BRN Discussion Ongoing

[TheDrooben]

#mit #machinelearning #neuromorphiccomputing #edgeai #decisionscience #brainchip #akida #parallaxadvancedresearch #aihardware #intelligentteaming | Dr. Steve Harbour, PhD

I’m honored to share that I’ve completed the MIT Professional Education program Machine Learning: From Data to Decisions – a 64-hour intensive focused on turning machine learning into rigorous, real-world decision frameworks. What made this program especially valuable is that it went far beyond...

www.linkedin.com

Happy as Larry

 

[DK6161]

Ironically the safest/quickest way to clear the Straight Of Hormuz for oil Tankers would likely be to deploy the Hive Mind that Kevin Johnson built in 5 days.
Kevin would drop 1000's of bouys with AKIDA sensors in the Straight for real time response to threats.
Real time response for sea and air drones plus mine detection.
The sensors 'act as one'.
Its pretty well a sure bet that Kevin's 'Hive Mind' will in one way or another be a part of the US Dome early warning system for land, sea, air and space. Then there is the security systems for 'secure' defense facilities and cybersecurity.
Bringing it back home Kevin's ' Hive Mind' would be great for Shark detection at our popular swimming beaches at low cost. They could also advise of a change in safety 'water' conditions potentially dangerous to swimmers and boating.
10000's deployed on our roads could warn us of traffic issues/incidents ahead on our chosen GPS route.
The Hive Mind potential is endless.

Great thinking, but can't help help to think that at the moment were relying on hobby projects to get more exposure on our technology.

This is cool and all, but we need companies to jump in, design their products and put it out the market.

Without that, Sean and co are reluctant to put out any announcements to the ASX. Cool project, but there is no money attached!

Out share price is stagnant and will continue to get eroded week by week. Help!

 

[Drewski]

Claude what Is Klepsydra Technologies?

Klepsydra is a Swiss company specializing in:
High-performance edge AI software
Heterogeneous computing optimization (mixing different processor types)
Real-time systems for aerospace, defense, robotics
Customers: ESA (European Space Agency), aerospace companies, defense contractors
This is NOT a random partnership - Klepsydra is serious enterprise software for mission-critical systems.

Happy Days

 

[poor guy]

Shouldn't Brainchip do an ASX announcement about this partnership (Klepsydra)?

 

[jrp173]

I cant believe a share hasnt had a run in 2 years

We haven't had a price sensitive deal announcement since December 2024 (I'm not talking about a company name slipped into a 4C, where the announcement is price sensitive because of the nature of the quarterly financial information, I'm talking about a stand alone announcement).

Yet again, BRN still doesn't announce partnerships (eg Klepsydra that's been found today). Shareholders (and the market) can only assume that these deals are not material to shareholders since they are not announced on the ASX. This is what Antonio said at the 2025 AGM - "we will absolutely, when something we deem is truly material to the shareholders, we will absolutely put it on ASX"

The market wants solid results, and IMO nothing short of a significant (hopefully IP) deal will move the share price.

 

[Bravo]

Klepsydra Technologies and BrainChip Announce Strategic Partnership for Heterogeneous AI Runtime for Akida™ Processors - The Desert Sun



Extract :- “



Klepsydra Technologies and BrainChip Announce Strategic Partnership for Heterogeneous AI Runtime for Akida™ Processors

Distributed by EIN Presswire

Brainchip Limited Holding Co (ASX:BRN)

“BrainChip’s Akida is the ideal neuromorphic partner in delivering the ‘performance layer’ that allows engineers to focus on building intelligent systems rather than fighting performance bottlenecks”.”

— Mike Carey, President of Klepsydra North America.

LAGUNA HILLS, CA, UNITED STATES, March 12, 2026 /EINPresswire.com/ — Klepsydra Technologies, a leader in high-performance edge computing software, and BrainChip Holdings Ltd (ASX: BRN, OTCQX: BRCHF, ADR: BCHPY), the world’s first commercial producer of ultra-low-power, event-based neuromorphic AI IP, today announced a strategic partnership to develop a revolutionary heterogeneous run-time environment for the BrainChip Akida™ processor family.

The collaboration integrates Klepsydra’s efficient software framework—capable of processing up to 10x more data with 50% less power consumption—directly with BrainChip’s Akida neuromorphic architecture. This partnership addresses a critical challenge in edge AI: the seamless orchestration of tasks between traditional CPUs and specialized neuromorphic accelerators.

