As China's Huawei looks for ways to become less dependent on American technology, the results could end up being very mixed.
Here's a quick look at some of the fields where Huawei has been trying to develop alternatives to U.S.-developed technology amid U.S. trade restrictions, and some of the challenges these efforts face.
Operating Systems
With U.S. export restrictions putting its future access to Alphabet/Google's (GOOGL) version of Android and other software from U.S. developers into question, Huawei has been prepping an operating system known as HarmonyOS that's promised to support the porting of Android apps. However, as tech analyst Patrick Moorhead has observed, Android apps typically rely on a number of programming interfaces (APIs) for interacting with things such as cameras and sensors, which in turn complicates the porting process.
Since the core Android OS is open-source, Huawei could continue using it without Google's support. However, if it lost access to Google's Android distribution, Huawei would also lose access to the Play Store and the many popular Google apps and services that come bundled with Google's version of Android (Google Assistant, YouTube, Gmail, Maps, etc.). That in turn would do tremendous damage to Huawei's smartphone sales outside of China, where Google's services are mostly blocked.
Mobile Processors and Modems
Though also relying on mobile processors and modems from the likes of Qualcomm (QCOM) and Taiwan's MediaTek, Huawei's mobile device business now also heavily relies on processors and modems that were internally developed by its HiSilicon unit. HiSilicon's R&D work has recently extended to 5G modems and base station processors.
There are also a few other Chinese mobile processor and/or modem developers, such as Unisoc (formerly Spreadtrum) and Rockchip. However, when it comes to RF chips, both Huawei and other Chinese OEMs are still quite dependent on U.S. suppliers such as Broadcom (AVGO) , Skyworks (SWKS) and Qorvo (QRVO) .
Chip Intellectual Property and Design Software
In May, British chip IP giant ARM Holdings (owned by Japan's SoftBank) made waves by announcing that it's suspending business with Huawei due to U.S. restrictions, while citing the fact that some of its technology was developed in the U.S.. Since CPUs relying on the ARM instruction set (if not fully-fledged CPU cores designed by ARM) power nearly all of the world's smartphones and billions of other devices, any extended loss of access to ARM's technology, which also includes GPU designs and other IP, would deal a massive blow to HiSilicon.
Along similar lines, losing access to electronic design automation (EDA) software provided by U.S.-based Cadence (CDNS) and Synopsys (SNPS) (by far the world's two biggest chip design software providers) would be a major problem for Huawei/HiSilicon. In May, Japan's Nikkei reported that Synopsys had told its employees to suspend the delivery of software updates to Huawei.
Huawei is looking to lower its dependence on ARM by embracing the open-source RISC-V instruction set architecture. And the company has also signaled that it's looking for alternatives to Cadence and Synopsys' tools. However, given the enormous R&D investments that ARM, Cadence and Synopsys have respectively made over the years, as well as the ecosystems that they've built up, any effort to develop serious alternatives will take many years of work.
AI/Deep Learning Chips
Last week, Huawei unveiled the Ascend 910, a powerful server accelerator that's focused on training AI/deep learning models to do things such as translate text, detect objects within photos and decipher voice commands. This followed last fall's unveiling of the Ascend 310, a less powerful chip meant for both AI training and inference (the less demanding task of running trained AI models).
Nvidia has long been dominant in the training accelerator market -- Chinese tech giants Alibaba, Tencent and Baidu are among its clients, and the company's large developer ecosystem and software investments help act as a moat. In addition, many of the firms looking to challenge Nvidia, such as Intel (INTC) , AMD (AMD) and private Cerebras Systems, are also U.S.-based. While battling Nvidia in this space won't be easy, one can imagine Beijing both encouraging Huawei's AI silicon investments and encouraging Chinese firms making AI investments to use its chips.
Alphabet is a holding in Jim Cramer's Action Alerts PLUS member club. Save 57% With Our Labor Day Sale. Join Jim Cramer's Action Alerts PLUS investment club to become a smarter investor! Click here to sign up!
