TSMC 230TSMC is rushing to produce 10's of millions of chips for the next iPhone, taking up nearly all their 2018 capacity at the new 7 nm node. Qualcomm and possibly Huawei have the chips at an advanced stage of design and should have sample chips in the next few months.

Rick Merritt at EE Times reports, "TSMC is in volume production of 7-nm chips today with more than 50 tapeouts expected this year. It’s making CPUs, GPUs, AI accelerators, cryptocurrency mining ASICs, networking, gaming, 5G, and automotive chips."

.

Rick adds, "The new normal for performance gains and power reductions generally fall in a 10% to 20% range, a reality that makes the new packaging and specialty processes increasingly important." In other words, Moore's Law ain't dead but it's slowing down. TSMC is already building the 5 nm plant and taking delivery of EUV lithography gear. Both TSMC and Samsung have committed to $20B 3 nm fabs early next decade.

The first millimeter wave chips are likely to be power-hungry and may run hot. Linley estimates it will require as much as 10X the processing power as early LTE chips. The radio frequency front end will need to work from 600 MHz to 30 GHz or more. Carriers will require many more bands, puching the state of the art. Thousands of engineers are required.

TSMC's next five years

In 2019, TSMC expects to produce 10's of millions of seven nanometer chips, maybe more. Some will be made with EUV lithography.  Numerous inprovements will come. They say, "Compared to its 10nm FinFET process, TSMC's 7nm FinFET features 1.6X logic density, ~20% speed improvement, and ~40% power reduction. TSMC set another industry record by launching two separate 7nm FinFET tracks: one optimized for mobile applications, the other for high performance computing applications."

5 nm "risk production" is expected late in 2019. "MC's 5nm Fin Field-Effect Transistor (FinFET) process technology is optimized for both mobile and high performance computing applications. It is scheduled to start risk production in the second quarter of 2019. Compared to its 7nm FinFET Plus process, TSMC's 5nm FinFET adopts EUV Lithography for more critical layers to reduce multi-pattern process complexity while achieving aggressive die area scaling." Merritt expects, "Compared to the initial 7 nm without EUV, the 5-nm node promises a 1.8x greater density than 7 nm. However, it is only expected to reduce power by up to 20% or raise speeds by about 15%, perhaps 25% using Extremely Low Threshold Voltage (ELTV),"

 6G anyone?

 

dave askOn Oct 1, Verizon will turn on the first $20B 5G mmWave network, soon offering a gigabit or close to 30M homes. The estimates you hear about 5G costs are wildly exaggerated. Verizon is building the most advanced wireless network while keeping capex at around 15%.

The Koreans, Chinese, and almost all Europeans are not doing mmWave in favor of mid-band "5G," with 4G-like performance. Massive MIMO in either 4G or "5G" can increase capacity 4X to 10X, including putting 2.3 GHz to 4.2 GHz to use. Cisco & others see traffic growth slowing to 30%/year or less. Verizon sees cost/bit dropping 40% per year. I infer overcapacity almost everywhere.  

The predicted massive small cell builds are a pipe dream for vendors for at least five years. Verizon expects to reach a quarter of the U.S. without adding additional small cells. 

In the works: Enrique Blanco and Telefonica's possible mmWave disruption of Germany; Believe it or don't: 5G is cheap because 65% of most cities can be covered by upgrading existing cells; Verizon is ripping out and replacing 200,000 pieces of gear expecting to save half. 

-------------------

 5G Why Verizon thinks differently and what to do about it is a new report I wrote for STL Partners and their clients.

STL Partners, a British consulting outfit I respect, commissioned me to ask why. That report is now out. If you're a client, download it here. If not, and corporate priced research is interesting to you, ask me to introduce you to one of the principals.

It was fascinating work because the answers aren't obvious. Lowell McAdam's company is spending $20B to cover 30M+ homes in the first stage. The progress in low & mid-band, both "4G" and "5G," has been remarkable. In most territories, millimeter wave will not be necessary to meet expected demand.

McAdam sees a little further. mmWave has 3-4X the capacity of low and mid-band. He sees an enormous marketing advantage: unlimited services, even less congestion, reputation as the best network. Verizon testing found mmWave rate/reach was twice what had been estimated. All prior cost estimates need revision.

My take: even if mmWave doesn't fit in your current budget, telcos should expand trials and training to be ready as things change. The new cost estimates may be low enough to change your mind.