In recent months Qualcomm, which happens to be one of the leaders in the field of processors and chipsets that make up smartphones, has been on the lips of the entire sector. A takeover bid by a competitor, Broadcom, was deactivated by the president of the United States himself, Donald Trump , when everything seemed to be heading for a dead end.
But in this blog we are not going to talk about this story, but about how Qualcomm is embarking on a strategy in the field of artificial intelligence (AI) that can keep it away from the big smartphone manufacturers on the market, such as Apple, Samsung or Huawei. , which happen to be the best sellers worldwide, and by far. Because? The latter, obsessed with leading the smart mobile market, with increasing capabilities in AI, have developed their own neural chips and cloud capabilities to integrate them into their flagship devices. This is the case of Apple with its A11 Bionic processor, which incorporates the iPhone X; Huawei with its Kirin 970, which includes the Mate 10 and the P20; or Samsung with its Exynos 9810, which ships the Galaxy S9 and S9 +.
To understand the reason for this particular strategy and to understand the path that Qualcomm has chosen, we have spoken with Gary Brotman, the head of artificial intelligence at the American company. Gary welcomed us last February at his booth at the Mobile World Congress, and attended us just a few days later after announcing AI Engine, the engine that will provide all the artificial intelligence capabilities to Qualcomm’s Snapdragon processors that carry smartphones. of the different manufacturers of smart terminals.
Gary Brotman commented to Very Interesting that “the model that others (Huawei, Apple or Samsung) are following of developing cloud-based neural processors requires a lot of computing power, a lot of data and a lot of time. What we do is provide the right tools so that developers can have access to optimized performance of the hardware and develop the system in a general way. We also believe that consumers should not care whether the processing is done locally or in the cloud. They must first know something to really experience why it is best for them. Qualcomm believes that if the intelligence resides in the device, the processing will be faster and the end-user experience will be perfect. “
With these words it can be deduced, although Brotman does not say so expressly, that Qualcomm does not think at the moment to develop a neural chip that is capable of recognizing objects in the cloud, improving the photographs taken by the cameras of the devices or intervening in the interface of smart mobile voice. “Space on a smartphone chip is so limited that we don’t want to sacrifice anything for a unit that only has one function. It’s no fun waiting for, and maybe not even getting, a response from a digital assistant or smartphone because there’s no connection to the Internet. People are impatient and we are increasingly demanding of real-time performance, “says Brotman.
And this is what some smartphone manufacturers have done, such as Xiaomi, OnePlus, Oppo, Motorola, ASUS, ZTE or LG, who have not developed their own neural processors like Apple, Huawei or Samsung and have announced that they will take advantage of the AI Engine. of Qualcomm’s Snapdragon to optimize the artificial intelligence applications of its smart terminals.
Gary Brotman also does not believe, so he explains, that Qualcomm should now dedicate itself to researching and developing its own neuromorphic chips, or processors that will mimic the behavior of the human brain, as does another of its competitors, Intel, which has announced that this year it will launch its first processor with these features.
What Qualcomm is immersed in is developing the intelligent brain of connected vehicles and of future autonomous cars. Although we asked him, the head of AI at Qualcomm did not want to reveal in which projects and with which partners his company is embarking. However, in December 2017 the American company was authorized by the California authorities to test Ford’s autonomous cars in which it has introduced its 9150 chip for C-V2X, a connectivity solution that allows vehicles to exchange information with other cars. and with pedestrian devices.
Qualcomm also continues with its idea of acquiring NXP, the company specializing in semiconductors for connected cars, and in February of this year made an offer to acquire it for about 44,000 million dollars. Also in early 2018 Qualcomm Technologies, a subsidiary of Qualcomm, announced a partnership with Continental, Ericsson, Nissan, NTT DoCoMo and OKI to conduct their first C-V2X connected vehicle trials in Japan.
Brotman, despite all this, is skeptical in our conversation that 5G networks are going to be the deciding factor for autonomous driving to spread. “Before the regulators will have to solve other problems that arise.” Who knows, but until the regulatory obstacles mentioned by Qualcomm’s head of artificial intelligence are resolved, the same thing is necessary to wait until 6G, in 2038, for autonomous driving to be general, as we talked about in one of the first articles of this blog.
