Scenarios

Huawei is a $120 billion company with 200,000 employees and the unlimited backing of the Chinese state. It will survive.

The U.S. President and Secretary of State are attacking with everything short of actual bombing raids. Huawei sales went up this year despite the blockade. But sales will be affected by cutting off chips from TSMC, the only practical source for the 5 nm and 7 nm chips essential for the best cell phones.

It has stockpiled enough chips to last past the U.S. election into 2021 and maybe longer. Nikkei in May reported Huawei has two years’ worth of some chips in inventory, such as Intel CPUs. The U.S. can’t win the tech war against China without prohibitive losses, but no one in D.C. will admit that before the election.

Half of Huawei’s sales are phones, many of which require chips only possible from TSMC or Samsung.

Donald Trump will make almost any compromise to win reelection. It’s possible but not likely that he will end the blockade in return for China buying U.S. crops.

TSMC also has some leverage. Trump wants TSMC to go ahead with a $12B plant in Arizona. He trails in the state after winning it in 2016. Although the NY Times reported a done deal, TSMC has made clear it will not go forward without $billions in government giveaways. TSMC has applied for an exemption from the blockade, but will probably concentrate on a bigger subsidy.

I see four plausible scenarios:

Bright: TSMC resumes shipping to Huawei

Huawei phones do not pose a security risk anyone has identified. Neither do Huawei PCs and televisions sold in China. Historically, the U.S. almost never used secondary boycotts like the ban on TSMC sales to Huawei except for nation-states like Iran or North Korea. They almost certainly violate world trade rules, but the U.S. ignores the rules when they go against us.

With chips fabricated a TSMC, Huawei pulled to #1 in phone sales in Q2. Kunpeng is a killer chip, benchmarking better than the Intel I9. China Telecom has ordered 10,000 PCs with Kunpeng. The Google blockade cost over $10 billion in European sales, but Hongmeng/Harmony is catching up quickly.

Huawei has a huge program to develop new markets, including television and consumer goods. The auto division is a leading supplier of telematics for cars around the world, including Daimler in Germany. Auto is strategic for Huawei, which is building parts for driver assistance and automation. It promises to cut the market price for lidar by more than half.

Huawei’s $17 billion research budget is delivering enough products the company could do well without phones. If the phone products are unimpeded, Huawei will pass $200 billion in sales in 3 or 4 years.

Calm: TSMC via Mediatek or Qualcomm chips

Huawei had a deal in place to buy 100 million Mediatek chips if TSMC wouldn’t deal with HiSilicon. HiSilicon chips are somewhat ahead of Mediatek but it is close enough in performance for practical purposes. Samsung was also interested in selling chips.

Presumably, buying 100 million produced a very low price. Mediatek’s markup would probably add $2 billion to $4 billion to Huawei’s 2021 chip costs. That’s significant, but manageable in a $60 billion product.

Trump or a successor might allow companies to sell chips to Huawei. Intel and Qualcomm, cut off from billions in sales, are throwing everything they have to get the blockade eased. Qualcomm officially reported over $7 million in lobbying, Intel over $5 million. Most DC lobbying is not reported, so the real total is often three or four times as high.

The Democrats have made a point of being “tougher on China” than Trump. Mark Warner and Chuck Schumer seem to be Chinaphobes, but others may just be posturing. Certainly, nothing before the election and hard to guess about 2021.

Challenged: Limited to SMIC 7 & 12 nm chips and eventual EUV

SMIC, China’s leading foundry, is producing mobile chips for Huawei at 14 nm and will deliver 12 nm by yearend. It’s raised $9 billion from investors and the government, nearly all set to be spent on raising capacity in 12-24 months.

12 nm going to 7 nm should be fine for most of Huawei’s product line. 5G radios, for example, might require two or three 12 nm chips to replace the TSMC 7 nm chip that could become unavailable. That might add $10-30 to the cost, require somewhat more heat and power, and warm things up a bit. That’s manageable in a radio which sells for $3,000 to $20,000. Most other Huawei products should be fine as well.

