Bravo
If ARM was an arm, BRN would be its biceps💪!
Hi All,
After watching Tony Lewis' demonstration, I decided put together this simple table (with the help of ChatGPT) to help compare AKIDA with Snapdragon in relation to on-device GenAI.
Breaking it down like this makes me think that BrainChip would be a great fit for Qualcomm because the areas that AKIDA excels in (smart sensors or ultra-low-power devices) in are not the same areas that Snapdragon excels in (mobile phones or AI PCs doing heavier GenAI inference) – so no-one would be cutting anyone else’s lunch, so to speak.
If Qualcomm wants to tap into the battery-powered edge market, or to improve the battery life of mobile phones, for example, then they would be mad not to consider partnering with us, wouldn’t they?
What does everyone think?
Verdict Power Efficiency) : Akida is dramatically lower power, especially in always-on, sensor-fusion, or incremental learning scenarios.
Verdict (Performance & Throughput Activity) : For raw performance, Snapdragon wins - especially for LLMs and GenAI where traditional tensor ops and memory throughput dominate. Akida trades off some performance for extreme energy efficiency and real-time learning.
www.androidcentral.com
After watching Tony Lewis' demonstration, I decided put together this simple table (with the help of ChatGPT) to help compare AKIDA with Snapdragon in relation to on-device GenAI.
Breaking it down like this makes me think that BrainChip would be a great fit for Qualcomm because the areas that AKIDA excels in (smart sensors or ultra-low-power devices) in are not the same areas that Snapdragon excels in (mobile phones or AI PCs doing heavier GenAI inference) – so no-one would be cutting anyone else’s lunch, so to speak.
If Qualcomm wants to tap into the battery-powered edge market, or to improve the battery life of mobile phones, for example, then they would be mad not to consider partnering with us, wouldn’t they?
What does everyone think?
Feature | Akida | Snapdragon |
Power Draw | Micro-watt to milliwatt range | Typically several watts (2–10W) |
On-device Learning | Yes, continuous & incremental | No (inference only) |
Active Only When Needed | Yes (event-driven) | No (clocked cycles) |
LLM Scale (Billion+) | Demonstrated 1B+ parameter GenAI | Snapdragon 8 Gen 3 supports up to 10Billion parameters |
Verdict Power Efficiency) : Akida is dramatically lower power, especially in always-on, sensor-fusion, or incremental learning scenarios.
Verdict (Performance & Throughput Activity) : For raw performance, Snapdragon wins - especially for LLMs and GenAI where traditional tensor ops and memory throughput dominate. Akida trades off some performance for extreme energy efficiency and real-time learning.

Qualcomm is chasing Google when it comes to AI and your battery will be the victim
Faster charging is not the answer. Bigger batteries are.

Last edited: