Ambarella (NASDAQ:AMBA) CEO Fermi Wang said edge AI is developing quickly even as much of the broader AI investment and attention remains concentrated in data centers, arguing that the company is seeing expanding use cases across enterprise security, portable video devices, drones, and emerging “edge infrastructure” deployments.
Edge AI use cases expanding beyond endpoints
Wang described a shift from thinking only about AI at the endpoint (such as individual cameras) to building an “AI box” deployed at the edge that can aggregate multiple sensor inputs and run generative AI-type models. He gave an example of a CES demonstration with a software vendor that used security camera feeds in a retail store to turn cameras into operational tools, such as monitoring customer traffic and purchasing behavior.
Software portability and rapid partner adoption
Wang said Ambarella has been investing for years in a portable software platform intended to reduce customers’ effort when moving applications between silicon generations. He characterized this as a response to customer complaints about time-consuming software porting whenever chip platforms change.
As an example of the platform’s impact, Wang said a software vendor received Ambarella’s Cooper SDK roughly three months before CES, ported its software from a competitor’s platform to Ambarella’s in “less than a few weeks,” and was able to demonstrate at CES. Wang emphasized that while advanced silicon development takes time, software porting often takes longer and is outside Ambarella’s direct control, making portability a key time-to-market lever.
IoT demand: wearables, fleet management, and performance requirements
In discussing IoT opportunities, Wang pointed to wearable cameras as a market evolving beyond law enforcement into retail settings, where store clerks may use cameras to document interactions. He also highlighted fleet management as an area where AI-enabled cameras paired with telematics can extend beyond GPS tracking to support safety and operational insights, such as monitoring driving environments and conditions related to vehicles, drivers, and cargo.
Wang said Samsara is Ambarella’s largest customer in fleet management and that the company has seen “huge growth” in that segment, alongside broader adoption of edge AI in the category.
On product cycles, Wang said Ambarella experienced 37% growth in fiscal 2026 after beginning the year guiding for “high teens” growth, attributing the outperformance to customers ramping faster than expected. Looking to the current year, he said Ambarella is confident in its product ramps, including CV75, CV72, and CV7, but is watching customer ramp timing in the second half as a determinant of annual performance.
Wang also described increasing AI compute requirements in edge applications. He said early CV2-family use cases might have needed 2-3 TOPS, while current customers may require 20-50 TOPS due to application complexity and larger generative AI models. He said CV7, Ambarella’s first 4-nanometer chip, delivers 2.5x the AI performance of CV5, its first 5-nanometer chip, and that design activity around CV7 is “huge” across multiple applications.
Insta360 lawsuit: “absolutely no impact”
Addressing investor concerns raised after an earnings-related stock move tied to an Insta360 lawsuit, Wang said Ambarella did not initially have details during the conference but later spoke with the customer and referenced a public statement indicating “absolutely no impact” to the customer—and therefore no impact to Ambarella. He said the customer had worked around the patent issue for products involved in the case.
Automotive, robotics, and a semi-custom path to 2nm
On automotive, Wang said Ambarella continues to pursue autonomous OEM design wins and has gained greater visibility through increased invitations to bid on RFQs and RFIs, even after losing the Volkswagen decision. He said Ambarella disclosed a $13 billion total automotive opportunity identified over the next six years, noting the figure increased from the prior year and reflects additional projects the company is pursuing.
Wang also said investments made for autonomous driving are translating to robotics and drones, particularly where autonomous operation requires similar hardware and software capabilities. In discussing robotics, he outlined three product approaches:
- Single-camera solutions for basic object detection
- A “perception box” aggregating multiple cameras and other sensors for sensor fusion (the type he said was referenced in a recent earnings call win)
- A longer-term “domain controller” concept using a single chip to manage perception, planning, and control functions
Wang said Ambarella’s go-to-market approach in fragmented robotics markets will rely on partners—ISVs, system integrators, and OEMs—who can tailor models and software for specific robots.
He also discussed semi-custom ASIC opportunities tied to Ambarella’s move to 2-nanometer technology. Wang said a customer approached the company about sharing costs to add custom features without funding an entire 2nm tape-out, in exchange for allowing Ambarella to sell the chip to non-competing customers. He said the structure helped offset R&D costs for Ambarella’s first 2nm project and that additional customers have since expressed interest. Wang said the company is evaluating whether to engage with two to three customers to assess the opportunity and long-term viability.
On 2nm strategy, Wang said Ambarella intends to define 2nm products carefully, with identified customers and market potential, adding that the company does not plan to build another automotive chip at 2nm “anytime soon.” He framed 2nm as important for power efficiency in applications like robotics and HMI.
Wang said Ambarella has grown more confident in Samsung Foundry’s 2nm process over the past 12 months, expects to go to production in the first half of next year, and said Ambarella has secured Samsung capacity for 2026 and 2027. He contrasted this with what he characterized as tighter capacity availability at TSMC.
Looking ahead, Wang said he expects edge AI to become more visible over the next five years as more real-time AI applications move to the edge. He said Ambarella will continue investing heavily in edge AI, with video remaining a core business, while also expanding into edge AI aggregation “box” use cases.
About Ambarella (NASDAQ:AMBA)
Ambarella, Inc is a global semiconductor company headquartered in Santa Clara, California, specializing in video compression, image processing and computer vision technologies. The company designs low-power, high-definition system-on-chip (SoC) solutions that enable the capture, processing and streaming of video in a variety of embedded applications. Ambarella’s platforms combine advanced video encoding, multi-core central processing units and hardware accelerators to deliver high-resolution imaging with low power consumption.
Ambarella’s product portfolio caters to multiple markets, including security and surveillance, automotive vision, wearable cameras, drones and robotics.
