Tenstorrent Demo Shows Ultra‑fast Video Generation on 256 BlackHole Chips
.@tenstorrent will launch its new gen cluster-scale systems next week, but in the mean time, Jim Keller @jimkxa and Jasmina Vasiljevic @JasminaVas gave me a sneak preview of their video generation demo running on 256 BlackHole chips (spoiler - it's fast): https://t.co/7oVYauf0VE
Edgecortix AI Chip Evolves From Compiler to NASA‑Qualified Hardware
Had a great conversation with @edgecortix CEO about the interesting details of the company's reconfigurable AI chip. It started out as a compiler licensed to Renesas, but today it's entering its third generation of hardware and space-qualified by Nasa. https://t.co/HbuOiXbTu5
Nvidia's Groq Acquisition Expands AI Hardware Portfolio
Here's a long (but hopefully worth it) story about how the @nvidia-Groq deal came about, the new Groq chip and how it fits into Nvidia's portfolio alongside Vera Rubin, how much of Groq's software stack will be used and what's...
Nvidia and Arm Disrupt Legacy Models in One Week
Last week Nvidia admits its GPUs need help and unveils, with fanfare, a chip from a 10yo startup competitor many had written off. This week Arm breaks its 35yo business model and becomes a merchant silicon provider. What a time...
TinyEngine NPU Expands to Third MCU Family for Sensor Edge
.@TXInstruments has added its home-grown NPU, TinyEngine, to a third MCU family that will go into sensor edge applications. It's not a scaled-down version of the C7, the accelerator in some of TI's bigger processors, which has a DSP component: https://t.co/fZU4XOsdBF
Multi-Physics Design Must Shift Upstream in Workflow
Had fun moderating a very interesting multi-die/multi-physics panel at @Synopsys Converge yesterday with four esteemed expert speakers: Shankar Krishnamoorthy, Asad Khamisy, Raja Swaminathan and Aveek Sarkar. All agreed multi-physics design needs to move upstream in the workflow. https://t.co/Ee6bVVmGLf
Nvidia's Near Groq Deal Validates AI Chip Startups
When Nvidia all-but acquired Groq for a reported $20B, it inadvertently validated the entire AI chip startup landscape (featuring commentary from @vsora_dsp @SambaNovaAI @dMatrix_AI) https://t.co/65Jk6Fe7I5
Ambarella Onboards ISV Partners in Weeks
.@Ambarella_Inc: "We moved ISV's software over in a matter of weeks" The company onboarded new ISV partners extremely quickly, as it evolves its go-to-market strategy to an ecosystem approach. Catch the full episode on the EE Times channel at https://t.co/NbBpWDeyi1...
AxeleraAI Secures Record €250M EU Chip Funding
. @AxeleraAI has raised the biggest semi round in EU history at $250 million. CEO Fabrizio Del Maffeo told me investor confidence in European chip companies is growing, and talked about commercial expansion, a next-gen chip and the startup's 500...
Key Criteria for Choosing the Right ISVs
@Ambarella_Inc's Muneyb Minhazuddin on how to select ISVs. View the full episode of AI With Sally on the EE Times channel at: https://t.co/NbBpWDeyi1 https://t.co/QSE5d8xBfe
SambaNova Skips Intel Acquisition, Secures Funding and Partnership
.@SambaNovaAI cancelled plans to be acquired by Intel, instead raising an oversubscribed Series E. SN's CEO and chief architect told me about a new multi-year tech and go-to-market partnership with Intel, and SN's new silicon for multi-agent workflows: https://t.co/JLMl9PIUGk
Taalas Chip Excels at Llama3.1, but Lacks Flexibility
AI chip startup Taalas @taalas_inc is showing off a chip that can do 16,000 tps/user on Llama3.1-8B, many multiples of its nearest competitor. The catch? The chip ONLY runs Llama3.1-8B, and a model like DeepSeekR1-671B would need 30 separate tapeouts: https://t.co/IJuprQZqaE
HyperAccel Unveils Bertha 500, Low‑latency AI Chip
South Korean AI chip startup HyperAccel @HyperAccel0112 is preparing to launch Bertha 500, an LLM inference accelerator aimed at cost-efficient, low-latency token generation. The company also plans an edge SoC version in partnership with Korean giant LG: https://t.co/TRgQrMNik2
Ex-Altera CEO Leads Vsora to Challenge Nvidia, AMD
Former Altera CEO Sandra Rivera has joined French AI chip startup @vsora_dsp as chair of the board. I had a chat with her about raising Vsora's profile, raising funding, and competing with Nvidia, AMD and the entire landscape of startups...