Hardware & Compute
Chips, data centres and the infrastructure behind AI.
-
Trends in Sam Altman’s $500B Stargate Project and GPT-6 deployment
OpenAI’s GPT-6 deployment introduces tiered intelligence with GPT-6 Astra achieving a 32% higher average score than GPT-5.6 Sol. The rollout includes cost-efficient…
-
The risks in Cerebras’ WSE-3 rollout
Cerebras faces significant market volatility after OpenAI chose NVIDIA GPUs for its GPT-6.1 Sol Ultrafast mode, causing Cerebras stock to drop nearly…
-
Tenstorrent scales with RISC-V to challenge NVIDIA dominance
Tenstorrent secured a $1.5 billion Series E funding round to scale its Blackhole processor and RISC-V AI infrastructure. Jim Keller leads the…
-
The people shaping Groq’s LPU revolution and silicon strategy
Groq’s LPU architecture achieves 241 tokens per second on Llama 2 70b, outperforming conventional GPUs. Following a $20 billion deal with NVIDIA,…
-
Positron AI’s $5 billion valuation and the shift to LPDDR5X inference
Positron AI reached a $5 billion valuation following an $875 million funding round to develop its Asimov chip. The company’s Atlas system…
-
Lightmatter’s photonic interconnect and Passage silicon photonics chiplet
Lightmatter addresses AI data center bandwidth bottlenecks using the Passage platform, which provides 114 Tbps of bidirectional bandwidth via a 3D vertical…
-
Wasted Infinity Fabric bandwidth in Mixtral 8x22B clusters
Engineers often stall Mixtral 8x22B performance by ignoring ROCm 7 kernel optimizations and incorrect NUMA modes. Proper configuration of AITER kernels and…
-
CoreWeave closes $3.1 billion debt facility and $1.5 billion notes
CoreWeave secured a $3.1 billion delayed draw term loan and a $1.5 billion senior notes offering to fund AI infrastructure. The company…
-
The economics of Groq’s LPU and the NVIDIA inference price war
Groq’s LPU architecture delivers 750 tokens per second for Llama 3.1 8B, providing 21x more value per dollar than NVIDIA H100 hosting.…
-
The history of Graphcore and its IPU architecture
Nigel Toon and Simon Knowles founded Graphcore in 2016 to challenge NVIDIA with Intelligence Processing Units. Despite the Colossus MK2 GC200 chip…
-
Cerebras CS-3 vs SambaNova DataScale SN40L comparison
Cerebras generates 4,400 tokens per second on the GPT-OSS 120B model using a single CS-4 rack, significantly outperforming SambaNova’s 708 tokens per…
-
NVIDIA Blackwell manufacturing failures and CoWoS-L supply gaps
Thermal expansion mismatches and TSMC CoWoS-L capacity shortages are causing three-month delays for NVIDIA Blackwell shipments. Massive orders from Google for 400,000…