forget about the noisy supercomputers and gigantic data centers that today give AI its lead in the world. Nvidia. According to the CEO of QualcommBrazilian Cristiano Amon, a new front in the “chip war” is emerging.
In an interview with the television channel CNBCthe executive stated that the company is betting on more than 40 different “wearable” devices with AI, including glasses, jewelry, headphones with cameras, brooches and watches.
The strategy is quite different from the Qualcomm has done so far, since, instead of focusing on the silicon that powers traditional smartphones (its main territory), the company now wants to shape chips into new formats.
Study abroad
Upgrade your career!
To begin paving this path, the Qualcomm announced two new solutions: Snapdragon Reality Elite, a platform for mixed reality glasses designed to run AI on the device itself, and START (acronym for “Scalable Turnkey AI-Ready Toolkit”), a combination of hardware and software modules aimed at accelerating the creation of AI devices, starting with smart glasses.
The movement was anticipated by Cristiano Amon last year, during the Summit Lisboa websitewhen he told journalists that the chip battle is just beginning. According to the CEO, the real competition is in processing efficiency versus energy consumption, something the company already does well. “We came from processors that are used to working with batteries, low battery consumption”, he said at the time.
From data centers to edge computing
To understand the strategic movement of Qualcommit is necessary to look at the current architecture of the technology market. Today, the “chip war” is concentrated in data centers, with companies massively purchasing GPUs from Nvidia to train gigantic models in the cloud.
But one of the industry’s biggest bets is that an increasing part of artificial intelligence will start to be executed at the edge of the network, on devices close to the user, complementing the processing carried out in data centers. According to forecasts from institutes such as IDC (International Data Corporation)the next stage of this race could be marked by the expansion of AI to “edge computing” (called “edge computing”, in the English term).
In practice, edge computing distributes processing between different layers. Some of the simplest and most urgent tasks happen directly on the user’s device (such as a smart glasses or smartphone) while more complex operations continue to be executed on servers in the cloud. In other words, instead of sending absolutely everything to the cloud via the internet, which generates delays (latency) and requires constant connection, the technology divides the work.
Devices such as smartphones and smart glasses are responsible for faster and more contextualized responses, while large data centers continue to be responsible for activities that require enormous computational capacity, such as training AI models, storing large volumes of data and executing complex operations.
Continues after advertising
It is precisely in this scenario that the Qualcomm sees an opportunity. Unlike the Nvidia, whose leadership was built on GPUs for servers and data centers, the Qualcomm developed its expertise in chips for smartphones, tablets and other mobile devices. If the next wave of AI requires local processing on compact, low-power equipment, the company believes it is well positioned to compete.
Interestingly, the Qualcomm does not want to give up either side of the coin, as the company itself is already trying to enter the data center market with the creation of the brand Dragonflyannounced at the beginning of the month.
Content produced by
Continues after advertising
