Even though artificial intelligence (AI) has been in development for years, experts believe it is still in its infancy, especially for driving AI use in chips.
One company that hopes to capitalize on AI chip growth is Intel. The legacy chipmaker will announce Q4 2019 earnings on Thursday and is drawing an unusual amount of positive attention from several financial analysts, largely because of Intel’s AI achievements and prospects.
When its last reported earnings in October, Intel raised its full-year 2019 revenue outlook to $71 billion, up by $1.5 billion from July.
The increase came even as Intel has suffered continued criticism for falling behind on PC client chip shipments, with CEO Bob Swan admitting, “We are letting customers down.” In November, Intel even sent customers an open letter apologizing for the PC chip shipment delays.
RELATED: Intel apologizes for PC chip shipment delays in open letter
In December, Dell cited the Intel chip shortage as contributing to its Q3 decline in PC revenues. Dell CFO Tom Sweet said the PC maker was evaluating AMD chips in light of the shortage.
RELATED: Dell mulls using more AMD chips as Intel chip shortage drags on
Dell ended up finishing 2019 in third place to Lenovo and HP, but the overall PC sector did well in the fourth quarter and all of 2019 thanks chiefly to Windows 10 upgrades. With volumes of PCs up by 2.7 %, it was the first full year of PC growth since 2011, according to IDC. It was that surge that helped exacerbate Intel’s PC chip shortage.
RELATED: Lenovo, HP and all PCs finish 2019 on wild ride
Even with Intel’s PC chip shipment delays, Intel’s focus on AI chips offers a hedge against AMD competition and a doorway to long-term growth.
“AMD has no credible AI solution as yet,” wrote analyst Arne Verheyde in a recent overview in Seeking Alpha. “As the leader in AI, Intel is well-positioned to capitalize from edge to cloud with its comprehensive silicon portfolio.”
Verheyde described AI as a “key catalyst” for valuing Intel as a growth company.
Analysts at Citi also caught the optimism over Intel recently, raising their price target for Intel from $53 to $60 on Thursday. At noon on Friday, Intel shares traded at $59.40.
Verheyde’s overview of Intel AI looks into several competitors in AI chips for either training or inference or data center/edge acceleration. They include: Nvidia (V100 and T4 chips for training and interference), Qualcomm (Cloud 100 for inference), Huawei (Ascend 310 and 910 for edge and data center), Google (TPU for inference) and Baidu (Kunlun for data center).
“Intel intends to defend its established position by increasing the AI performance of its Xeons each generation while developing its own line of dedicated AI accelerators alongside the Movidius chips for the edge,” Verheyde wrote. Intel has purchased Nervana, Movidius and, recently, Habana in its quest for AI superiority.
Against Nvidia, he said that Intel’s NNP-T and Gaudi neural network processors and upcoming Ponte Vecchio GPU (due in late 2021 on a 7nm process) can grab data center AI training future revenues from Nvidia.
AI has generated $3.5 billion in revenue for Intel in 2019, when IoTG and Mobileye are included, which makes it a larger player in AI than Nvidia, he said. “If Intel beats Nvidia at what is seen as Nvidia’s market, investor sentiment will further swing in Intel’s favor,” Verheyde concluded.
Part of what is driving AI revenues are emerging examples of successful AI experiments and rollouts. In one example, Big River Steel is using Noodle.ai applications to improve productivity and save on energy costs.
At John Deere, work is proceeding on commercialization of a machine called See and Spray that uses AI to accurately target weeds in crops. It relies on an embedded Nvidia Jetson Xavier graphics processor that has been trained to know a weed from a crop plant.
RELATED: Deere shows off smart weed control machine that relies on AI with Nvidia chip
What makes See and Spray dramatically different from the Big River Steel example of AI is the use of the Nvidia chip in Deere’s edge herbicide device. It can operate independently (or nearly so) of the cloud or a data center, using stored data that it has been trained to differentiate a weed from a healthy plant and infer the difference in the two types of plants as it passes among the crops, spraying only the weeds.