Femtosense raises funds to power AI inside consumer electronics

Femtosense raises funds to power AI inside consumer electronics.

Large AI systems, such as OpenAI’s GPT-3, may make headlines, but practical constraints such as power consumption and cost, rather than capabilities, often determine where AI is deployed. Highly capable systems are typically relegated to the cloud because they are too complex to run on edge devices with inferior hardware and limited connectivity.

This is something Sam Fok laments. He is the CEO of Femtosense, a startup that is developing edge hardware to enable AI processing in low-cost consumer electronics. Femtosense, founded by members of Stanford’s “Brains in Silicon” group, which seeks to reverse-engineer how the brain uses relatively little data to learn, aims to tackle use cases such as noise suppression and speech enhancement for hearing aids and earbuds, as well as security cameras, TVs, and cars.

Femtosense today closed an $8 million round in support of the vision, valuing the company at $27 million and bringing its total raised to $11 million. Fine Structures Ventures led the round, with J2 Ventures, SV Pacific Ventures, Quest Venture Partners, Amino Capital, Sand Hill Angels, and Gaingels also participating.

The final day of “Overwatch 1” is today

“Hardware developers continue to create new hardware to support existing workloads, while algorithm developers optimize for existing hardware.” “There’s an inherent bias in what gets built toward what already exists,” Fok told TechCrunch via email. “Femtosense is fairly unique in that it is designed with the intention of pushing the hardware and algorithm design space into sparse computing that has yet to be exploited while still working well with existing technology.”

The SPU-001, Femtosense’s first-generation processor, has yet to be demonstrated, and mass production is at least several months away (sometime in 2023). However, Fok claims that it will allow product developers to run 10 MB AI models at the same power required to run 100 KB models. (The size of an AI model is usually proportional to its sophistication; smaller models are less accurate.)

Femtosense raises funds to power AI inside consumer electronics
Femtosense raises funds to power AI inside consumer electronics

According to Fok, a company using the SPU-001 could run an AI-based noise cancellation algorithm that, when combined with applications such as speech-to-text, provides a better user experience (think a voice-controlled TV remote that can better understand you in loud surroundings). Such a setup could also drive personalization, allowing important safety sounds — such as a siren, honking, or crying — to pass through while the rest is noise-canceled.

“The earbud market is expanding, OTC hearing aids will become a new market, and voice interfaces are becoming more common.” In general, consumer product developers are competing to include more AI capabilities in their offerings, according to Fok. “The market is ready to embrace AI.”

Google Pixel 7 preorder incentives may include the Pixel Watch and Pixel Buds Pro

The main rivals of Femtosense. Fortunately, the startup stands to benefit — and is already benefiting — from a boom in the semiconductor industry. According to PitchBook data, VC funding for global semiconductor startups more than tripled year over year in 2021, with $9.9 billion invested across 170 deals.

The administration’s recent rule allows access to over-the-counter hearing aids — the type of electronics that the SPU-001 is designed to power. The policy change implemented in August allows consumers to buy hearing aids directly from stores without requiring a medical exam, prescription, or fitting by an audiologist, creating a potentially lucrative new market category.

Despite the go-to-market challenges that lie ahead — fabricating a new chip isn’t easy — Fok claims that Femtosense, which is still in the early stages of development, already has several enterprises and public sector customers lined up. (Previously, the company was awarded a $2.1 million contract from the US Office of Naval Research.) The 10-person company intends to hire two new employees this year and is already planning the next generation of its chip, which Fok describes as “general purpose” hardware designed to accelerate applications such as computer vision and natural language processing.

“By raising funds immediately prior to the [current economic headwinds], Femtosense is well-positioned to weather the [current economic headwinds],” Fok said.

Read also:

Leave a Reply

Your email address will not be published. Required fields are marked *