Xperi Eyeing Tier 2 OEMs for TiVo Stream Platform Due Next Year

Xperi is eyeing opportunities for growth from the platform’s search, discovery and monetization “that we believe will be significant as we move the technology stack off of the dongle directly into connected TVs,” he said. Stream is “content-specific” and the platform integrates live and over-the-top video content. The dongle's installed base is in the “hundreds of thousands” in its first year, Kirchner said.

The company hopes to leverage its legacy relationships with TV brands in the U.S., after buying TiVo last year (see 2005070052), Kirchner said, giving a late-2022-2023 time frame for Stream-based connected TVs. Commenting on whether it’s looking at Tier 1 OEMs, Kirchner said the initial focus is on the second-tier makers due to “competitive dynamics.” For Xperi, “it all comes down to the quality of the solution and the monetization it drives” through revenue or advertising share. The company has had “productive discussions” and “there is both demand and interest in working with us.”

Xperi has a “sizable beachhead” in the pay-TV space, providing services to Tier 2, Tier 3 and Tier 4 providers, said Kirchner. He believes pay-TV subscription churn will eventually “bottom out.” MVPDs are upgrading offerings primarily through IPTV, an area where Xperi is investing to drive growth, he said. TiVo’s purchase of MobiTV assets in a Chapter 11 bankruptcy this month gives it some 100,000 subscribers and an “end-to-end IPTV delivery system,” Kirchner said. Moderate subscriber declines will continue in pay-TV in the next three to five years, he said, “but you’re going to see a revenue offset as you see people shift into IPTV, and we can monetize.”

Xperi’s IP licensing business has over 5,000 patents, and many from the pay-TV business are relevant to OTT-based delivery, Kirchner said. The company is just starting new licensing cycles that will add to revenue growth, he said.

Commenting on its Perceive subsidiary that’s developing edge-based machine-learning technology, Kirchner said it’s initially targeting security cameras, with future uses in mobile, white goods, automotive and wearables. The technology “over time will find its way into lots of things as there’s a desire for greater intelligence on the edge across a number of applications.” First products are due in early 2022, he said.

Perceive had hoped to have products by year-end, but its customers have been affected by chip shortages, Perceive Vice President-Marketing David McIntyre told us. “The biggest issue is, lead times have gotten a little longer.” Its customers “have to assemble chips for many people, and they can only release the product with the slowest chip that shows up.” Shortages have stretched out timelines “a little bit,” but Perceive’s Ergo chip isn't affected, he said.

The company’s edge AI inference processor is targeted to CE devices, enabling them to run machine learning models, or neural networks, with power as low as 20 milliwatts, McIntyre said. In the first implementation, the processor will take information from a camera, mic, gyro or thermal sensor and transform it into perceiving so it knows what it’s seeing. “Instead of seeing pixels, and then maybe sending a firehose of pixels to the cloud to figure out it’s a dog, the device should know it’s a dog, or it should know it’s interacting with a person,” he said. “The device itself is intelligent enough to know who it’s dealing with and understand its environment.”

Videoconferencing, a hot category accelerated by work-at-home trends, could be an opportunity for Perceive, said McIntyre. Blurring or background-swapping features available for Zoom use a lot of a PC’s processor resources, he said. Those, along with relighting or face improvement features, could be done in camera instead using Ergo’s AI capabilities, he said. Noise reduction in the chip could remove keyboard tapping sounds. The chip could also integrate voice or gesture control in a conference room camera, he said.