Big AI systems like OpenAI’s GPT-3 may grab the headlines, but the practical limits of power consumption and cost — not capacity limits — often determine where AI is deployed. For the most part, high-capacity systems are relegated to the cloud because they are too complex to run on edge devices with poor hardware and limited connectivity.
That’s the case, laments Sam Falk. He is the CEO of Femtosense, a startup developing edge hardware designed to make AI processing viable for low-cost consumer electronics. Founded by members of the “Brains in Silicon” group at Stanford, which seeks to reverse how the brain uses relatively small amounts of information to learn, Femtosense aims to address issues such as noise cancellation and hearing aids and headphones, as well as speech enhancement. Security cameras, TVs and cars.
To support the vision, Femtosuns closed $8 million today, bringing the company’s $27 million valuation to $11 million. Good Structures Ventures is led by J2 Ventures, SV Pacific Ventures, Quest Venture Partners, Amino Capital, Sandhill Angels and Gengels.
“Hardware developers still build new hardware for existing workloads, and algorithm developers optimize for existing hardware. There’s an inherent bias towards building for what’s already there,” Fok told TechCrunch in an email interview. Space computing is pushing and working well with current technology that is yet to be used.”
Image Credits: Femtosense
Femtosense’s first-generation processor SPU-001 has yet to begin rolling out, and mass production is at least several months off (sometime in 2023). But Fok says it allows product developers to run 10 megabyte AI models with the power it would normally take to run 100 kb models. (The size of an AI model often correlates with complexity, smaller models tend to be less accurate.)
Fok says a company using the SPU-001 can run an AI-based noise-cancellation algorithm, providing an improved user experience when integrated with speech-to-text apps (think a voice-controlled TV remote that might help you better. In a loud environment). Such a setup can also bring about personalization, for example, when sounds important to safety – such as sirens, screams or cries – are canceled out by the rest of the noise.
“The headphone market is growing, over-the-counter hearing aids are becoming a new market, and voice communication is becoming commonplace. In general, consumer product makers are racing to add more AI capabilities to their offerings,” Fok said. “The market is primed for AI adoption.”
Folk sees Syntiant, Greenwaves Technologies and AONDevices as Femtosense’s main competitors. Fortunately, the startup has an advantage — and it is. Profitability – An important factor in the semiconductor industry. In the year By 2021, VC funds for global semiconductor startups more than tripled year over year, with $9.9 billion invested in 170 deals, according to PitchBook data.
Another finding in FemtoSense’s favor is the U.S. Food and Drug Administration’s recent rule allowing over-the-counter access to hearing aids — the type of electronics the SPU-001 is designed to light up. The August policy change allows consumers to buy hearing aids directly from stores without needing a medical exam, prescription or fit by an audiologist, creating a new market category.
Despite the go-to-market challenges ahead — making a new chip isn’t easy — Fock says FemtoSense is ahead of its revenue, with several enterprise and government sector customers lined up. (The company previously received a $2.1 million contract from the U.S. Office of Naval Research.) The 10-person company plans to hire two new employees this year and is already shaping the next generation of what Fock describes as a “general purpose.” ” hardware to accelerate applications including computer vision and natural language processing.
“Femtosense is well positioned to withstand the weather. [current economic headwinds] Raising money immediately before it, “Fok said.
