261 summarised stories in Hardware & Infrastructure, each linking back to the original source. Browse all topics →
OpenAI 8 years ago 22
Optimized GPU kernels for block-sparse neural networks have been released, enabling faster computation than standard libraries like cuBLAS and cuSPARSE. The kernels achieved state-of-the-art performance on text sentiment analysis and generative image and text modeling tasks. Networks using these kernels can execute orders of magnitude faster depending on their sparsity configuration.
The daily briefing
TLDRocket reads all relevant sources, removes duplicate coverage, and summarises the day in two minutes. Follow companies and topics for alerts, or get the briefing in Slack. Free, no spam, unsubscribe anytime.
TLDRocket
7 August 2026
Demis Hassabis's step back from running Google DeepMind signals a broader reorganization at Alphabet as the company recalibrates its AI leadership. Hassabis moves to chairman and chief scientist—a more research-focused role—while Koray Kavukcuoglu takes operational control, suggesting Alphabet wants nimbler execution alongside moonshot thinking. The shuffle arrives as smaller, more practical AI models prove their worth: Liquid AI's new 2.6-billion-parameter model runs entirely on-device with 128K context and tool-calling abilities, outperforming much larger competitors while keeping user data local. That shift toward edge deployment reflects where real money is moving. European VCs just backed it with conviction—White Star Capital closed a €250 million fund dedicated to industrial AI and resilience tech at Series A and B stages, betting on companies solving concrete problems rather than chasing frontier capabilities.
Read the full briefing →