With AI pushing everything in tech these days, that momentum is also boosting Edge AI optimism and economic activity despite some pronounced concerns by engineers over how to move early Edge AI pilot projects to full production.
The Edge AI momentum is being expressed by tech leaders who are explicitly optimistic, including Qualcomm CEO Cristiano Amon who has called the edge “the biggest opportunity of 2026” and added “the winner of edge is going to be the winner of the AI race.” Qualcomm’s CFO has seen Qualcomm’s advantage over Nvidia and more cloud-focused players because Qualcomm focuses on power-efficient, on-device AI across phones, PCs, cars and robotics.
And Arm is making a bold bet on AI chips that evolve into edge uses, while all of Nvidia, Broadcom, Intel and TSMC will also benefit after the cloud AI buildout of recent years, financial analysts have noted. Both Nvidia and Intel have put focus on the AI edge-dependent humanoid sector in the coming decade.
Some financial analysts view Qualcomm as a “moderate buy” for investors because they see it as an undervalued stock when compared to pure AI plays.
How optimistic are Edge AI developers?
But what about developers on the ground, who look at Edge AI as somewhat frustrating for the past eight years? The majority of Edge AI pilots have failed to move into production rollouts, according multiple surveys and industry accounts.
(Editor’s Note: Fierce is completing an online survey into this concern to be released soon. At Sensors Converge 2026, May 5-7 in Santa Clara, CA, the EDGE AI FOUNDATION is once again setting up a pavilion of several Edge AI companies.)
Fierce spoke Tuesday to a Lattice Semiconductor official who has worked with Edge AI developers to gain more perspective. He offered guarded optimism for the Edge AI sector after years of experiments and pilots.
“The last year with Edge AI shows an intensity of opportunities and how frequently [they occur] and what developers want to do. That is very different from previous years and causing me to be more optimistic,” said Hussein Osman, director of marketing at Lattice, in an interview.
Lattice specializes in field programmable gate arrays that work in Edge AI to sort through sensor data to help find the actionable data for a GPU or SoC to apply compute. Lattice FPGAs “make sensors into smart sensors,” according to company literature.
Osman added, “Versus the Edge AI market of seven to eight years ago, there’s a definite difference today. Customers have more concrete requirements for use cases -- and that’s not across the board -- but there are definite bright spots and areas where customers can monetize projects with the final customers.”
He attended an EDGE AI FOUNDATION conference in March where industry leaders compared notes about what they had accomplished with Edge AI. “Everybody was saying it is early days of Edge AI deployments. Everybody said there’s huge investment there.”
In recent years, Osman estimated that pilot Edge AI projects failed to make it to production as much as 80% to 90% of the time. Now, Lattice sees the markets for robotics and industrial uses for visual inspection and maintenance as prime growth areas. “We’re also seeing aerospace and defense with huge investments moving forward,” he added.
“If we really look at the market, security cameras have already taken off [with Edge AI] and the phone has Edge AI, and IoT and sensors are deployed everywhere. I’m at 60 percent to 70 percent, but not 100 percent certain, that everything is perfect and is going to work out.”
Still, he noted that Japan and China are “moving quite a bit faster” on Edge AI, while other countries are going to wait.
Lattice, based in Hillsboro, Oregon, was founded in 1983 and saw $509 million in revenue in 2024. In March, Lattice joined the Nvidia Halos ecosystem to build Halos-certified Holoscan Sensor Bridge-based designs for physical AI. Lattice will contribute expertise in low-power FPGAs to enable trusted, scalable AI systems across robotics, industrial automation and autonomous applications.
