$Hundreds of millions in extra chip costs prove Open-RAN not quite ready

Tareq Amin’s Rakuten is the most advanced Open-RAN and virtual carrier on earth. But he wasn’t able to get there using standard parts and equipment.

Rakuten pays hundreds of millions of dollars in non-recurring engineering fees to chipmakers like Qualcomm to obtain the components it needs. …We cannot find the right material at the right cost, the right architecture, to address the future requirements for 5G radios

Tareq Amin Rakuten to John Hendel of Politico

That’s an enormous sum, enough to design state-of-the-art chips. I would guess the extra cost ate up all the expected cost savings from the initial deployment. (Below) Important: Please don’t infer from this that Open-RAN is a failure, especially for new networks. Most of these problems will be solved. Meanwhile, don’t believe the hype.

Alex Choi of Deutsche Telekom is enthusiastic about Open RAN and the other buzzwords as near-future technologies, still with challenges. That’s the near-universal consensus of the top network engineers. That doesn’t mean the new networks shouldn’t pioneer, but I’ve been seeing some unfortunate datapoints:

  • Vodafone CTO Scott Petty has been one of the most enthusiastic supporters and is doing some deployments. But Laurie Clarke of New Scientist quotes Petty, “We believe by 2023, we may be able to deploy some scale in the rural parts of our network, but it will take until 2025 to be able to deploy at real scale in our denser urban and suburban areas.”
  • AT&T was the first strong backer of SDN/NFV/Ecomp, pouring a fortune into open software. It is quietly cutting back. A friend has been receiving resumes from senior people at AT&T, expecting layoffs.

Contrary to general belief, the initial saving from open and virtual RAN is modest. The hardware is cheaper, but hardware is only a small part of the network cost. For large carriers with bargaining power and good negotiators, the price of the hardware is already low.

China Mobile & China Telecom are paying US$23,000 per cell. The antenna, power supply, and a high-performing processing unit will still be required. How much could they really save if they bought radios from Mavenir or Jio/Radisys?

Smaller telcos, like AT&T or Telefonica, order closer to 10,000 cells. Nokia & Ericsson are masters at extracting maximum revenue from customers but if pressed will give great prices on the initial purchases.

Over time, the more flexible systems should be much cheaper to upgrade. Moore’s Law may be slowing, but the systems in five years will be far more capable than today. If open systems deliver on their promises, the upgrade should be cheap and easy. (That’s still unproven.)

Carl Russo at Calix has produced some of the best SDN systems, Verizon tells me. He tells me carriers who just look for initial savings will not reap the full benefit of SDN. Only those who use the flexibility and integrate better management, especially of customer offerings, receive a full return.

Nearly all telcos upgrading and densifying existing systems will likely follow a similar timeline as Vodafone, limiting new systems for several years. New builds are almost all choosing Open.

All of this is speculation until we have good data from the field, of course.

2020-2021: Networks Doing Fine in the Time of Corona

Amazingly, there have been close to zero reports of additional traffic causing important problems for people at home. The networks are successfully handling 20-50% traffic increases while still staying fast enough for almost all practical purposes. Traffic has now plateaued and speeds improving weekly.

Median landline download speeds in 90% of U.S. cities were within 20% of norm. The majority of cities tested over 40 Mbps on downloads. Virtually all connections were fast enough for three 4 Mbps HD streams except on DSL links that already had problems and should have been upgraded years ago.

Wireless traffic was actually down in some places and rarely increased by more than 20%. Many people at home used Wi-Fi instead of cellular, leaving wireless data networks in good shape. 4G & 5G wireless technology has improved enormously in the last few years. Verizon’s cost to carry a bit has dropped ~40% annually. Speeds are now often 50-150 Mbps down.

Continue reading

$23,000 for 5G Base Stations, $56,000/cell Complete. Quantity 200,000

China Mobile is upgrading over 250,000 cells to 2.5 GHz, 100+ Mbps 5G in 2020. It projects a total capex expenditure for 5G of ~US$14 billion. That’s about US$56,000 per upgraded site. It will reach between 500 million and 700 million Chinese, about half the country. China Telecom and China Unicom are doing a slightly smaller cooperative build at a similar cost.

It is also buying 230,000 base stations for ~US$23,000. Huawei won 55% of the order, ZTE ~30%, Ericsson ~10%, and Datang 5%. Nokia Shanghai Bell was shut out.

Continue reading

5G Phones Will Keep Getting Cheaper

In 2020, the difference in 5G and 4G manufacturing costs will fall to US$15-35 per phone. Xiaomi in January is offering a decent 5G phone for $285. The 4G version is only $55 less.

Huawei and Samsung, who make their own parts and have enormous volume, will have an advantage. Xiaomi, Oppo, and Vivo believe their prices must be especially aggressive or they will lose market share.

As the price goes down, more people will choose 5G over 4G phones. In the fall of 2019, a third of Korean sales have been 5G phones, as the carriers are using subsidies to bring down the price. LG Uplus has been particularly aggressive, seizing what it considers a “once-in-a-decade” opportunity to gain share with 5G.

Although the first 5G phones cost about US$1,000, by October prices were down to $536 (Oppo) and $570 (Vivo). In January, Xiaomi will sell 5G for $285.Prices will continue to fall after that, although the $50-150 phones are still a while off.

Continue reading