set aside some of the capacity in its semiconductor fabrication facilities for other companies to use. Qualcomm and Amazon will be among the first customers of the new Intel Foundry Services (IFS) division of the company.
Qualcomm currently makes Snapdragon processors for Windows PCs that compete against Intel’s current Core chips. In the future, Intel will make Qualcomm processors based on the Intel 20A node, Intel announced. Cooperation is nice on its own, but additional fabrication capacity is good for business and could even help alleviate or prevent future chip shortages. By selling fab space to Qualcomm, Intel would pad its bottom line while simultaneously churning out more chips to power future PCs.