Here's an interesting article I stumbled upon this morning. Sounds like AMD and BRN would make great partners since their embedded edge AI goals are very well aligned.
The article also mentions AMD's acquisition of Xilinx. This made me think about the project that integrated Xilinx and BrainChip technologies in 2017; with Brainchip accelerator implemented on Xilinx Kintex UltraScale FPGA.
For added context on the possibility of AMD being interested in neuromorphic or accelerated computing, it's also worth reading this previous post from June 2024.
April 22, 2025 11:19:29 AM
AMD’s Embedded Edge: Leadership, Differentiation, and AI Opportunity
- By Mark N. Vena
- April 21, 2025 6:00 AM PT
- Email Article
0 8 0 8
Image Credit: AMD
Over the last decade, AMD has undergone a remarkable transformation. From a company once struggling for relevance, it has risen under CEO Lisa Su’s steady and strategic leadership to become a serious contender across the data center, client computing, and now embedded and adaptive edge markets.
One of AMD’s fastest-rising stars is its embedded business, now armed with a broad portfolio and a sharp focus on AI. As competitors like Intel falter, AMD’s differentiated approach could put it on a trajectory to gain significant market share, particularly at the embedded edge.
AMD’s Embedded Rebound and Push Into Edge AI
AMD’s acquisition of Xilinx laid the foundation for its embedded success. This purchase brought a robust adaptive computing portfolio — FPGAs, SoCs, and RF technologies — that AMD has tightly integrated with its x86 CPUs, GPUs, and NPUs.
In an invite-only fireside chat with analysts last week, Salil Raje, SVP and GM of AMD’s Adaptive and Embedded Computing Group, the depth of that integration was on full display.
Raje outlined AMD’s five-pillar strategy:
- Fortifying its adaptive portfolio
- Improving usability for developers
- Growing x86 embedded market share
- Winning high-value custom silicon deals
- Leading in embedded AI
AMD isn’t just positioning itself as a component vendor — it’s becoming a “platform enabler” across industries like automotive, aerospace, communications, and robotics.
AMD’s Edge Over Intel in Embedded Strategy
It’s clear that AMD is not playing catch-up — it’s pushing ahead where others stall. AMD has achieved revenue leadership in adaptive computing, leaving Intel’s Altera (soon to be spun out again) in its wake.
In embedded CPUs, AMD holds just a 7%-8% market share but sees that as an opportunity, not a weakness. Raje said, “We believe we can grow significantly faster in that business over the next four to five years.”
What makes AMD’s approach stand out? Flexibility and openness. AMD’s edge strategy doesn’t depend on any single compute architecture. Instead, it uses a modular mix of x86, Arm, GPU, and FPGA — whatever the application demands.
The company also avoids a black-box approach to AI software stacks, instead partnering with ecosystem players to keep platforms open and customizable. This open strategy contrasts sharply with more closed approaches from some of its competitors, particularly in automotive and robotics.
AI at the Edge: AMD’s Next Big Move
Perhaps the most exciting element of AMD’s embedded strategy is its continued aggressive push into edge AI. “There will be a ChatGPT moment at the edge,” said Raje, and AMD wants to be ready for it.
AMD is integrating NPUs into nearly every product, from AI PCs to embedded SoCs. The goal is simple: deliver low-latency, power-efficient AI acceleration in markets like industrial automation, medical imaging, and autonomous vehicles.
AMD’s recent product launches reflect this vision.
From the versatile Versal AI Edge Gen 2 — which fuses ARM cores, FPGA fabric, ISPs, and NPUs — to the powerful EPYC Turing 9005 with 192 Zen 5 cores, the company is scaling across performance tiers and verticals. It’s already winning sockets in security, networking, and auto sectors.
Moreover, AMD’s AI software tools enable seamless migration from cloud-trained models to edge deployments, a unique value proposition that strengthens customer loyalty.
Custom Silicon: Playing Offense, Not Defense
AMD’s edge isn’t just in off-the-shelf products. Its custom silicon business, once limited to gaming consoles, is gaining traction across automotive, defense, and data center sectors. Importantly, AMD doesn’t pursue custom silicon unless it brings differentiated IP or platform value, such as integrating x86, GPU, or RF IP into unique packages. It’s a targeted, value-driven strategy that avoids commoditization.
Chiplets add another layer of flexibility. AMD’s leadership in chiplet architectures allows it to offer semi-custom solutions more cost-effectively, integrating customer IP on shared platforms. As chiplet adoption increases, AMD’s ability to modularize compute elements will become a powerful differentiator.
Leadership That Delivers
Much of AMD’s rise can be traced back to the discipline and strategic clarity of CEO Lisa Su. Her turnaround of the company isn’t a product of bold promises but of measured execution that prioritizes innovation, product roadmaps, and market focus. That same DNA is evident in AMD’s embedded and adaptive computing.
Su’s leadership enabled AMD to avoid the pitfalls that have plagued Intel — missed process nodes, delays in AI strategy, and an overdependence on legacy business lines. AMD, by contrast, now delivers products that are often competitive but ahead of the curve, particularly in power-performance ratios and time-to-market in AI.
The Intel Factor: A Window of Opportunity
Intel’s struggles in recent years have opened the door for AMD. From manufacturing delays to uncertainty around the Altera spinout, Intel’s position in the embedded market is vulnerable. While Intel still dominates in x86 embedded CPUs, its distracted execution allows AMD to grab market share, especially with AI reshaping workloads at the edge.
AMD’s strength in heterogenous computing, openness to Arm, and software-developer focus make it far more agile in responding to evolving edge AI needs than Intel’s more rigid portfolio. If AMD executes on its roadmap and continues to differentiate with a modular platform approach, it could become the go-to supplier for a broad swath of edge workloads.
A Platform for the Future
AMD’s embedded business is no longer just a side bet. It’s rapidly becoming a cornerstone of the company’s long-term growth strategy.
Once viewed as a niche, the embedded market is now a critical front in the broader computing landscape, especially as AI workloads migrate from centralized data centers to distributed, real-time environments at the edge.
Under Lisa Su’s direction, AMD’s leadership team has positioned the company to capitalize on this shift with discipline, clarity, and a sharp focus on execution.
The strategy is not just about having a broad portfolio but about how it works together. AMD is offering customers a consistent, scalable compute platform from cloud to edge, blending the flexibility of adaptive hardware with the performance of CPUs, GPUs, and NPUs.
This point is especially relevant in today’s fragmented edge environments, where power efficiency, latency, and customization define competitive advantage. AMD’s modular approach, enabled by chiplets and customizable silicon, ensures customers can get precisely what they need without compromise.
AMD’s Opportunity To Redefine Embedded Computing
The company’s solid stance on open software ecosystems resonates in markets that have grown weary of closed, proprietary solutions. This customer-centric approach and a differentiated product roadmap make AMD more than a component supplier: it’s becoming a strategic partner across industries.
As Intel contends with a
new CEO, internal restructuring, and struggles to execute cleanly in the embedded and AI domains, AMD has a rare and meaningful opportunity to gain market share and mindshare.
Already, the momentum is evident: new design wins, expanding market share in adaptive and embedded CPUs, and deepening traction in custom silicon engagements. The path ahead won’t be without challenges. Arm-based players, vertical integration trends, and software complexity will continue to test all major players, but AMD appears better prepared and positioned than ever.
I
t’s clear that AMD is not just catching up; it’s redefining the rules of the embedded game. If it continues to execute with the same precision that defines its turnaround story, AMD won’t just lead in AI at the edge but will help determine what the edge becomes.
What a remarkable juxtaposition AMD’s position in the semiconductor industry is today compared to just 10 years ago.
must have some kind of noise cancellation, if such a thing exists it would be included.
