On August 13, 2026, Signal65 published a report, commissioned by Qualcomm, comparing three retail premium thin-and-light 16-inch Windows laptops. The ASUS Zenbook A16, equipped with the Snapdragon X2 Elite Extreme, significantly outperformed the Intel and AMD machines on key CPU and AI benchmarks, with a gap of up to 87% in Cinebench 2026 multi-thread scores. Previous Snapdragon family evaluations used reference designs, but this time the products compared were purchased at market prices in July 2026. That said, this is the result of comparing three finished products, not a test that establishes a definitive ranking of the chips themselves.
Up to 87% Gap Among Three Retail 16-Inch Machines
The machines compared were the Zenbook A16 with the Snapdragon X2 Elite Extreme X2E-96-100, the MSI Prestige 16 Flip AI+ with the Core Ultra X9 388H, and the HP OmniBook X Flip 16 with the Ryzen AI 9 465. According to Signal65, all three machines ran Windows 11 Home with virtualization-based security (VBS) enabled, and each test used the average of multiple runs. Comparisons were made using "Best Performance" and "Balanced" modes on AC power, and Balanced mode on battery.
In Geekbench 7's Best Performance mode, multi-core scores were 27,519 for the Snapdragon machine, 17,990 for the Intel machine, and 15,023 for the AMD machine—the Snapdragon machine outperformed the Intel machine by 53% and the AMD machine by 83%. Cinebench 2026 multi-thread scores were 6,998, 3,732, and 3,785 respectively, with gaps of 87% over the Intel machine and 85% over the AMD machine. Signal65 discloses that it provides research, advisory, and lab services to tech companies including Qualcomm, so these figures should be read as measurements from a commissioned party.
On the other hand, the conditions across the three machines were not equal. The Zenbook A16 has 48GB of memory and a 66.5Wh battery, the MSI machine has 32GB of memory and an 82.0Wh battery, and the HP machine has 32GB of memory and a 67.3Wh battery. Screen resolutions also differ: the Zenbook A16 and MSI machine are 2880×1800, while the HP machine is 1920×1200, and storage capacities also vary. There is no guarantee that the same ranking or gap would reproduce if the same CPU were placed in a different OEM design.
Where Did the 18 Cores and 80 TOPS Make a Difference?
According to Qualcomm's product overview, the X2E-96-100 features an 18-core CPU with a maximum clock of 5.0GHz, 53MB of cache, an 80 TOPS Hexagon NPU, and 228GB/s of LPDDR5X memory bandwidth. In comparison, the Core Ultra X9 388H has 16 cores and a 50 TOPS NPU, while the Ryzen AI 9 465 has 10 cores/20 threads and a 50 TOPS NPU. These differences in core count, NPU performance, and memory bandwidth form the backdrop for the widening gap in CPU multi-thread and AI performance. However, this benchmark alone cannot isolate how much each specification contributed to the scores.
Single-core performance also showed differences. In Geekbench 7's Best Performance mode, the Snapdragon machine scored 3,438, outperforming the Intel machine's 2,639 by 30% and the AMD machine's 2,614 by 32%. Cinebench 2026 single-thread scores were 622, 502, and 460, with gaps of 24% and 35%. High multi-thread CPU scores cannot be generalized to mean that all typical tasks will speed up by the same proportion.
For AI, Procyon AI Computer Vision's Best Performance scores were 4,414 for the Snapdragon machine, 2,265 for the Intel machine, and 1,809 for the AMD machine, which Signal65 described as 1.9x over Intel and 2.4x over AMD. In Computer Vision 2.0 WinML, scores were 2,170, 1,601, and 1,143 respectively, with the Snapdragon machine leading by 36% and 90%. This image-recognition-focused measurement does not verify compatibility or real-world application responsiveness.
CPU and AI Rankings With the Charger Unplugged
Even when the Snapdragon machine was set to battery power/Balanced mode and the Intel machine to AC power/Best Performance mode, the Snapdragon machine still led in Signal65's CPU and AI tests. The gaps were 43% for Geekbench 7 multi-core, 29% for single-core, 26% for Cinebench 2026 multi-thread, 20% for single-thread, and 79% for AI Computer Vision. However, under these conditions, the Intel machine led in Office, and browser performance was effectively equal. There is no guarantee that categories beyond CPU and AI will show the same ranking.
