AMDGPU retweeted
SHOCKING: We have confirmed that multiple Neoclouds have been refused access to GTC Berlin due to them adding AMD GPUs, TPUs, and other chips to their cloud offerings
Note that this doesn't apply to hyperscalers, as they have more buying power. But for neoclouds, executives feel NVIDIA has been using high-pressure tactics to keep them NVIDIA-only.
As such, neocloud executives are starting to consider offering TPUs or AMD GPUs quietly to avoid getting pressured by NVIDIA and to avoid the feeling that NVIDIA will retaliate.
xitter.cf/SemiAnalysis_/status/2…
SHOCKING: Many neocloud executives we spoke with feel that if they have non-NVIDIA networking gear on their cluster, or if their cloud has an AMD GPU or TPU offering, NVIDIA will retaliate. They feel that retaliation includes not giving early allocation or no longer supporting a potential IPO/VC raise.(1/3)🧵
AMDGPU retweeted
ALERT🚨: AMD JUST LANDED SUPPORT FOR MEGAKERNEL DECODE IN UPSTREAM VLLM WELL BEFORE NVIDIA. This marks one of the first times that a major vLLM feature is only supported on ROCm and not on CUDA yet.
MegaKernels is a technique that is massively important for high-interactivity decoding by reducing the # of kernel launches by 10x.
This marks an❗️important milestone❗️for ROCm vLLM to land major ROCm-exclusive features. CUDA vLLM doesn't even have draft PRs for upstream vLLM megakernel support yet.
AMDGPU retweeted
DLSS IS NOW RUNNING ON AMD HARDWARE 🚨
A new mod called HelixSR runs NVIDIA's DLSS 3.7 neural network directly on AMD GPUs. No NVIDIA card, no drivers, no cloud.
It even works on the Steam Deck, where early tests have it running FASTER than AMD's own FSR 4 (3.5ms vs 4.5ms per frame at 800p) while looking way better than FSR 3.
Video below is DLSS vs FSR 4 on an RX 5700 XT, a 7 year old AMD card.
NVIDIA IS F*CKED
AMDGPU retweeted
AMD Software: Adrenalin 26.9.2 is now available (Optional Update)! New games are compatible, and issues have been fixed!
More can be found in the full release notes here: bit.ly/4hC6mCE
AMDGPU retweeted
Does anyone know if @huggingface (owned by @NVIDIA) still has access to the MI455X that AMD gave them under embargo? Does anyone know if AMD continues to give exclusive access to next-gen hardware to Hugging Face (owned by NVIDIA)? Will Hugging Face be working on making Nemotron3 Ultra work on MI455X?
AMDGPU retweeted
AMD ALERT🚨🚨: On Kimi K3 2.8T, AMD has even better profit margins per gigawatt compared to B300, GB200 NVL72, and B200. This is due to AMD's lower TCO and better performance, leading to better profit margins of 53.6% on Mi355X compared to 44.3% on GB200 NVL72. This is from our open-source Agentic Inference benchmark, AgentX.
Shoutout to @roaner & @EmadBarsoumPi 10x engineers for optimizing this 2.8T model!
AMDGPU retweeted
Anthropic, OpenAI y xAI pueden ralentizar sus modelos cuando quieran. Incluso apagarlos hoy mismo.
El problema es vender esa pausa bajo el disfraz del "bien común" y la "seguridad".
Lo que realmente estamos viendo es captura regulatoria clásica:
- La frontera tecnológica se democratiza rápido (open-source pisándoles los talones).
- Mantener la ventaja competitiva se vuelve exponencialmente caro.
- La solución más rentable: usar el miedo al "apocalipsis de la IA" para que el Estado imponga barreras de entrada imposibles de pagar para startups y proyectos open-source.
Quieren monopolio por decreto.
Los mismos "EmpreSaurios" prebendarios de siempre, pero con sede en Silicon Valley y disfrazados de salvadores de la humanidad.
AMDGPU retweeted
"The CUDA moat is gone" says @DavidBennett__ , the CEO of a new Japanese neocloud called ai& (aka @aiand_). A chunk of their infrastructure is built on @tenstorrent hardware, one of the largest deployments outside of the USA.
morethanmoore.substack.com/p…
AMDGPU retweeted
"Ocarina of Time parece que lleva DLSS5, se ve muy ia" 😪
Como se vería el putísimo Zelda con DLSS 5:
AMDGPU retweeted
Stable lighting, frame after frame: new research on a one-step diffusion model for GI, presented at #ECCV2026.