But it’s impractical to add more chips and power demand to a mobile phone. If Huawei becomes dependent on Chinese chips, phone sales, especially at the premium end, will drop a great deal.

Chinese chipmaker UNICOM is delivering a 5G chip using TSMC’s 12 nm. See Coolpad $199 5G phone with Unisoc Ziguang Zhanrui Chinese chip It could be manufactured at SMIC and Huawei certainly has developed similar. It should be fine for low-end and possibly mid-range phones.

SMIC is testing a process that should be similar to TSMC’s early 7 nm. That was used for Huawei’s first 5G chip, the Kirin 980. Two years later, the Kirin 1020 and the new Apple 5 nm chip are the stars of premium phones. They are made using ASML’s $200 million EUV machines, which the U.S. is preventing the Dutch company from shipping to China.

Almost everyone believes it will take a miracle for China to develop its own EUV machine in less than 5-10 years. I’ve seen China deliver many miracles in technology and hope a Chinese company becomes a second source of EUV.

Struggle: SMIC gives in to U.S. pressure

Donald Trump believes he has the right to prevent SMIC or other Chinese fabs from selling to Huawei because some of their equipment was made in the United States. Some analysts believe that is likely, although it’s almost certainly a breach of international law.

I think it highly unlikely the U.S. can prohibit a Chinese foundry from selling to a Chinese company in China. I’d expect any rational American regime to avoid raising the issue.

Every Chinese schoolchild learns about the national shame of the Uneven Treaties imposed after the Opium Wars by the British, French, Germans, and Americans. The Chinese could not match the weapons the Europeans had developed in three centuries of virtually unceasing warfare. 20% of Germans died in the 30 Years War. England and France were at war most of the time between 1756 and 1815.

The Chinese will not tolerate anything like that today. Anyone who doesn’t realize that knows nothing of Chinese history.

The Chinese have already begun construction of several fabs without any American components. Trump is simply wrong that chip plants cannot be built without U.S. tools. ASML in Holland, Tokyo Election in Japan, and dozens of other companies outside the U.S. make semiconductor gear.

I am systematically going through more than a dozen different types of equipment needed to make advanced semiconductors. There are non-U.S. choices for everything.

China has budgeted over $200 billion for semiconductor manufacturing. There have been several successes and some catastrophes. Wuhan Hongxin may go bankrupt after an $18 billion investment. But as the Americans discovered in Korea, Communist China is able to learn from mistakes and persist.

The U.S. no longer dominates the world of technology but some will learn that the hard way.

7 Comments

  1. […] unlikely the U.S. can prohibit a Chinese foundry from selling to a Chinese company in China,” he writes. “I’d expect any rational American regime to avoid raising the issue.” Given the level of […]

    Reply
  2. […] U.S. can prohibit a Chinese language foundry from promoting to a Chinese language firm in China,” he writes. “I’d anticipate any rational American regime to keep away from elevating the difficulty.” […]

    Reply
  3. […] unlikely the U.S. can prohibit a Chinese foundry from selling to a Chinese company in China,” he writes. “I’d expect any rational American regime to avoid raising the issue.” Given the level of […]

    Reply
  4. […] unlikely the U.S. can prohibit a Chinese foundry from selling to a Chinese company in China,” he writes. “I’d expect any rational American regime to avoid raising the issue.” Given the level of […]

    Reply
  5. […] unlikely the U.S. can prohibit a Chinese foundry from selling to a Chinese company in China,” he writes. “I’d expect any rational American regime to avoid raising the issue.” Given the level of […]

    Reply
  6. […] unlikely the U.S. can prohibit a Chinese foundry from selling to a Chinese company in China,” he writes. “I’d expect any rational American regime to avoid raising the issue.” Given the level of […]

    Reply
  7. […] unlikely the U.S. can prohibit a Chinese foundry from selling to a Chinese company in China,” he writes. “I’d expect any rational American regime to avoid raising the issue.” Given the level of […]

    Reply

Say something

Scroll to top