BRN Discussion Ongoing
- Thread starter TechGirl
- Start date
I pointed out in my earlier post that Onsor demonstrates the lengthy adoption style and this is outlined on the Onsor website.
No doubt its been frustrating for all holders but unfortunately there are no shortcuts.
Directors.
"In practical terms, acting in a responsible and ethical manner towards stakeholders (such as customers, employees, community and the environment) is necessary for the promotion of the interests of the company and its sustainability, and builds long-term value for shareholders.
Of course, this does not mean that all stakeholders will be satisfied with every decision taken, given stakeholder interests will be diverse."
My bold above.
The BOD have to prioritise long term value for holders over short term value - its the horse we are riding..
As the timeframes to revenue are lengthy (we are closing in on the future though) the BOD has had to remain firmly focused on the long term. This the reason for LDA cap raisings - the alternative is and was going broke.
Another aspect is that case law is very firm that Directors fiduciary duty is the entity and not individual holders.
The cornerstone case was Percival v Wright (1902) - UK but applies here as much as it did in 1902.
Briefly a group of individual holders of Nixon’s Navigation Co. approached the Directors to buy their shares. The directors after negotiations purchased the shares. Unbeknown to the holders during negotiations the Directors were also negotiating sell the entire company to a third party (Holden) which would have substantially increased the share value - the negotiations subsequently fell through though.
The shareholders later sued alleging nondisclosure of the negotiations.
The shareholder's lost their case on the basis that Directors duties are to the entity and not individual holders.
The case pretty much lets holders know where they stand - its basically 'like it or lump it' when it comes to the BOD setting a course for the business.
www.bing.com
No doubt its been frustrating for all holders but unfortunately there are no shortcuts.
Directors.
"In practical terms, acting in a responsible and ethical manner towards stakeholders (such as customers, employees, community and the environment) is necessary for the promotion of the interests of the company and its sustainability, and builds long-term value for shareholders.
Of course, this does not mean that all stakeholders will be satisfied with every decision taken, given stakeholder interests will be diverse."
My bold above.
The BOD have to prioritise long term value for holders over short term value - its the horse we are riding..
As the timeframes to revenue are lengthy (we are closing in on the future though) the BOD has had to remain firmly focused on the long term. This the reason for LDA cap raisings - the alternative is and was going broke.
Another aspect is that case law is very firm that Directors fiduciary duty is the entity and not individual holders.
The cornerstone case was Percival v Wright (1902) - UK but applies here as much as it did in 1902.
Briefly a group of individual holders of Nixon’s Navigation Co. approached the Directors to buy their shares. The directors after negotiations purchased the shares. Unbeknown to the holders during negotiations the Directors were also negotiating sell the entire company to a third party (Holden) which would have substantially increased the share value - the negotiations subsequently fell through though.
The shareholders later sued alleging nondisclosure of the negotiations.
The shareholder's lost their case on the basis that Directors duties are to the entity and not individual holders.
The case pretty much lets holders know where they stand - its basically 'like it or lump it' when it comes to the BOD setting a course for the business.
law case Pecival V Wright 1902 - Bing
Intelligent search from Bing makes it easier to quickly find what you’re looking for and rewards you.
#ai #edge #technology #marketresearch | Melvin Sörum | 10 comments
We just released a 90-page report covering the on-device AI market. 🧠 On-device AI is sometimes overshadowed by the hype surrounding cloud-based generative AI. However, on-device AI has quietly been transforming applications for over a decade, particularly those requiring real-time...
Pom down under
Top 20
Isn’t this on loihi2?As expected (and predicted), Markus Schäfer has posted one more time about neuromorphic computing in the context of the NAOMI4Radar project before stepping down as the Mercedes-Benz CTO at the end of this month:
If you’ve followed my updates, you’ll know I’ve shared insights into NAOMI4Radar before – and today, I’m excited to round off the story with our latest results. In the recently concluded public… | Markus Schäfer
If you’ve followed my updates, you’ll know I’ve shared insights into NAOMI4Radar before – and today, I’m excited to round off the story with our latest results. In the recently concluded public research project NAOMI4Radar, our colleagues in Future Software & AI at Mercedes-Benz AG joined...www.linkedin.com
Fiendish
Regular
Would it be possible for car manufacturers to use both akida and loihi in a single vehicle?
