The funding landscape for AI chip startups, once bright and optimistic, is now facing uncertainty as Nvidia continues to assert its market dominance.
Recent reports indicate that U.S. chip firms raised only $881 million between January and September 2023, a sharp decline from $1.79 billion during the same period in 2022. Notably, AI chip startup Mythic faced a financial crisis in 2022, nearly shutting down, while Graphcore, once a well-funded competitor, is now grappling with significant losses.
However, one startup is making waves in the competitive AI chip market. Hailo, co-founded in 2017 by Orr Danon and Avi Baum, who previously led wireless connectivity efforts at Texas Instruments, specializes in designing chips tailored for AI workloads on edge devices. Hailo’s innovative chips deliver AI processing with significantly lower memory and power consumption compared to traditional processors, making them ideal for compact, battery-powered devices such as cars, smart cameras, and robotics.
“I co-founded Hailo with the goal of democratizing high-performance AI beyond data centers,” Danon shared. “Our processors facilitate tasks such as object detection and semantic segmentation, and are increasingly utilized for AI-powered image and video enhancements. Recently, they’ve also been adapted for running large language models (LLMs) on edge devices like personal computers and infotainment systems.”
While many AI chip startups struggle to secure a single major contract, Hailo boasts over 300 clients across diverse sectors, including automotive, security, retail, industrial automation, medical devices, and defense.
To bet on Hailo’s potential, a group of investors, including Israeli entrepreneur Alfred Akirov, automotive importer Delek Motors, and venture capital platform OurCrowd, recently injected $120 million into Hailo, extending the company's Series C funding round. This capital will “empower Hailo to seize upcoming opportunities” and “lay the groundwork for sustained growth,” according to Danon.
“We are strategically positioned to bring AI to edge devices in transformative ways, enhancing the reach and impact of this groundbreaking technology,” Danon emphasized.
Could a startup like Hailo genuinely compete with industry leaders like Nvidia, as well as Arm, Intel, and AMD? Stanford electrical engineering and computer science professor Christos Kozyrakis believes the answer is yes—he argues that accelerator chips like Hailo’s will become increasingly essential as AI continues to expand.
“The energy efficiency disparity between CPUs and accelerators can’t be overlooked,” Kozyrakis explained. “Accelerators are vital for efficiency in key tasks like AI, while traditional processors can still support general programmability.”
Yet, Kozyrakis acknowledges potential longevity challenges for Hailo’s leadership—if the AI model architectures their chips are built to support become outdated. Additionally, he points out that software adoption could pose a challenge if developers don’t embrace the tools designed for Hailo's technology.
“Most challenges related to custom chips revolve around the software ecosystem,” Kozyrakis noted. “Nvidia, for example, holds a significant edge due to its long-term investment in software for its architectures.”
With $340 million in funding and a team of around 250, Danon remains optimistic about Hailo’s future—at least in the short term. He believes the startup's technology effectively addresses numerous issues encountered with cloud-based AI inference, particularly regarding latency, cost, and scalability.
“Traditional AI models reliant on cloud infrastructure often suffer from latency and other hurdles,” Danon remarked. “They struggle with real-time insights and alerts due to their need for network connectivity, leading to data privacy concerns. Hailo’s solutions operate independently of the cloud, allowing for significantly enhanced AI processing capabilities.”
Intrigued about Danon’s take on generative AI and its reliance on the cloud, I asked if Hailo perceives the current cloud-centric trend (like OpenAI’s approach) as a potential threat.
On the contrary, Danon believes the rise of generative AI is fueling increased demand for Hailo's hardware.
“In recent years, we’ve noticed a spike in demand for edge AI applications across a range of industries—from airport security to food packaging,” he explained. “The surge in generative AI is further amplifying this need, with customers in compute, automotive, industrial automation, security, and more requesting local processing for LLMs.”
Indeed, the future looks promising for Hailo.
Ranking