💡 Geometrically aligned, temporally stable lighting.
🧠 Aligns well with real-world features & illumination.
Read more ⬇️
gpuopen.com/learn/temporally…
AMDGPU retweeted
You missed the entire point of local AI
1. Data Privacy
2. Censorship control
3. Cost
The prospect of unlimited, private, uncensored, tokens far outweighs the relatively small fixed upfront cost.
I wouldn't expect a former meta/Microsoft employee to understand.
i still don’t understand the “local AI is free” point of view
it literally costs thousands of dollars to get started on anything useful, a small data center worth of cost and complexity to do something near frontier
if you sum up all the money spent buying hardware and paying power bill and your time spent fiddling with it, it’s almost decades worth of consumer subscription from frontier labs which give much better intelligence at much larger concurrent throughput
the only reasonable explanation for doing full-on local AI seems:
1. it’s fun
2. it’s cool
3. you really hate the labs
i do think we can look at it in another way - for those who already bought a capable computer, use the idle gpu capacity to do some secondary tasks that don’t require frontier intelligence while you continue to rely on cloud models for primary usage
did i miss anything? genuinely curious if there’s a sound argument
AMDGPU retweeted
Clem: "They will be a great home for Open models"
Jensen: "I paid $12.9 Billion because I just love open models so much."
Lies.
Huggingface is now owned by The closed AI agenda:
- Data collection and surveillance
- Token Monopolization
- State funded censorship/propaganda
AMDGPU retweeted
📈 CPU Retail Sales AMAZON US 🇺🇸 - August 2026
AMD holds nearly 79% of unit sales, while X3D CPUs alone account for more than 37% of the entire CPU market.
ℹ️ Units
AMD: 28 750 units sold (78.7%), ASP: $314
Intel: 7 800 units sold (21.3%), ASP: $304
Total: 36 550 CPUs
ℹ️ Revenue
AMD: $9 033 082 (79.2%)
Intel: $2 369 870 (20.8%)
Total revenue: $11 402 951
ℹ️ By socket
AM5: 17 050 (46.6%)
AM4: 11 700 (32.0%)
1851: 4 050 (11.1%)
1700: 3 700 (10.1%)
1151: 50 (0.1%)
ℹ️ By gen / arch
Zen 5 X3D: 8 000 (21.9%)
Zen 5: 4 500 (12.3%)
Zen 4 X3D: 2 400 (6.6%)
Zen 4: 2 150 (5.9%)
Zen 3 X3D: 3 100 (8.5%)
Zen 3: 8 550 (23.4%)
Zen+: 50 (0.1%)
Arrow Lake: 4 050 (11.1%)
Raptor Lake Refresh: 3 000 (8.2%)
Raptor Lake: 50 (0.1%)
Alder Lake: 650 (1.8%)
Coffee Lake: 50 (0.1%)
ℹ️ By SKU (units / share)
Ryzen 7 9800X3D: 5 000 (13.7%)
Ryzen 5 5500: 4 000 (10.9%)
Ryzen 7 5800X3D: 3 100 (8.5%)
Ryzen 5 9600X: 3 000 (8.2%)
Ryzen 7 7800X3D: 2 000 (5.5%)
Ryzen 9 9950X3D: 2 000 (5.5%)
Core Ultra 7 270K Plus: 2 000 (5.5%)
Ryzen 5 5600X: 1 000 (2.7%)
Ryzen 5 7600X: 1 000 (2.7%)
Core Ultra 5 250K Plus: 1 000 (2.7%)
Core i9-14900K: 600 (1.6%)
Ryzen 7 5700: 550 (1.5%)
Ryzen 5 5600: 500 (1.4%)
Ryzen 7 5700G: 500 (1.4%)