Redefining Edge Performance through Heterogeneity
By utilizing Klepsydra’s proprietary approach, the new runtime will allow developers to offload compute-heavy AI layers to the Akida accelerator while maintaining high-speed synchronization with host processors like ARM, RISC-V, and x86. This “heterogeneous” approach ensures that mission-critical systems in space, defense, and automotive sectors can achieve breakthrough performance without the need for increased hardware complexity.

“The synergy between Klepsydra’s high-performance orchestration and Akida’s event-based processing creates a power-efficient powerhouse for the edge,” said Steven Brightfield, Chief Marketing Officer at BrainChip. “This partnership provides our customers with a unified, space-ready API that simplifies the deployment of complex AI models in a heterogeneous compute for the most demanding requirements”.

Great to see that BrainChip has partnered with Klepsydra Technologies to develop a heterogeneous AI runtime for the Akida neuromorphic processor.

Back in August 2024 I floated the possibility that BrainChip might be working with Klepsydra. At the time it seemed like the two companies technologies might fit really well togther, so it’s interesting to now see that collaboration formally confirmed.

As mentioned in the highlighted post below, Klepsydra has also worked with Frontgrade, European Space Agency and other aerospace programs, Their runtime software has been used for onboard satellite data processing, autonomous spacecraft operations and high-frequency sensor streams in space environments.

From what little I understand, one of the challenges with neuromorphic chips like Akida is that they don’t usually operate on their own. In most real-world systems you tend to have several different types of processors working together (CPUs, GPUs and other AI accelerators) each doing different parts of the job. The key phrase in the announcement seems to be “heterogeneous AI runtime.” That’s where Klepsydra appears to come in.

The other thing I also noticed is the “10× more data with up to 50% less power/CPU usage” claim mentioned in the announcement. From what I can see, Klepsydra has been citing similar performance improvements in their product materials for a couple of years, with references appearing in product pages and technical briefs around 2024.

Partnerships like this often take 12 - 24 months before a public announcement, so it wouldn’t surprise me if some level of collaboration had been happening behind the scenes for a while. Hard to know whether those performance figures are related specifically to BrainChip’s integration or simply Klepsydra’s underlying technology.

Klepsydra’s runtime appears to be designed for this kind of streaming AI architecture, which seems naturally compatible with Akida’s event-driven approach and is hopefully really well suited to applications such as robotics, autonomous drones, satellites, radar detection and industrial monitoring. I'm also hoping it links us to NASA's High Performance Space Computing chip at some point in the future.

 

[HopalongPetrovski]

Great thinking, but can't help help to think that at the moment were relying on hobby projects to get more exposure on our technology.

This is cool and all, but we need companies to jump in, design their products and put it out the market.

Without that, Sean and co are reluctant to put out any announcements to the ASX. Cool project, but there is no money attached!

Out share price is stagnant and will continue to get eroded week by week. Help!

We haven't had a price sensitive deal announcement since December 2024 (I'm not talking about a company name slipped into a 4C, where the announcement is price sensitive because of the nature of the quarterly financial information, I'm talking about a stand alone announcement).

Yet again, BRN still doesn't announce partnerships (eg Klepsydra that's been found today). Shareholders (and the market) can only assume that these deals are not material to shareholders since they are not announced on the ASX. This is what Antonio said at the 2025 AGM - "we will absolutely, when something we deem is truly material to the shareholders, we will absolutely put it on ASX"

The market wants solid results, and IMO nothing short of a significant (hopefully IP) deal will move the share price.

26 February 2026
LETTER TO SHAREHOLDERS
From the BrainChip Holdings Board of Directors

To maintain the highest standards of accuracy, we do not announce:

• Early-stage discussions, non-binding engagements or verbal commitments.
• Aspirational outcomes, conjecture, or unverified projections.
• Information protected by Non-Disclosure Agreements (NDAs).
  The Company has prepared this IR policy after obtaining appropriate external advice and having regard to the ASX Listing Rules.
  To provide shareholders with greater visibility between major milestones, we are introducing structured quarterly commercial progress summaries, in addition to our regular newsletters. These will provide clear operational insights without breaching commercial confidentiality.

When the going gets tough..............

 

[smoothsailing18]

Asking Ai about cars, brainchip and Valeo.

