As the pandemic disrupted our lives and made us acutely aware of our digital dependencies, Intel went for broke under new CEO Pat Gelsinger, making strategic moves to correct its innovation curve, one that it hopes will shape the semiconductor industry over the next decade.
Let¡¯s be honest, this disruptive reset button has been long overdue: It¡¯s been a rocky past couple of years for Intel, with their innovation supremacy losing steam and challenged by several serious competitors. We spoke to Sanjay Natarajan, General Manager of Intel¡¯s Logic Development Group, who¡¯s responsible for defining Intel¡¯s semiconductor process innovation pipeline, one that aims to restore them to the top of the chip manufacturing world by 2025. No pressure, right?
For Natarjan, the journey of a thousand miles, back to the top of the podium, begins with a whole new lexicon of process nodes -- one that¡¯s less complex, more consistent and a little more accurate than it ever was. Everything else comes after, as it should.
[ALSO READ:?Sneak Peek Into Intel Labs To See What's Cooking In 2021 And Beyond]
¡°Apart from renaming everything that comes after the existing 10-nanometer (nm) SuperFin node to something more accurate and easy to remember -- like Enhanced SuperFin is Intel 7, the next 7nm node is simply Intel 4 (Q2 2021), the one after that is Intel 3 (2021 2nd half), and beyond that is Intel 20A (1st half 2024) which is a nod to the Angstrom era of node scaling,¡± Sanjay said, in clear and no uncertain terms, ¡°In fact, Intel 18A¡¯s semiconductor technology innovation that¡¯s coming in 2025 will give us clear leadership in the industry in terms of chip process technology.¡±
Sanjay Natarajan is overseeing an important phase of Intel¡¯s technological transformation, where microchip fabrication will move beyond nanometers into the angstrom (10 angstrom = 1 nanometer) scale, manifesting first through Intel¡¯s RibbonFET transistor design and PowerVia interconnect innovation by 2024-2025, unlocking a new paradigm for Moore¡¯s Law.
Of course, for Intel to be able to innovate and scale up its process node manufacturing, it will need sophisticated devices operating at the cutting edge of physics. ¡°Intel is going to be the first to receive ASML¡¯s high-NA EUV scanner and implement it in a high-volume chip production environment,¡± said Sanjay Natarajan, mentioning how Intel¡¯s working very closely with ASML to help them deliver the equipment on time after much delay.
Having spoken to IBM earlier -- who showed off their ¡°world¡¯s first 2-nm¡± chip architecture in May 2021 -- I couldn¡¯t help but ask Sanjay how Intel¡¯s nanoribbons architecture in RibbonFET due in 2024 looked awfully similar to IBM¡¯s 2-nm nanosheets layers.
¡°Well, every leading chip manufacturing and design company is working on similar stuff that we are working on here at Intel, so you will see architectural similarities,¡± he said. ¡°My key message is that we have a track record of leadership in this area, and we have a real timeline and plan for production. For us this isn¡¯t a research project, this is serious engineering involving RibbonFET in a process node in 2024 fully in production, that¡¯s our goal post,¡± emphasized Sanjay.
And just to be clear, these aggressive milestones that Intel has set for itself require a lot of problem solving at the very cutting edge of applied sciences, at the intersection of physics and engineering (among other things) to unlock scalable process development -- no easy task, as Sanjay Natarajan very well knows.
[ALSO READ:?'We Are 10 Years Away From A Chip That Will Change Our Life,' Says Intel's Quantum Computer Guy]
Computer chips are made of billions and billions of tiny transistors arranged in a complex pattern, all of which begins life as an electronic circuit design. The process of taking a circuit design and transferring it onto a silicon wafer is basically where all the magic happens, thanks to photolithography. It all depends on the wavelength of light and its manipulation, requiring highly sophisticated tools and equipment.
