AWS Graviton5 M9g Is GA: Benchmarks Confirm 36% Gains
AWS Graviton5 M9g EC2 instances are generally available. ClickHouse and Honeycomb independently measured 36% throughput gains. Here is the migration case and how ...
DeepSeek Builds Its Own Inference Chip: What Developers Need to Know
Reuters broke the story: DeepSeek is designing a custom inference chip via SMIC. Here is ...
SK Hynix’s $28B Nasdaq IPO: What Developers Must Know
SK Hynix controls 60% of the HBM market and just launched a $28B Nasdaq IPO. ...
AMD Ryzen AI Halo: $3,999, 128GB, and Reviews Are In
AMD Ryzen AI Halo launches at $3,999 with 128GB unified memory for 200B-param models locally. ...
Oxmiq OxCore: Run CUDA Code on Custom AI Silicon Without Changes
Raja Koduri raised $35M for Oxmiq's OxCore — a licensable RISC-V GPU architecture that lets ...
Qualcomm Buys Modular for $3.9B: What Developers Need to Know
Qualcomm's $3.9B acquisition of Modular gives it Mojo and MAX, the best CUDA alternative on the market. Here's what changes for developers and ...
Anthropic Custom AI Chip: Samsung 2nm Talks Start
Anthropic is the last frontier AI lab without custom silicon. Samsung 2nm chip talks and ...
OpenAI’s Jalapeño Chip: What Developers Need to Know
OpenAI and Broadcom unveiled the Jalapeño inference chip targeting 50% cheaper AI per-token costs. Here ...
OXMIQ Raises $35M to License GPU Architecture: What Devs Must Know
Raja Koduri’s OXMIQ raised $35M to license OxCore GPU architecture. OxPython runs CUDA code on ...
OxCore: Run CUDA Code on Non-NVIDIA Hardware Without Rewriting It
Oxmiq's OxPython runs existing CUDA and PyTorch code on non-NVIDIA hardware with zero changes. Here's ...