Ryzen 7 5700X: 500 (1.4%)
Ryzen 7 5800XT: 500 (1.4%)
Ryzen 7 7700X: 500 (1.4%)
Ryzen 7 9700X: 500 (1.4%)
Ryzen 7 9850X3D: 500 (1.4%)
Ryzen 9 5900XT: 500 (1.4%)
Ryzen 9 9900X: 500 (1.4%)
Ryzen 9 9950X: 500 (1.4%)
Core Ultra 9 285K: 500 (1.4%)
Core i7-14700K: 500 (1.4%)
Core i7-14700KF: 500 (1.4%)
Ryzen 5 7600X3D: 400 (1.1%)
Core i5-14400F: 350 (1.0%)
Core i5-14600KF: 350 (1.0%)
Ryzen 9 7900X: 300 (0.8%)
Ryzen 9 9950X3D2 Dual Edition: 300 (0.8%)
Core i9-14900KF: 300 (0.8%)
Ryzen 5 5600GT: 200 (0.5%)
Ryzen 7 5800X: 200 (0.5%)
Ryzen 9 9900X3D: 200 (0.5%)
Core Ultra 5 225: 200 (0.5%)
Core i9-12900KF: 200 (0.5%)
Ryzen 5 5600G: 100 (0.3%)
Ryzen 5 7600: 100 (0.3%)
Ryzen 5 8500G: 100 (0.3%)
Ryzen 5 8600G: 100 (0.3%)
Core Ultra 5 245K: 100 (0.3%)
Core Ultra 7 265: 100 (0.3%)
Core Ultra 7 265KF: 100 (0.3%)
Core i3-14100: 100 (0.3%)
Core i5-12400F: 100 (0.3%)
Core i5-12600KF: 100 (0.3%)
Core i5-14400: 100 (0.3%)
Core i5-14600K: 100 (0.3%)
Core i7-12700KF: 100 (0.3%)
Ryzen 3 3200G: 50 (0.1%)
Ryzen 7 8700G: 50 (0.1%)
Core Ultra 5 245KF: 50 (0.1%)
Core i3-14100F: 50 (0.1%)
Core i5-12400: 50 (0.1%)
Core i5-13600K: 50 (0.1%)
Core i7-12700K: 50 (0.1%)
Core i7-14700: 50 (0.1%)
Core i7-9700K: 50 (0.1%)
Core i9-12900K: 50 (0.1%)
$AMD $INTC
AMDGPU retweeted
DLSS 5 Testing Ends in a Melted RTX 5090 Connector, Power Shoots Past 600 W tpu.me/vehw
AMDGPU retweeted
📈 GPU Retail Sales August 2026 (mf) 🇩🇪
AMD takes a commanding volume lead, driven by the RX 9070 XT and RX 9060 XT.
RDNA 4 alone accounts for more than half of all GPUs sold, while Nvidia remains heavily concentrated in Blackwell.
ℹ️ Units
AMD: 1,570 units sold, 55.87%
Nvidia: 1,110, 39.50%
Intel: 130, 4.63%
Total: 2,810 units
ℹ️ by GPU
RX 9070 XT: 870 (31.0%)
RX 9060 XT: 600 (21.4%)
RTX 5070 Ti: 270 (9.6%)
RTX 5060 Ti: 250 (8.9%)
RTX 5080: 240 (8.5%)
RTX 5070: 190 (6.8%)
RX 9070: 80 (2.8%)
RTX 3050: 50 (1.8%)
ARC B580: 30 (1.1%)
RTX 5060: 25 (0.9%)
ARC B570: 20 (0.7%)
RX 9070 GRE: 20 (0.7%)
ARC A310: 20 (0.7%)
Arc Pro B60: 20 (0.7%)
RTX 3060: 10 (0.4%)
RTX 5090: 10 (0.4%)
ARC A750: 10 (0.4%)
ARC A770: 10 (0.4%)
Arc Pro B70: 10 (0.4%)
Arc Pro B65: 5 (0.2%)
Arc Pro B50: 5 (0.2%)
ℹ️ by Architecture
RDNA 4: 1,550 units (55.2%)
Blackwell: 1,039 units (37.0%)
Battlemage: 85 units (3.0%)
Ampere: 71 units (2.5%)
Alchemist: 45 units (1.6%)
RDNA 3: 20 units (0.7%)
ℹ️ VRAM Share (8 GB vs 16 GB+)
16 GB+: 2,130 units (75.8%)
≤ 12 GB: 680 units (24.2%)
ℹ️ Top 3 best selling SKUs
1. PowerColor Radeon RX 9070 XT Reaper 16 GB: 140 units
2. Sapphire Radeon RX 9060 XT Pulse 16 GB: 140
3. MSI GeForce RTX 5060 Ti Ventus 2X OC Plus 8GB: 130 units
$AMD $INTC $NVDA
AMDGPU retweeted
Big thank you to David McAlvaney and his team at McAlvany Financial for having me on yesterday to talk about the AI bubble! Give them a thumbs up and be sure to tell them Nobody sent you.
youtu.be/dtMqqnHcFHI?si=ic1b…
