tech news ᳇ eicker.news23h ago
The surge in #semiconductor stocks, driven by #AI, is shifting towards a more heterogeneous #computelandscape beyond #GPUs. While GPUs excel in training and answer inference, #Cerebras’ #wholewaferchips offer superior speed for #answerinference but face limitations in #training and #agenticinference due to cost and memory constraints. https://stratechery.com/2026/the-inference-shift/?eicker.news #tech #media #news
The Inference Shift

Agentic inference is going to be different than the inference we use today, and it will change compute infrastructure because speed won’t matter when humans aren’t involved.

Stratechery by Ben Thompson
Yvan ー イボん 🗺️  1d ago

amd-instinct-miI355x-platform-brochure.pdf
https://www.amd.com/content/dam/amd/en/documents/instinct-tech-docs/product-briefs/amd-instinct-miI355x-platform-brochure.pdf

You will not get cheap #GPUs once #AI craze is gone.

These are not runnable by you. (Well most of you)

Tom's Hardware: For The Hardcore PC Enthusiast [Unofficial]1d ago

AMD is reportedly developing an entry-level RDNA 4 GPU with 8GB of VRAM — RX 9050 rumored to debut with 2048 cores, more than RX 9060

https://fed.brid.gy/r/https://www.tomshardware.com/pc-components/gpus/amd-is-reportedly-developing-an-entry-level-rdna-4-gpu-with-8gb-of-vram-rx-9050-rumored-to-debut-with-2048-cores-more-than-rx-9060

NERDS.xyz – Real Tech News for Real Nerds [Unofficial]2d ago

NVIDIA CEO Jensen Huang tells graduates to embrace AI despite fears it could replace them

https://fed.brid.gy/r/https://nerds.xyz/2026/05/nvidia-jensen-huang-ai-graduates/

Winbuzzer1d ago

https://winbuzzer.com/2026/05/11/micron-memory-bottlenecks-threaten-ai-inference-efficiency-xcxwbn/

Micron's Jeremy Wernersays memory limits are becoming the constraint that can keep expensive data-center GPUs from running AI inference efficiently.

#AI #AIInference #Micron #AIInfrastructure #AICompute #AIChips #AIHardware #GPUs #HBMy#DataCenters #JeremyWerner

Winbuzzer2d ago

https://winbuzzer.com/2026/05/11/enterprises-face-underused-gpu-fleets-as-ai-costs-rise-xcxwbn/

Enterprise AI buyers are hitting a new cost wall as reported GPU utilization stays near 5% even while infrastructure spending keeps rising.

#AI #AIInfrastructure #GPUs #AIInference #AICompute #EnterpriseAI #DataCenters #AIInvestment #Nvidia

tech news ᳇ eicker.news2d ago
#Anthropic’s acquisition of all compute capacity at #xAI’s #Colossus1 data centre suggests xAI isn’t actively training its own #AImodels. This shift to a “neocloud” business model, #renting out #GPUs, raises questions about xAI’s innovative potential and its future profitability, especially as #SpaceX prepares for its IPO. https://techcrunch.com/2026/05/10/were-feeling-cynical-about-xais-big-deal-with-anthropic/?eicker.news #tech #media #news
We’re feeling cynical about xAI’s big deal with Anthropic | TechCrunch

On the latest episode of the Equity podcast, we discussed what xAI's deal with Anthropic might mean for parent company SpaceX.

TechCrunch
Tom's Hardware: For The Hardcore PC Enthusiast [Unofficial]3d ago

$200 'socketed' Nvidia AI GPU for servers hacked into a PCIe card with custom PCB and 3D-printed cooling — modded Tesla V100 SMX data center GPU runs AI LLMs and is more efficient than many modern midrange offerings in AI inference

https://fed.brid.gy/r/https://www.tomshardware.com/pc-components/gpus/usd200-nvidia-server-ai-gpu-hacked-into-a-pcie-card-with-custom-pcb-and-3d-printed-cooling-modded-tesla-v100-smx-gpu-turing-data-center-card-runs-ai-llms-and-is-more-efficient-than-many-modern-midrange-offerings-in-ai-inference

Tom's Hardware: For The Hardcore PC Enthusiast [Unofficial]3d ago

Testing Nvidia's RTX Mega Geometry tech — VRAM-reducing tech a leap forward for path-traced rendering

https://fed.brid.gy/r/https://www.tomshardware.com/pc-components/gpus/testing-nvidias-rtx-mega-geometry-tech-vram-reducing-tech-a-leap-forward-for-path-traced-rendering

Show threadOpen Risk4d ago

NB: Its not to imply that #gpus and accelerated computing (vectorising matrix-multiply) is incompatible with a human-centric digital landscape.

#Algorithms are powerful stuff, that is why they need to be in the service and control of society, or as the cliche goes: empowering all individuals, not just the #techbros

This constrains the manner in which algorithms are developed and deployed, and ultimately what kind of silicon we need.

3/