This site may earn affiliate commissions from the links on this page. Terms of use.

TSMC-Fab

If you lot'd told me 5 years ago that Intel was near to stumble into arguably the worst manufacturing problems of its existence, I would accept been surprised. Our 2012 story on why Intel leads (led) the globe in semiconductor manufacturing now looks more like a historical retrospective than a confident projection of time to come trends. The company'south sideslip — and the potential consequences of that slip — have driven a lot of assay over the past few years, some of it written by yours truly.

Merely there's some other question underlying the 10nm problems that deserves attention. How much does it actually matter to the silicon manufacture as a whole, that Intel slipped on the 10nm node? Apparently, there have been real repercussions as far as the visitor's power to see demand, and Intel would undoubtedly prefer to retain its reputation for being the virtually-advanced silicon manufacturer in the business. Merely the long-term impacts are surprisingly tricky to sort out. I was reminded of this by a recent Bloomberg story, titled "A Company Few Americans Know Is About to Dethrone Intel." It's about TSMC and its overall process node leadership, and information technology makes a lot of expert points. Intel's 10nm is expected to exist roughly equivalent to TSMC'southward 7nm, but TSMC has 7nm in-marketplace already in mobile hardware and AMD will almost certainly ship 7nm CPUs before Intel starts moving 10nm in book. It's been virtually xx years since AMD and Intel transitioned to the same node at the same time. From the original Athlon to the present day, AMD has never beaten Intel to a node. Add together up the total R&D spending by all of TSMC's customers (some of which is surely being spent at TSMC itself) and the sum is larger than what Intel itself spends on R&D.

Similarly, nosotros nevertheless expect the shift to 10nm to have an touch on on silicon operation besides. At that place's a reason why AMD chose to limit its 14nm fries to 32 cores but volition push upward to 64 cores on 7nm. The firm projects a 25 percent performance improvement in the same power envelope or a 50 percent power consumption improvement at the same frequency. Even allowing for manufacturer fudging, that's still a substantial heave. Nosotros've also seen larger GPU generational performance improvements when companies shift to new nodes as opposed to when they release new products on the same node. The spring from Maxwell to Pascal was larger than the improvement from Kepler to Maxwell, for example. So on the one hand, yep — semiconductor performance improvements matter, and given that 10nm will provide some of them, it matters to some extent. The real question is — how much?

All the Intervening Variables

On the other hand, Intel's 10nm node is already years late. At IDF 2013, Intel talked about having chips in-market by 2015. Despite this, Intel's business is positively booming. Data heart growth is huge. The company may take famously missed mobile, merely it has a 5G modem apparently well positioned for the appearance of that advice standard, a growing IoT business, surging data center shipments, and plans for new 48-core Pour Lake CPUs to ship next year with a new MCM blueprint. In the Bloomberg commodity, Intel'due south server sectionalisation head, Navin Shenoy, argues that what customers care about isn't process node, every bit such. "I'k confident that we're going to deliver what our customers intendance almost, which is system performance," he said. This is a real — and important — signal.

Efforts to push ARM into data centers have, to date, not enjoyed tremendous success. Amazon's conclusion to build its own ARM cores is absolutely a warning sign that Intel could face up long-term intensified competition from alternative CPU architectures, merely the company has done a very skillful job fending off those challenges so far. Rory Read'due south 2012 prediction that ARM would constitute 15 pct of the server marketplace by now isn't fifty-fifty inside an order of magnitude of being true. The only vendor cut into Intel'southward information center share is AMD, and just by single digits. That'southward still a strong consequence for AMD, but the virtually-term threat to Intel just isn't very large. Chipzilla has made it clear that cloud and data center is a major pillar of its business going forward, and these customers tend to be more conservative than the typical consumer marketplace. They besides tend to be less price sensitive.

Intel's IoT revenue is growing adequately well.

Information technology's difficult to say how all of this plays out in burgeoning markets like AI, machine learning, and self-driving cars. I remember you can make a very off-white argument that Moore's law and Dennard scaling collectively functioned similar a type of straitjacket. With an assumed 50-100 percent operation comeback arriving every few years, at that place was precious little room for alternative architectures or new approaches to computing problems. If you started in 1980 with a custom architecture that was 15x more efficient than CPUs of the solar day, but information technology took you until 1988 – 1990 to bring your product to marketplace, chances were practiced that conventional processors had either cutting your do good to cipher or eaten into information technology to such a degree that your scrap could never compete on price after economies of scale were taken into account. CPU clocks have only improved modestly since 2004, while Intel's overall CPU operation has scaled modestly (outside of certain AVX-friendly workloads) since 2011. These two developments opened the door and the funding streams for alternate approaches to computing. Only Intel has positioned itself to play in these spaces every bit well, with acquisitions similar Movidius. Information technology may have canceled its Knights Hill product, but the company'due south GPU efforts will likely target the aforementioned infinite with a better, more than competitive architecture.

Estimating the impact of a 10nm filibuster on Intel'south overall position in the semiconductor business as it moves into these markets is difficult because process node is merely one factor impacting overall product adoption. Intel's addiction of offering higher performing CPU cores but fewer cores at the same price point compared with AMD may non exist a negative if you buy software that'southward licensed per-core as opposed to per-socket. There are specific applications that favor Intel to a sufficient caste to starting time some of the cost concerns. Companies also tend to sign long-term license or supply agreements that buffer the bear upon of performance shifts that favor one vendor over some other. Intel is working hard to scale its Optane business and 5G modem engineering science with the stated goal of offering comprehensive solutions to its partners, not "merely" a CPU core. It's piece of cake to run into how discrete GPUsSEEAMAZON_ET_135 See Amazon ET commerce could fit into that stack also. By 2020, Intel will be able to make a play to provide a arrangement's Wi-Fi and cellular connectivity, GPU, CPU, and retentivity. That's a much wider stack than the company has previously fielded and it's difficult to predict how customers will respond.

Bloomberg'due south observation that calculation upwardly the R&D budgets of all of TSMC's customers results in more total spend than Intel seems rather banal to me and I'm not certain it has any predictive power at all. First, some of that R&D spending is going to exist duplicative, reflecting things like mask and tool costs. 2nd, not every product TSMC builds even has an Intel analogue. Tertiary, the question of full R&D spending is less of import than what a visitor is spending R&D dollars on. Fourth, R&D spend as a raw figure may non exist a good predictor of last product performance no matter what. Plenty of people looked at AMD's anemic R&D spend from 2012 – 2016 and concluded in that location was no gamble of Ryzen being any kind of competitive production.

Intel has made process node leadership disquisitional to its concern reporting for many years, simply at the Credit Suisse 22nd Annual TMT Conference, the company acknowledged that its ain guidance is shifting. Hither's Intel acting CEO Bob Swan:

I think there's a few dynamics near the visitor that nosotros created and we asked you to judge us by them. And over time, if you stick with those, I think you get shortsighted. And we've talked well-nigh a couple. Moore'due south Law leadership is all that matters. That's non true. Product leadership is what matters. But we told your forever that Moore'southward Law leadership matters. And at present we're trying to tell y'all, hey, nosotros get paid on product. And then, that's a fundamental shift that I think it's very important – it's very important for us in terms of how we run the concern and I believe increasingly important for you well-nigh how you make your decisions.

It might be tempting to dismiss this equally Intel pivoting away from a metric only because it isn't winning in that category any longer, but the terminate of conventional Moore's law scaling isn't unique to Intel. It's something the entire semiconductor industry is grappling with. But nosotros're as well in a reality where no ane is sure which AI/ML products volition succeed, or how rapidly alternative architectures volition ramp, or how long it will be before autonomous vehicles are common. The path frontward for conventional silicon scaling below 5nm is securely uncertain.

It's easy to say that sure, not being in a leadership position is going to cost Intel something. It'southward much harder to predict exactly what that toll is going to be. If I personally had to guess, I'd bet on less disruption rather than more than. With the gains from each new node smaller than the node before, the chances that missing a node will prove a major problem are lower as well. The factors that thing to future success are likely to be rooted in the types of products a company brings to market place and how well those products are positioned to meet the needs of the new workloads we meet rise by the day. Process node positioning/leadership is important to that question, just non the sole determinant.

Now Read: Intel Is at a Crossroads, Intel CPU Shortage Could Ease in Q1 2019, and The Myths of Moore'due south Police