1. Mercedes-Benz: Mercedes-Benz has been a public pioneer in testing BrainChip’s technology. The VISION EQXX concept car utilized Akida for "keyword spotting" (voice control), achieving efficiency levels five times higher than conventional systems.[3] By 2026, this technology is expected to migrate from high-end concept vehicles into production models within the Mercedes-Benz Modular Architecture (MMA) and the MB.OS ecosystem.[4]
2. Valeo: As a major global Tier-1 supplier, Valeo partnered with BrainChip to integrate neuromorphic processing into their sensor suites.[5] Valeo provides components to a vast array of manufacturers, including BMW, Ford, and the Volkswagen Group. In 2026, cars from these brands utilizing Valeo’s "smart" LiDAR or ultrasonic sensors will effectively be powered by BrainChip IP at the edge.[6

 

[Fullmoonfever]

New Patent up

Full details:

Here

 

[buena suerte :-)]

Klepsydra Technologies – Put Intelligence into your system

klepsydra.com

Brainchip to be added to the list then with a few other names we know!!

 

[Rskiff]

Why again doesn't BRN announce on ASX new patent ffs. I asked google this question about Aus co's announcing patents.... answer, "Yes, Australian companies listed on the Australian Securities Exchange (ASX) frequently announce new patents and patent applications on the ASX platform" If anyone wanted to know more about the company by looking at ASX announcements, not much happening

 

[Diogenese]

Great to see that BrainChip has partnered with Klepsydra Technologies to develop a heterogeneous AI runtime for the Akida neuromorphic processor.

Back in August 2024 I floated the possibility that BrainChip might be working with Klepsydra. At the time it seemed like the two companies technologies might fit really well togther, so it’s interesting to now see that collaboration formally confirmed.

As mentioned in the highlighted post below, Klepsydra has also worked with Frontgrade, European Space Agency and other aerospace programs, Their runtime software has been used for onboard satellite data processing, autonomous spacecraft operations and high-frequency sensor streams in space environments.


View attachment 96118


From what little I understand, one of the challenges with neuromorphic chips like Akida is that they don’t usually operate on their own. In most real-world systems you tend to have several different types of processors working together (CPUs, GPUs and other AI accelerators) each doing different parts of the job. The key phrase in the announcement seems to be “heterogeneous AI runtime.” That’s where Klepsydra appears to come in.

The other thing I also noticed is the “10× more data with up to 50% less power/CPU usage” claim mentioned in the announcement. From what I can see, Klepsydra has been citing similar performance improvements in their product materials for a couple of years, with references appearing in product pages and technical briefs around 2024.


View attachment 96119


Partnerships like this often take 12 - 24 months before a public announcement, so it wouldn’t surprise me if some level of collaboration had been happening behind the scenes for a while. Hard to know whether those performance figures are related specifically to BrainChip’s integration or simply Klepsydra’s underlying technology.

Klepsydra’s runtime appears to be designed for this kind of streaming AI architecture, which seems naturally compatible with Akida’s event-driven approach and is hopefully really well suited to applications such as robotics, autonomous drones, satellites, radar detection and industrial monitoring. I'm also hoping it links us to NASA's High Performance Space Computing chip at some point in the future.

Hi Bravo,

I read the performance claim as being in relation to Klepsydra software and this is confirmed from their interweb:

https://klepsydra.com/#explained
Klepsydra’s acceleration technology combines two-dimensional parallelization with lock-free programming to deliver a flexible and efficient inference engine, achieving up to 4x lower latency, 10x higher throughput, and 50% reduced power consumption. Its autotuning tool optimizes AI inference for various constraints, while a unified API eliminates processor lock-in, enabling greater flexibility and improved performance across hardware platforms.

The fact that they are teaming up with BRN is a strong endorsement of Akida's performance, adding additional improvements above their claimed efficiencies, although it's not clear what the benchmark for their comparison is. In any case, a software solution will always be inherently inferior to the Akida hardware SoC.

