US chip giant Qualcomm has set up a $100 million investment fund to support artificial intelligence (AI) startups as it looks to bring AI technology into a range of mass-market devices and cut out the cloud.
Known as the world's biggest designer of communications chips for smartphones, Qualcomm Inc. (Nasdaq: QCOM) said the fund would build on its own research into low-power processing and connectivity, which it considers "essential" to AI. Most of the funding will go toward companies developing AI for autonomous cars, robotics and machine-learning platforms.
Qualcomm also announced that an Israeli startup called AnyVision would be the first company to receive financial support from the new fund, although it did not say how much it will invest.
AnyVision is developing AI systems that can identify people and objects. By installing its technology in devices, it hopes to mitigate some of the privacy concerns about storing data in central cloud facilities.
Qualcomm, meanwhile, is reported to have argued that using the cloud to support AI is too "computationally intensive." Given its focus on end-user devices, and relatively weak position in data centers, it clearly has a vested interest in ensuring the device becomes the main "host" of AI capabilities.
Keeping AI out of the data center could help Qualcomm to sell the more sophisticated and pricier chips that will be needed to handle AI processing on devices.
"As a pioneer of on-device AI, we strongly believe intelligence is moving from the cloud to the edge," said Steve Mollenkopf, the CEO of Qualcomm, in a company statement about the new investment fund. "Qualcomm's AI strategy couples leading 5G connectivity with our R&D, fueling AI to transform industries, business models and experiences."
However, its latest initiative is bound to bring Qualcomm into conflict with other technology giants that have similarly put AI at the forefront of company strategy.
China's Huawei, the world's biggest maker of telecom networks and a growing force in the smartphone sector, unveiled a new line-up of AI chips in October under the "Ascend" brand. While the Ascend 910 is aimed at the data center market, the Ascend 310 is intended for gadgets including smartphones -- ensuring Huawei Technologies Co. Ltd. has covered all the various bases.
Ken Hu, Huawei's rotating chairman, says one of his main goals is to drive down AI costs so the technology becomes mainstream. "We are trying to make computing power more affordable," he said during a Huawei conference in London last week. "Nowadays it is still very expensive and that is a big bottleneck for AI." (See Bill Shock: Orange, China Telecom Fret About 5G Energy Costs.)
While some of the most game-changing AI innovations have come from the startup community, analysts may be skeptical that Qualcomm's $100 million funding commitment can really make a difference. In February this year, Huawei promised to spend between $10 billion and $20 billion annually on R&D starting this year, although it has not indicated how much of this will go specifically toward AI. (See Huawei Commits Up to $20B for Annual R&D, Fleshes Out AI Pitch.)
Qualcomm spent $5.6 billion on total R&D in its last fiscal year. The company is under pressure to find alternative sources of revenue, with its licensing business under attack from regulatory authorities and threatened by a legal battle with iPhone maker Apple.
In the last fiscal year, licensing revenues dropped a fifth, to about $5.2 billion, while revenues from chip sales were up 5%, to $17.3 billion.
On a GAAP basis, Qualcomm reported a net loss of $4.9 billion, compared with a profit of $2.5 billion a year earlier.
— Iain Morris, International Editor, Light Reading