Thursday, April 25, 2019

The US is the world leader in 5G; We’re just missing one piece

There is widespread concern in the US, ranging from cable TV shows right up to the White House, that the US is losing the race for 5G wireless networks. The concern is right, but the analysis is wrong. The US is the world leader in 5G networks. We are missing just one piece: the systems integrator. A careful examination of the industry shows how strong our leadership is, and that it would not be hard for the US to provide that one missing piece. 


Most people have little idea of what goes into a modern Internet network. Essentially, it consists of wireless transmitters and receivers in the cellphone, in a radio base station, larger devices at the base station to aggregate signals from hundreds or thousands of phones, and then devices to send those signals on an underground fiber-optic network back into the Internet. 

The aim of 5G is to crank up the speed from today’s 4G rate, where you typically get speeds in the low megabits-per-second range when you watch a YouTube video on your phone, to 5G levels, where you will get gigabit (billions of bits per second) speeds. At every point in the network--phones, radio towers, base stations, network core--this will require new devices that are faster, better, and more efficient in their use of electric power. To accomplish that, engineers have developed new ways to use the radio spectrum, enabling them to pump more bits more rapidly through the air, over dozens of channels, and then recombine them in the right order. It might sound like magic, and it almost is.

The devices that determine the speeds and capabilities of the network are microchips. In almost every case, the chip companies leading the way to developing these breakthroughs to 5G speeds are American. In fact, it is shocking (and a bit embarrassing for Europe and Asia) how overwhelming is the lead of American companies in this business. Some of the names of the chip companies now developing 5G technology are Qualcomm, Broadcom, Xilinx, Marvell, Altera, Intel, Nvidia, Maxim, Skyworks…and I could go on. 

If you took a bright red Sharpie and a map of Silicon Valley, and traced the oval that is formed by Highways 85 and 101, inside that oval, no more than 15 miles long and six miles wide, you will have circled the majority of the world’s chip design talent that is making 5G possible. And by the way, that oval includes a couple of the companies that you probably have heard of that will be customers for 5G chips, and their input into the process is highly valuable. Those customer companies are Apple Inc. and Google. 

It’s the Supply Chain, Stupid!
Why then do we keep hearing that China, and Huawei specifically, is leading the race for 5G? Huawei makes the systems that contain those chips. By systems, I mean boxes that get installed at radio towers and underground to carry the 5G signals. Those boxes are typically a few inches wide up to a couple of feet. They have the logo stamped on the outside in bright colors, so people get to know their names. 

The systems are not trivial, they do require design and engineering, but the intellectual property (IP) in the systems is a fraction of the IP in the chips. Huawei got to be the leader in systems because of an unnatural series of deaths in the US systems industry. 

Back in the 1990s, the US had three world-leading systems companies: Lucent Technologies, Nortel Networks, and Motorola. All three were decimated by the collapse of the first Internet boom in the year 2000. In the three years that followed (sometimes called the telecom nuclear winter), all of them shriveled. Lucent shrunk and was acquired by French networks maker Alcatel. Motorola withdrew from public telecom networks and today is mostly focused on smaller public safety and security networks. Nortel went bankrupt in 2004. (Full disclosure: I worked for Nortel from 1998 to 2000. I worked in the networking industry for a total of 16 years.)

These companies shriveled away for several reasons: first, they acquired too much debt and made too many bad acquisitions in the euphoric 1990s boom years; after March 2000, when the tide went out, everyone saw they were all swimming naked. Secondly, around the beginning of the awful “nuclear winter,” Huawei unleashed its campaign of copying the products of western systems makers and selling them to western telcos at prices some 20%-50% lower than the originals. But finally and most fundamentally, the systems makers shriveled because they were trapped between chipmakers (that were able to demand high prices and high profit margins for their innovative products) and the telecom companies that ruthlessly drove down the prices they were (and still are) willing to pay for complete networks. 

Caught in that squeeze, the three US network makers quit the business. European network makers, Ericsson and Nokia, are able to soldier on because their telecom companies are not as ruthless as American ones and their stock market is not as demanding. But those two European companies are not funded well enough, nor ruthlessly enough, to stand nose-to-nose with Huawei. Huawei has dominated the standards-setting, the patents, and other parts of the 5G process. Huawei has done this through tried-and-true tactics: dominating meetings, sending 20 people to a meeting when other companies send one or two; cajoling or bullying its customers  and suppliers to back its proposals in these meetings. It’s not about engineering breakthroughs, but it is effective.

The Trump White House is very concerned that Huawei 5G networks pose a security threat. Even if we successfully keep Huawei out of the US, major Huawei networks overseas pose a threat to US national security. They are right to be concerned. The nature of these networks is to carry additional channels for information that is never seen by customers or users. Usually those channels are used to update the control centers on the health of the network. But they can easily be made to carry other information secretly, and the telco would likely never know. 

If we want to preserve US leadership in technology, we need to think of this industry as a complete supply chain, not a bunch of companies selling individual products to each other. In the tech industry, they call this an “ecosystem.” The US needs a systems maker to keep our Internet networking ecosystem healthy. We can’t trust the Chinese producers. Our friends at Ericsson and Nokia are innovative companies, but sadly, not big enough, tough enough, or even physically close enough to sustain US technology leadership. 

We need to find a way to shoehorn into the ecosystem a new US-owned and US-based systems maker, for national security reasons and for economic reasons—because that systems maker will help keep that oval from Sunnyvale down to San Jose alive and healthy.  That oval is the beating heart of the US technology industry. 

How? The federal government has to step up. It has many tools for this fight. It can provide business to a new American systems maker, at generous profit margins. It can provide research funds—as it once did for Bell Labs--and in those days Bell Labs was the envy of the world. It can lean on the telcos to ensure they are paying decent prices for their networks. 

The problem with globalization is that it views everything as an isolated product that can be outsourced and sweated down to the lowest possible cost. It ignores the externalities--i.e. the effects on related industries--and it ignores the consequences over time, i.e. the failure of investment-intensive industries, if they are sweated too much. A globalist, to quote Oscar Wilde, knows the price of everything, and the value of nothing. 

The Chinese understand this, which is why they are desperately trying to build out a chipmaking industry to complement the network systems industry built by Colonel Ren, and eliminate their dependence on our chipmakers. We can’t stop them trying (although it would be nice if they tried to do it legally for a change). But we must build out our own.

It’s the supply chain, stupid!

No comments:

Post a Comment