In any event, Klepsydra seems a good vehicle to hitch our wagon to:

https://klepsydra.com/klepsydra-tec...-neaixt-project-to-advance-edge-ai-in-europe/

Klepsydra Technologies Awarded Horizon Europe NeAIxt Project to Advance Edge AI in Europe​

Zug, Switzerland — February 3, 2026 — We are proud to announce that we have been awarded a Horizon Europe project as part of the NeAIxt initiative, a strategic European effort aimed at strengthening Europe’s leadership in edge artificial intelligence. Europe is strongly positioned to play a pivotal role in the technological revolution, particularly in the field of edge AI, which promises sustainable growth, high performance, and reliability. NeAIxt is designed to foster European independence and control over edge AI technologies, benefiting both industry and society. The project brings together SMEs, research organizations, universities, and industrial partners to advance key AI enablers, evolve embedded non-volatile memory technologies for edge applications, and demonstrate AI capabilities at both chip and system levels. By advancing 18nm technology and next-generation embedded memory, NeAIxt will enable secure, low-power, AI-capable microcontrollers and deliver a fully European solution for safe, reliable, and sovereign edge AI applications across the entire value chain.

Within the NeAIxt consortium, Klepsydra Technologies will play a key role in enabling AI deployment at the edge. Klepsydra will support inference on a range of emerging microcontrollers and hardware accelerators by implementing optimized, in-device AI inference capabilities. This includes upgrades to its optimization toolchain to deliver efficient and optimized inference on constrained devices, as well as the integration of our optimization tools into AI/ML learning and development environments. In addition, Klepsydra will update its mathematical backends and frameworks to support the diverse processor and SoC architectures developed throughout the project, ensuring portability and performance. These capabilities will be validated and transferred across multiple use cases, contributing to the practical adoption of edge AI technologies.

Through NeAIxt, the consortium aims to reinforce Europe’s technological sovereignty in edge AI and embedded memory technologies, while ensuring safety, security, and compliance with EU regulations. We are happy to contribute to the advancement of European technology and, in particular, to the development of the JU chip, strengthening Europe’s position at the forefront of secure and sustainable edge AI innovation.

NeAIxt Project website: https://neaixt.eu/

This project has received funding from the European Union’s Horizon Europe research and innovation programme under the HORIZON-JU-Chips-2024-1-IA grant agreement No 101194172.The project is also supported by its members France, Italy, Czech Republic, Germany, Sweden, Danmark, Greece, Spain, Netherland, Portugal, Turkey, Switzerland. Views and opinions expressed are however those of the author(s) only and do not necessarily reflect those of the European Union or CHIPS JU. Neither the European Union nor the granting authority can be held responsible for them.

The chronological proximity is encouraging.

We have several edge applications in common:

https://klepsydra.com/#application

• Edge AI Applications: Ideal for automotive applications where numerous sensors generate massive data that must be processed on-board. It also supports IoT real-time applications, including cybersecurity, surveillance, and GPS-denied environments.

This is their base patent:

US12561183B2 Parallel data processing in embedded systems 20190913

The invention claims a computer-implemented method of lock-free parallel data processing in autonomous embedded systems comprising: one or more producer providing data, a Smart Object Pool, an asynchronous publisher object capable of cloning event objects, a circular batching queue, an event object handler, and a subscriber being arranged for sending the event objects to one or more consumers and for returning the event object to the smart object pool once it determines that no more consumer needs the event.

 

[Diogenese]

Great to see that BrainChip has partnered with Klepsydra Technologies to develop a heterogeneous AI runtime for the Akida neuromorphic processor.

Back in August 2024 I floated the possibility that BrainChip might be working with Klepsydra. At the time it seemed like the two companies technologies might fit really well togther, so it’s interesting to now see that collaboration formally confirmed.

As mentioned in the highlighted post below, Klepsydra has also worked with Frontgrade, European Space Agency and other aerospace programs, Their runtime software has been used for onboard satellite data processing, autonomous spacecraft operations and high-frequency sensor streams in space environments.


View attachment 96118


From what little I understand, one of the challenges with neuromorphic chips like Akida is that they don’t usually operate on their own. In most real-world systems you tend to have several different types of processors working together (CPUs, GPUs and other AI accelerators) each doing different parts of the job. The key phrase in the announcement seems to be “heterogeneous AI runtime.” That’s where Klepsydra appears to come in.

The other thing I also noticed is the “10× more data with up to 50% less power/CPU usage” claim mentioned in the announcement. From what I can see, Klepsydra has been citing similar performance improvements in their product materials for a couple of years, with references appearing in product pages and technical briefs around 2024.


