I need to have a little lie down after reading about Taalas, the startup that claims to offer a 1000x speed up in LLM inferencing by building dedicated ASICs per model. Well done lads, you’ve invented what? A LUT? A Rom? They claim to be able to churn new models out in two months but their first prototype runs a two year old Llama model badly quantised down to 3 bits. I’m about 90% certain there will be a Theranos outcome for this company.

#ai #inference #taalas #theranos #asic

https://www.ctol.digital/news/taalas-hc1-review-17000-tokens-per-second-219m-raised-five-risks-every-investor-must-know/

And here we are, ladies and gentlemen. I told you. The same happened to #crypto. Someone went and made an #ASIC built specifically for #AI inference with #Llama models - and not a DRAM chip in sight. #GPU s are still be necessary for training, but this is the first crack in market saturation we've been seeing.

THERE'S LIGHT AT THE END OF THE TUNNEL, FOLKS!!!

#Taalas Launches Hardcore Chip With ‘Insane’ #AI #Inference Performance
https://www.forbes.com/sites/karlfreund/2026/02/19/taalas-launches-hardcore-chip-with-insane-ai-inference-performance/

Taalas just emerged from stealth with a claim that’s shaking the hardware world: 17,000 tokens per second on Llama 3.1 8B.

How? By physically etching the AI model directly into the silicon transistors. No HBM. No liquid cooling. Just raw, hardwired performance that is 10x faster and 20x cheaper than traditional GPU inference.

https://www.buysellram.com/blog/17000-tokens-second-is-taalas-hardwired-silicon-the-ultimate-solution-to-the-ai-memory-wall-and-hbm-shortage/

#AI #ArtificialIntelligence #AIHardware #DataCenter #MemoryWall #HBMShortage #InferenceFactory #HardcoreAI #ASIC #Taalas #NVIDIA #technology

Taalas just emerged from stealth with a claim that’s shaking the hardware world: 17,000 tokens per second on Llama 3.1 8B.

How? By physically etching the AI model directly into the silicon transistors. No HBM. No liquid cooling. Just raw, hardwired performance that is 10x faster and 20x cheaper than traditional GPU inference.

The Breakthrough: Taalas has unveiled the HC1 chip, achieving a massive 17,000 tokens/second on Llama 3.1 8B. It is roughly 10x faster and 20x cheaper than traditional GPU inference.

The “Hardwired” Secret: Unlike GPUs that load software, Taalas etches the AI model directly into the silicon transistors. By physically embedding the weights, they eliminate the need for High-Bandwidth Memory (HBM).

Solving the Memory Wall: By removing the “data movement” between external memory and the processor, Taalas bypasses the industry’s biggest bottleneck—the Memory Wall—and operates entirely on standard air cooling.

The Trade-off: The chip is model-specific. While it offers “insane” efficiency for stable, high-volume production (like 24/7 chatbots), it lacks the programmability and flexibility of a GPU.

Market Impact: The rise of these specialized “Inference Factories” actually increases the long-term value of your GPUs. Because GPUs are versatile and can be repurposed for any new model, they remain the “Gold Standard” for resale and training.

Demo LLM: chat jimmy

https://www.buysellram.com/blog/17000-tokens-second-is-taalas-hardwired-silicon-the-ultimate-solution-to-the-ai-memory-wall-and-hbm-shortage/

#AI #ArtificialIntelligence #AIHardware #DataCenter #MemoryWall #HBMShortage #InferenceFactory #HardcoreAI #ASIC #Taalas #NVIDIA #technology

Taalas just emerged from stealth with a claim that’s shaking the hardware world: 17,000 tokens per second on Llama 3.1 8B.

How? By physically etching the AI model directly into the silicon transistors. No HBM. No liquid cooling. Just raw, hardwired performance that is 10x faster and 20x cheaper than traditional GPU inference.

https://www.buysellram.com/blog/17000-tokens-second-is-taalas-hardwired-silicon-the-ultimate-solution-to-the-ai-memory-wall-and-hbm-shortage/

#AI #ArtificialIntelligence #AIHardware #DataCenter #MemoryWall #HBMShortage #InferenceFactory #HardcoreAI #ASIC #Taalas #NVIDIA #technology

Taalas just emerged from stealth with a claim that’s shaking the hardware world: 17,000 tokens per second on Llama 3.1 8B.

How? By physically etching the AI model directly into the silicon transistors. No HBM. No liquid cooling. Just raw, hardwired performance that is 10x faster and 20x cheaper than traditional GPU inference.

https://www.buysellram.com/blog/17000-tokens-second-is-taalas-hardwired-silicon-the-ultimate-solution-to-the-ai-memory-wall-and-hbm-shortage/

#AI #ArtificialIntelligence #AIHardware #DataCenter #MemoryWall #HBMShortage #InferenceFactory #HardcoreAI #ASIC #Taalas #NVIDIA #technology

How Taalas "prints" LLM onto a chip?

https://www.anuragk.com/blog/posts/Taalas.html

#HackerNews #Taalas #LLM #chip #technology #AI #innovation #chip #design

Applied for #TAALAS API access for a little thing I've been working on using cerebras. Hope they accept 🤞

#llama #AI

✨ Taalas HC1 raggiunge 17.000 token al secondo
Pesi nel silicio invece che nella RAM: la strategia di Taalas con HC1 per eliminare HBM, latenza e raffreddamento a liquido dall'inferenza AI.

https://gomoot.com/taalas-hc1-raggiunge-17-000-token-al-secondo/

#ai #hc1 #news #taalas #tech

Sopii kuvioon, että kun maassa on konservatiivisin #hallitus aikoihin, on valtiopäivien avajaisten jumalanpalveluksessa saarnaamassa #JariJolkkonen.

Tunnuksettoman tilaisuuden osallistujamäärä kasvaa koko ajan. Nyt puhujina olivat #CMI:n toimitusjohtaja #JanneTaalas ja #EvaBiaudet. Tilaisuuden teemana on #rauha ja #demokratia.

https://yle.fi/a/74-20141492

#eduskunta #uskonnonvapaus #uskonnottomuus #vakaumus #politiikka #yhteiskunta #uskonto #Jolkkonen #Biaudet #Taalas