Yes, it's not only possible but increasingly likely for a car manufacturer like Mercedes-Benz to integrate both the BrainChip Akida and Intel Loihi 2 neuromorphic processors into a single production vehicle, particularly for specialized edge AI tasks in advanced driver-assistance systems (ADAS), autonomous driving, and in-cabin monitoring.
This approach aligns with the automotive industry's trend toward heterogeneous computing architectures, where multiple specialized chips handle distinct workloads to optimize for power efficiency, latency, and performance—much like how modern vehicles already combine NVIDIA GPUs for high-level perception, Qualcomm SoCs for infotainment, and dedicated MCUs for sensors.
Why This Makes Sense for Different Purposes
Mercedes has demonstrated active use of both chips in separate projects, with complementary strengths that lend themselves to modular deployment:
Akida for Ultra-Low-Power In-Cabin and Edge Inference: Deployed in the 2022 Vision EQXX concept vehicle, the Akida processor excels at always-on, battery-sensitive tasks like keyword spotting ("Hey Mercedes"), voice authentication, and contextual in-cabin monitoring. It achieves 5–10x energy efficiency over traditional AI by processing spiking neural networks (SNNs) at microwatt-to-milliwatt levels, ideal for non-safety-critical, continuous operations that don't drain EV range. BrainChip positions Akida for broader sensor-edge integration, such as with partners like Prophesee for event-based vision.
Loihi 2 for Radar and Real-Time Safety
Processing: Integrated into the ongoing Naomi4Radar project (a German-funded initiative announced in 2024), Loihi 2 focuses on event-driven radar data optimization for forward-facing sensors in ADAS. It enables up to 90% energy reduction and microsecond-latency responses for detecting objects in poor visibility, mimicking brain-like spiking for faster, more adaptive autonomous driving decisions. This is geared toward safety-critical, high-reliability tasks requiring on-chip learning.
These aren't mutually exclusive; Mercedes' research collaborations (e.g., with Intel and the University of Waterloo) emphasize neuromorphic computing's scalability across vehicle subsystems. In a full vehicle stack:
Akida could handle distributed, low-power nodes near microphones and interior cameras.
Loihi 2 could sit closer to radar/LiDAR fusion modules for external perception.
Evidence from Industry Trends
Heterogeneous Integration in AVs: Autonomous vehicles routinely use multi-chip setups today—e.g., NVIDIA's DRIVE platforms combine Orin SoCs with dedicated accelerators for redundancy and specialization. Neuromorphic chips fit this model, with projections for the neuromorphic AV market reaching $59 billion by 2034 (20.7% CAGR), driven by sensor fusion and edge processing. Research from Infosys and IBM highlights neuromorphic modules embedded directly in sensors (e.g., one chip per radar unit), enabling parallel use without central bottlenecks.
Mercedes' Multi-Partner Approach: As of late 2025, Mercedes continues exploring Akida for voice/sensor AI while advancing Loihi in radar R&D, with no reported conflicts. Broader partnerships (e.g., BrainChip with Renesas/Valeo for automotive IP, Intel's Hala Point scaling Loihi) suggest OEMs like Mercedes will mix vendors to avoid single-supplier risks.
Precedents and Future Outlook: While no public confirmation exists for both in one Mercedes vehicle yet (the EQXX remains a concept), 2025 reports from AI News Hub and Global Market Insights note accelerating commercialization, with OEMs like Renault/Bosch adopting multiple neuromorphic types for fleet adaptability. By 2026–2027, as Level 3+ autonomy rolls out in Mercedes' EQS/EQV lines, combined deployment could become standard for 90%+ efficiency gains in data processing.
Challenges like software standardization (e.g., MetaTF for Akida vs. Lava for Loihi) and ASIL safety certification remain, but Mercedes' CTO has publicly endorsed neuromorphic for "breaking new ground" in architectures. If energy and latency demands intensify with EV adoption, this dual-chip strategy could extend range by 10–20% while enhancing safety.