Looking at how well Balanced-mode AC performance was retained on battery, averaged across six tests, the Snapdragon machine retained 92%, the Intel machine 94%, and the AMD machine 81%. The Snapdragon machine did not maintain the same level consistently across all tests—its Cinebench multi-thread retention rate dropped to as low as 67%. The strength shown when unplugged is a result that should be viewed separately for averages versus individual workloads.
Intel Holds the Edge in GPU and Office Battery Life
On 3DMark Steel Nomad Light, which measures integrated GPU performance, the Core Ultra X9 388H machine outperformed the Snapdragon machine. Best Performance DX12 scores were 5,848 for the Intel machine, 5,419 for the Snapdragon machine, and 3,373 for the AMD machine—the Intel machine scored 8% higher than the Snapdragon machine. Under Vulkan, the scores were 6,203, 5,555, and 3,087, a 12% gap. The CPU and AI advantage cannot be extended to gaming or GPU performance.
Office results also varied by use case. Procyon Office's Best Performance overall scores were 8,737 for the Snapdragon machine, 7,518 for the Intel machine, and 7,296 for the AMD machine, but under Balanced mode they converged to 7,709, 7,316, and 7,065, narrowing the advantage to 5% and 9%. In Outlook, the Intel machine, and in Word, the AMD machine, were effectively on par with the Snapdragon machine. In light productivity tasks, the maximum gaps seen in CPU and AI do not necessarily carry over.
Battery life measured under Best Power Efficiency showed 14.2 hours for the Snapdragon machine, 12.2 hours for the Intel machine, and 6.9 hours for the AMD machine during local video playback. For Chrome browsing, the figures were 10.3, 9.0, and 10.4 hours, with the AMD machine effectively on par; for Procyon Office, they were 11.4, 13.4, and 6.2 hours, with the Intel machine lasting two hours longer than the Snapdragon machine. Per watt-hour of battery capacity, the Snapdragon machine had the longest runtime in both video and Office tests. However, capacity-adjusted efficiency is a separate metric from how many hours a device can actually be used in practice.
Looking Beyond Price-to-Performance to Configuration Differences
The July 2026 purchase prices were $1,699 for the Snapdragon machine, $2,299 for the Intel machine, and $1,649 for the AMD machine. Signal65 calculated price-to-performance ratios using three metrics—Geekbench 7 multi-core, Cinebench 2026 multi-thread, and AI Computer Vision—and with the Snapdragon machine set to 1.00, the Intel machine scored 0.48, 0.39, and 0.38, while the AMD machine scored 0.56, 0.56, and 0.42. Averaged across the three metrics, the Snapdragon machine comes out to 2.4x the Intel machine and 1.9x the AMD machine, but this is not an average across all tests or all types of work.
The price difference also reflects configuration differences beyond the CPU. The Snapdragon machine has 48GB of memory and 1TB of storage, the Intel machine has 32GB and 1TB, and the AMD machine has 32GB and 512GB, with battery capacity and screen resolution also differing. Subsequent discounts or regional price variations are not reflected either. Therefore, the price-to-performance ratio should be treated as an index dividing three specific measurements—taken under matched conditions—by the July 2026 purchase prices.
An independent Notebookcheck review of the Zenbook A16 also found that the lower-tier X2E-94-100 outperformed the Core Ultra 7 356H and Ryzen AI 9 465 machines in Cinebench 2024 multi-core, though the gap narrowed to 3% and 8% in Cinebench R23. While this does not corroborate Signal65's direct comparison of identical machines, it offers a point of reference showing how gaps can shift depending on the test generation and implementation. Compatibility, real-world application performance in games, driver maturity, external GPUs, and Linux support are outside the scope of the Signal65 report. When choosing a product, it's only by checking GPU performance, battery life, and compatibility with the software you actually use—in addition to the CPU and AI figures—that one can properly judge the meaning of this comparison.