topic
gpu
11 dispatches on gpu.
27,500 GPUs for Japan: NVIDIA's Vera Rubin Bet Is About Cost Per Token, Not Peak Performance
NVIDIA's Vera Rubin AI factory in Japan—27,500 Rubin GPUs, 140 megawatts, backed by the Japanese government—signals that the real AI arms race is now about cost-per-token infrastructure, not raw benchmark scores.
FSR 4.1 Reaches Older AMD Cards, and It's the Most Exciting GPU News of a Very Sad Year
AMD's FSR 4.1 just landed on RX 7000-series GPUs via a driver update — a meaningful software win in a year when neither AMD nor Nvidia has shipped new gaming hardware, thanks to a memory shortage that's reshaping the whole industry.
NVIDIA's RTX Spark Is a PC That Happens to Play Games, Not a Gaming PC
NVIDIA's RTX Spark superchip pairs a Grace CPU with Blackwell GPU and 128GB unified memory in thin laptops shipping this fall. It's built for AI agents and creative workflows — gaming is the third priority, not the first.
NVIDIA's Software Gambit: Selling 8GB Cards While Promising a Fix That Doesn't Exist Yet
NVIDIA launched the RTX 5060 with just 8GB VRAM and is betting that Neural Texture Compression will make it viable—but the technology is unshipped, adoption is near zero, and real games are already punishing 8GB owners right now.
NVIDIA Wants to Own the Whole PC Now
At Computex 2026, NVIDIA unveiled RTX Spark — a Blackwell GPU fused with a MediaTek-designed CPU and unified memory. It's NVIDIA's first real play for the consumer PC stack, and the real competition isn't Apple. It's the idea that the PC platform should have an owner at all.
AMD's RX 9070 GRE Goes Global, and the Price Tag Tells You Everything About This Market
AMD's China-exclusive Radeon RX 9070 GRE launches worldwide at $549 with cut-down Navi 48 silicon and 12GB GDDR6. Here's what the pricing strategy reveals about the midrange GPU market in 2026.
Apple's Siri Runs on Google Now. That's Not a Bug — It's the Compute Crisis Made Visible.
Apple's WWDC Siri AI reveal and Google's deal with SpaceX arrived days apart. Together, they expose the structural reality of the AI era: even the biggest device makers and hyperscalers can't build fast enough alone.
Gamers Are Funding the AI Buildout Now: The RTX 5090 Is the Receipt
The RTX 5090 sells for $2,700 to over $5,000 because the memory it needs is being bought up for AI servers. This isn't a shortage story — it's a story about who pays for the AI boom, and the answer is increasingly: gamers.
The Midrange GPU Is Being Squeezed Out — And the Memory Crisis Is Finishing the Job
Intel killed its Arc B770 midrange GPU, AMD pivoted the RX 9060 XT lineup toward 8GB, and a grinding DRAM shortage is erasing the $300–$450 GPU tier that sustained PC gaming for two decades.
Nvidia's RTX Spark Is a Bet That AI PCs Are Just APUs With Better Marketing
Nvidia's RTX Spark superchip combines CPU and GPU on one die and claims to 'reinvent the PC.' The architecture isn't new — AMD and Intel have been doing this for years. Nvidia's real edge is CUDA, DLSS, and brand power. That's real, but it's not a revolution.
Intel's Arc B390 Is the Integrated GPU That Makes Entry-Level Discrete Cards Obsolete
Intel's Core Ultra Series 3 Panther Lake chips pack an Arc B390 iGPU that matches the RTX 4050 laptop GPU and runs Battlefield 6 at high frame rates — no discrete card required. Here's what that means for gaming laptop economics.