View attachment 96119


Partnerships like this often take 12 - 24 months before a public announcement, so it wouldn’t surprise me if some level of collaboration had been happening behind the scenes for a while. Hard to know whether those performance figures are related specifically to BrainChip’s integration or simply Klepsydra’s underlying technology.

Klepsydra’s runtime appears to be designed for this kind of streaming AI architecture, which seems naturally compatible with Akida’s event-driven approach and is hopefully really well suited to applications such as robotics, autonomous drones, satellites, radar detection and industrial monitoring. I'm also hoping it links us to NASA's High Performance Space Computing chip at some point in the future.

Hi Bravo,

As you mention, NNs don't usually operate without a coprocessor. Akida 1 needs a coprocessor only for setup/configuration (wieghts/layers & NPUs) and for any downstream processing of the inference, but does the inference independently. I think the coprocessor has some smallish involvement in TENNs inference, so combination with software or processor architecture, eg, RISC-V becomes a more important consideration for Akida 2 +.

 

[Diogenese]

BrainChip exits 300:


https://www.bing.com/images/search?...&itb=0&idpp=overlayview&ajaxhist=0&ajaxserp=0

 

[Rach2512]

BrainChip exits 300:


https://www.bing.com/images/search?view=detailV2&ccid=UIUohpNB&id=D763E80C6639225BCF0038D4227E00D63D14F007&thid=OIP.UIUohpNBsFEWbNlhCddglwHaEg&mediaurl=https://www.emugifs.net/wp-content/uploads/2023/02/Ill-Be-Back-GIF-and-MEME-Police-Station-Assault-Scene-The-Terminator-Film-Movie-With-Arnold-Schwarzenegger-Download-and-Share-on-WhatsApp-Group-Messenger-or-Facebook-Post-Like-Reaction-GIF.gif&exph=390&expw=640&q=i'll+be+back+gif&FORM=IRPRST&ck=9222D787DAC1B7042D5CC952E6BB7F89&selectedIndex=0&itb=0&idpp=overlayview&ajaxhist=0&ajaxserp=0

What does this mean for the shorters?

 

[JoMo68]

I agree JoMo,

I think the situation between the US govt and Anthropic is something BrainChip shareholders should be aware of.

From what I understand, Anthropic has reportedly fallen out of favour with the US Department of War after refusing to allow its technology to be utilised (without additional permissions) for certain applications, particularly large-scale facial recognition, surveillance and autonomous weapons. Anthropic argued that the proposed use extended beyond the original scope of the agreement and raised ethical and safety concerns, especially around certain autonomous and lethal military applications.

The US government’s position seems to be that decisions about national security use cases should ultimately rest with the state, whereas Anthropic argues that companies should retain some control over how their technologies are deployed.

The Pentagon has punished Anthopic for it's views on AI safety in warfare by blacklisting it and labelling it a "supply chain risk", effectively restricting other defense contractors from using it's Claude AI models and Anthropic, as a result, has sued the US government, calling the ban unlawful retaliation and arguing it threatens billions of dollars in revenue

What makes this situation quite unusual is that the US govt has taken action against foreign technology providers in the past over national security concerns, but it’s extremely rare to see such a public dispute with a domestic AI company.

Advanced AI technologies are becoming strategically important in areas such as electronic warfare, surveillance, and defense systems and that creates a grey area where governments may feel compelled to assert stronger control over technologies they consider critical.

This is potentially relevant to BrainChip IMO because its neuromorphic processors are being explored in some similarly sensitive defense and intelligence-related applications.

You captured my concerns perfectly Bravo.

 

[Jacob Kiraz]

We haven't had a price sensitive deal announcement since December 2024 (I'm not talking about a company name slipped into a 4C, where the announcement is price sensitive because of the nature of the quarterly financial information, I'm talking about a stand alone announcement).

Yet again, BRN still doesn't announce partnerships (eg Klepsydra that's been found today). Shareholders (and the market) can only assume that these deals are not material to shareholders since they are not announced on the ASX. This is what Antonio said at the 2025 AGM - "we will absolutely, when something we deem is truly material to the shareholders, we will absolutely put it on ASX"

The market wants solid results, and IMO nothing short of a significant (hopefully IP) deal will move the share price.

Your a sitting duck when you dont ,
Lets hope this is the bpttom,
Come to the AGM i feel old mate who rocks up every year will just unload on them, unfortunately for them ha hs if no news or movement