Yes, it's not only possible but increasingly likely for a car manufacturer like Mercedes-Benz to integrate both the BrainChip Akida and Intel Loihi 2 neuromorphic processors into a single production vehicle, particularly for specialized edge AI tasks in advanced driver-assistance systems (ADAS), autonomous driving, and in-cabin monitoring.
This approach aligns with the automotive industry's trend toward heterogeneous computing architectures, where multiple specialized chips handle distinct workloads to optimize for power efficiency, latency, and performance—much like how modern vehicles already combine NVIDIA GPUs for high-level perception, Qualcomm SoCs for infotainment, and dedicated MCUs for sensors.
Why This Makes Sense for Different Purposes
Mercedes has demonstrated active use of both chips in separate projects, with complementary strengths that lend themselves to modular deployment:
Akida for Ultra-Low-Power In-Cabin and Edge Inference: Deployed in the 2022 Vision EQXX concept vehicle, the Akida processor excels at always-on, battery-sensitive tasks like keyword spotting ("Hey Mercedes"), voice authentication, and contextual in-cabin monitoring. It achieves 5–10x energy efficiency over traditional AI by processing spiking neural networks (SNNs) at microwatt-to-milliwatt levels, ideal for non-safety-critical, continuous operations that don't drain EV range. BrainChip positions Akida for broader sensor-edge integration, such as with partners like Prophesee for event-based vision.
Loihi 2 for Radar and Real-Time Safety
Processing: Integrated into the ongoing Naomi4Radar project (a German-funded initiative announced in 2024), Loihi 2 focuses on event-driven radar data optimization for forward-facing sensors in ADAS. It enables up to 90% energy reduction and microsecond-latency responses for detecting objects in poor visibility, mimicking brain-like spiking for faster, more adaptive autonomous driving decisions. This is geared toward safety-critical, high-reliability tasks requiring on-chip learning.
These aren't mutually exclusive; Mercedes' research collaborations (e.g., with Intel and the University of Waterloo) emphasize neuromorphic computing's scalability across vehicle subsystems. In a full vehicle stack:
Akida could handle distributed, low-power nodes near microphones and interior cameras.
Loihi 2 could sit closer to radar/LiDAR fusion modules for external perception.
Evidence from Industry Trends
Heterogeneous Integration in AVs: Autonomous vehicles routinely use multi-chip setups today—e.g., NVIDIA's DRIVE platforms combine Orin SoCs with dedicated accelerators for redundancy and specialization. Neuromorphic chips fit this model, with projections for the neuromorphic AV market reaching $59 billion by 2034 (20.7% CAGR), driven by sensor fusion and edge processing. Research from Infosys and IBM highlights neuromorphic modules embedded directly in sensors (e.g., one chip per radar unit), enabling parallel use without central bottlenecks.
Mercedes' Multi-Partner Approach: As of late 2025, Mercedes continues exploring Akida for voice/sensor AI while advancing Loihi in radar R&D, with no reported conflicts. Broader partnerships (e.g., BrainChip with Renesas/Valeo for automotive IP, Intel's Hala Point scaling Loihi) suggest OEMs like Mercedes will mix vendors to avoid single-supplier risks.
Precedents and Future Outlook: While no public confirmation exists for both in one Mercedes vehicle yet (the EQXX remains a concept), 2025 reports from AI News Hub and Global Market Insights note accelerating commercialization, with OEMs like Renault/Bosch adopting multiple neuromorphic types for fleet adaptability. By 2026–2027, as Level 3+ autonomy rolls out in Mercedes' EQS/EQV lines, combined deployment could become standard for 90%+ efficiency gains in data processing.
Challenges like software standardization (e.g., MetaTF for Akida vs. Lava for Loihi) and ASIL safety certification remain, but Mercedes' CTO has publicly endorsed neuromorphic for "breaking new ground" in architectures. If energy and latency demands intensify with EV adoption, this dual-chip strategy could extend range by 10–20% while enhancing safety.
Similar threads
