In a recent deep dive into Intel’s next-generation silicon, I sat down with Akshay Kamath, Director, Enterprise Client and Data Center Category, Intel India, to discuss the upcoming Panther Lake architecture. The conversation touched on the revolutionary 18A process node, the strategic positioning of the value-oriented Wildcat Lake, and Intel’s ambitious plans for the handheld gaming market.
The 18A Revolution: Performance-per-Watt
Panther Lake is the first client SoC built on the Intel 18A node. Beyond the raw performance numbers, can you quantify what the move to 18A means for the ‘performance-per-watt’ battle? Are we finally at a point where 18A-based designs can achieve fanless, all-day battery life that matches or exceeds the top-tier ARM-based competition without throttling?
The 18A node is revolutionary for two primary reasons: RibbonFET and PowerVia.
- RibbonFET: This is our gate-all-around technology. It allows architects to better control the current flow in the transistor, which directly leads to better performance-per-watt. At the silicon level, we are seeing a 15% improvement over the Intel 3 process.
- PowerVia: By moving the power plane below the transistor plane, we separate power lines from data lines. This improves signal integrity and chip density, up to a 30% improvement in density for Panther Lake.
(Note: The 30% density jump is the cumulative result of the 18A process; PowerVia specifically contributes roughly 10% of that through improved cell utilisation.)
Regarding fanless designs, while Panther Lake is delivering multi-day battery life (up to 27 hours in some tests), the implementation is up to our OEM partners. We also introduced Wildcat Lake (Core Series 3) with a lower base power of 15W (compared to Panther Lake’s 25W), which opens up even more possibilities for ultra-thin, highly efficient designs.
Bridging the Value Gap
I just saw the launch of Wildcat Lake (Core Series 3) for the value segment. It lacks DirectX 12 Ultimate support and has a 17 TOPS NPU, which is below the Copilot+ requirement. Is Intel intentionally creating a ‘two-tier’ AI ecosystem? How do you explain to a budget-conscious buyer in India why their laptop won’t support flagship AI features found in the ‘Ultra Series 3’?
It’s about balance. For the mainstream Indian market, typically laptops priced at ₹50,000 and below, the priority is productivity. These users are running browsers with 15 tabs, Office documents, and conference calls.
The Core Series 3 is architected to bring AI features to this price bracket for the first time. It uses the same Xe3 architecture as Panther Lake, but it’s right-sized for the segment. While it may not meet the specific TOPS requirement for Copilot+, it offers a massive leap in efficiency for background AI tasks like noise cancellation and blurring compared to traditional non-AI PCs.
XeSS 3 and 4x Frame Generation
Intel has been touting XeSS 3 with 4x frame generation for the Xe3 architecture in Panther Lake. Is this 4x boost achieved via a dedicated hardware block on the Xe3 tile, or is it a software evolution? More importantly, will this be backported to Lunar Lake users?
XeSS 3 uses AI to generate three frames for every single rendered frame. This process leverages the GPU’s AI capabilities to achieve that 4x boost. In our gamer software, users can toggle this Multi-Frame Generation (MFG) on or off.
Regarding the possibility of backporting this 4x frame generation to Lunar Lake laptops via a software or OTA update, I would have to get back to you on that, as I am not the technical expert on that specific detail.
The Handheld Exclusive Silicon for Gaming
Intel recently confirmed a ‘dedicated handheld gaming platform’ with Panther Lake. Can you confirm if Intel is developing unique silicon SKUs specifically for handhelds with a different core-to-GPU ratio, moving toward a ‘console-on-a-chip’ approach?
The excitement around handhelds is palpable. At CES, we confirmed that Panther Lake will power a dedicated handheld platform. While I can’t reveal specific SKU ratios yet, I can say that the Arc GPU in our Ultra series, especially with 12 Xe cores, is already delivering phenomenal integrated graphics performance. We are working closely with partners to ensure the silicon in these handhelds is optimised for that specific form factor. Stay tuned for more details.
The NPU vs. GPU Disparity
Panther Lake hits 120+ GPU TOPS but only 50 NPU TOPS. Over the next 24 months, does Intel see the NPU eventually taking over ‘heavy’ generative tasks, or will the Xe architecture remain the primary engine for high-end AI?
We believe in a multi-XPU strategy.
- NPU: Ideal for power-efficient, sustained background tasks like noise cancellation or background blur.
- GPU: The primary engine for heavy AI inferencing and large language models (LLMs) where sustained raw performance is required.
As NPUs move from 13 TOPS to 50 TOPS, they will handle richer models, but the GPU’s role in heavy AI workloads isn’t going away. We aren’t prioritising one engine at the expense of the other, we are giving developers the choice to run their features where they make the most sense.
Looking Towards The Future
As Intel pivots toward the 18A era, the conversation around Panther Lake reveals a company deeply focused on the intersection of power efficiency and specialised AI performance.
The dual-tier strategy with Wildcat Lake and the aggressive push into the handheld market suggest that Intel is no longer content with just being “inside” your PC, it wants to define the very architecture of our mobile and gaming futures.
For tech enthusiasts and everyday users alike, the promise of 27-hour battery life and console-grade handhelds marks the beginning of a fascinating new chapter in consumer electronics.
I’ll be keeping a close eye on how these innovations translate from technical specs to real-world performance as the first systems hit the market.
