Upstage partners with AWS to expand artificial intelligence offerings
South Korean startup Upstage names AWS its preferred cloud partner to scale Artificial Intelligence foundation models, expand the Solar large language model, and pursue asia-pacific and U.S. markets.
Connecting to a local LLM with Ollama
Run a local language model with Ollama and connect Mundi for offline, free Artificial Intelligence-powered geospatial analysis on your laptop.
NVIDIA could target 2027 for 3 nm HBM base die production
Market reports say NVIDIA plans to design its own High Bandwidth Memory base die on a 3 nm node, a move that could reshape memory sourcing for Artificial Intelligence accelerators.
Lightmatter achieves 16-wavelength bidirectional link on single-mode fiber
Lightmatter has demonstrated a 16-wavelength bidirectional DWDM link on a single strand of standard single-mode fiber, delivering 800 Gbps bidirectional capacity to ease bandwidth constraints for Artificial Intelligence model scaling.
First Descendant developers criticized for artificial intelligence ads impersonating streamers
Nexon faces community backlash after Artificial Intelligence-generated TikTok ads used streamers´ likenesses without consent, prompting an apology and a joint probe with TikTok.
U.S. government to take a 10% equity stake in Intel
Bloomberg says the U.S. is discussing converting delayed CHIPS Act grants into a roughly 10% equity stake in Intel to support foundry expansion and research.
AI token factory is the new unit of computing
Data centers are being reshaped around the ´Artificial Intelligence token factory´, a systems-level approach that prioritizes maximum token throughput across clusters of GPUs and specialized hardware.
NVIDIA updates Project G-Assist with lighter model, halves VRAM requirement
NVIDIA cut Project G-Assist´s video memory needs in half by shipping a lighter Artificial Intelligence model and smarter tool-calling, enabling the assistant to run on many 8 GB and 6 GB GPUs.