¡°Full use of EUV lithography¡± is an important bullet point in Intel 4¡¯s (or 7nm process node to debut in Meteor Lake) development, according to Sanjay Natarajan. ¡°Not only is the equipment required to be able to do that extremely complex, but there¡¯s an entire ecosystem behind it to make it all click together,¡± he explained, talking about resist development, transfer masks, inspection capabilities as some of the few sophisticated pieces of the jigsaw puzzle that need to fall into place for it to all work seamlessly. Sanjay underscored the importance of, ¡°an entire ecosystem that has to come up and support the equipment itself in order to get to something that Intel can put into its production line to print these even smaller features on the silicon wafer.¡±
¡°The scale of complexity, the boundaries of physics that we operate in, it¡¯s a huge team effort to overcome these challenges,¡± Sanjay explained, ¡°we have a large organization with incredibly talented engineers, and it takes the whole team¡¯s effort to innovate on a technical level, finalize a process node and get it into manufacturing.¡±
Just like manufacturing is impossible without a design phase, so is process development without a thought on the final product as it were, and which products Intel ultimately chooses to debut some of its latest process technologies on. ¡°Generally, we always take a fresh look on any given technology which products we should lead with, right from servers down to the smallest of client devices,¡± highlighted Sanjay, explaining how Intel studies all the gamut of products it manufactures to decide which ones are best fitted for and will benefit most from new process enhancements.
¡°Now that we have Intel Foundry Services (IFS), we'll make these new process node technologies available to our Foundry customers (AWS and Qualcomm) at the same time we're making them available internally for Intel¡¯s own use, and they may bring a class of product that they want to lead with these new technologies¡± Sanjay emphasized, before elaborating on Intel¡¯s lead products on the new process nodes. ¡°Specifically, we have already taped in Meteor Lake, which is the lead product for Intel 4 (7nm process node) -- a client product -- and on Intel 20A we are also going to lead with a client product, although we haven¡¯t been specifically gone public with its name yet,¡± said Sanjay.
Interestingly, IFS has announced Qualcomm as one of its first customers, something we would¡¯ve never imagined just a few years ago, when Intel failed to make inroads into the smartphone market and Qualcomm emerged as one of the biggest players to fill that vacuum over the past decade or so. Sanjay was understandably excited to talk about the Qualcomm partnership, in his limited capacity (deferring most of my questions to the IFS team).
¡°Qualcomm choosing Intel 20A to partner with is very exciting to me, to be able to work with a company of that caliber which has a proven track record of making great products, and to get their perspective on what we can do to make them successful is going to be great,¡± Sanjay suggested. ¡°And the fact that Qualcomm has taken a look at our Intel 20A roadmap and wants to engage with us is a pretty strong testimony to our process innovation capabilities.¡±
Sanjay Natarajan only returned to Intel in April 2021 (after Pat Gelsinger took over as Intel¡¯s CEO in February 2021) and he remained fully confident of Intel¡¯s aggressive process innovation roadmap, despite the pandemic¡¯s underlying impact.
¡°The majority of the people who used to be on campus, where we do this process development, are now working from home. Those onsite are really the people who have to be onsite because of the nature of their job that requires them to be near the tools and wafers. We really tried to minimize the number of people who have to be onsite to the bare minimum, taking significant precautions above and beyond those mandated by local, state and federal government to keep our people safe,¡± Sanjay said.
¡°We¡¯ve been fortunate enough to maintain our pace of execution throughout the past year, even though many of us have been working from home -- including me. We don't know where this pandemic is ultimately going to go in terms of how we come back to some sort of returning to work. But we¡¯re committed to doing five nodes in the next four years, which is a very aggressive, accelerated pace of development. I'm very confident, even in the current environment of how we have to work, that we'll be able to do it,¡± emphasized Sanjay Natarajan.
Only time will tell how prophetic Sanjay Natarajan¡¯s words turn out to be, of course. For now it¡¯s worth noting that an innovative, hyper competitive Intel doing what it does best is only going to be good for the semiconductor industry and ultimately the consumers at large.
Do you think Intel will be able to innovate aggressively over the next few years? Let us know in the comments below, and for more science and technology news and updates keep reading Indiatimes.com.