All articles tagged with #gpus
Advanced Micro Devices jumped after unveiling a multiyear agreement to supply Meta with 6 gigawatts of GPUs for Meta’s AI infrastructure, plus a performance-based warrant for up to 160 million AMD shares (~10% stake), signaling a bold push into AI data centers and potentially widening the gap with Nvidia despite dilution concerns.
Nvidia still dominates AI compute with GPUs and faster growth, while Alphabet offers a niche TPU alternative and steadier, diversified cash flows from Google Cloud and other non‑AI businesses. Nvidia appears cheaper on forward earnings with higher upside, but Alphabet provides resilience and a balance of AI exposure; both are solid AI plays, with Nvidia carrying AI spending risk and Alphabet offering steadiness.
WIRED reports Nvidia’s expanded deal with Meta—billions in GPUs plus Grace CPUs and the broader push into standalone CPUs and low-latency AI tech—to build hyperscale data centers for training and inference, signaling a shift from pure GPU purchases to an end-to-end AI compute stack. Analysts say CPUs are growing in importance for agentic AI, though GPUs remain central for heavy workloads, with Meta planning roughly $115–$135 billion in AI infra this year.
With hyperscalers like Amazon, Alphabet, and Meta driving a surge in AI infrastructure spend, Nvidia is set for continued growth in 2026. Despite AI-bubble concerns, the piece argues that ongoing AI improvements will sustain demand for Nvidia GPUs, and data-center capex is projected to rise from about $600B in 2025 toward $3–4T by 2030, suggesting the stock could be higher a year from now.
Alphabet beat Q4 estimates but warned that 2026 capex will surge to $175-185 billion to fuel AI, effectively backing the AI infrastructure boom that benefits Nvidia’s data-center GPUs; its partnership with Nvidia includes early access to the Vera Rubin platform, underscoring how hyperscalers’ AI spending sustains demand for Nvidia hardware even as stock rallies wobble. Meta’s similar capex trend also reflects the broader AI investment cycle driving Nvidia’s growth prospects.
Intel is reportedly launching the Arc Pro B70, a professional GPU built on the Battlemage G31 (BMG-G31) die, featuring 32 Xe2 cores (4096 shaders) and 32 GB GDDR6 VRAM on a 256-bit bus, with an estimated ~300 W TDP; the Pro card is said to arrive soon, while a consumer variant (Arc B770) is rumored to have 16 GB VRAM and launch later.
China is drafting purchase rules for Nvidia's H200 GPUs, likely including quotas and government approvals to control who can buy them; Alibaba, Tencent, and ByteDance are eyeing purchases pending these rules, while Nvidia shares rose in premarket trading on the news.
The article criticizes several GPU releases in 2025, highlighting the poor value and performance of 8GB models like Nvidia's RTX 5060 Ti and AMD's RX 9060 XT, which are seen as misguided and damaging to consumers, with Nvidia's RTX 5080 also underwhelming due to insufficient VRAM and high prices. Overall, the generation is marked by questionable product decisions and a lack of meaningful upgrades, especially in the entry-level and high-end segments.
The article reviews the top gaming GPUs of 2025, highlighting NVIDIA's RTX 5090 as the best high-end option, AMD's RX 9070 XT as the best overall, and the RX 9060 XT 16 GB as the best budget choice, with a focus on performance, availability, and technological advancements amidst supply chain challenges.
AMD is developing new 'Max' CPUs, such as the Ryzen AI Max+ series, which could challenge low-end GPUs by offering integrated graphics capable of high-performance gaming, potentially replacing or supplementing traditional GPUs in budget-friendly gaming devices and handheld PCs.
The Outer Worlds 2, built on Unreal Engine 5 with advanced graphics features, has been tested across over 30 GPUs to evaluate performance, image quality, and VRAM usage, providing insights into hardware needs for optimal gameplay.
The Trump administration is considering implementing tariffs on consumer CPUs and GPUs based on the value of their foreign chip content, which could significantly increase prices if enacted, aiming to promote domestic chip manufacturing but potentially raising costs for products like AMD Ryzen and NVIDIA GPUs.
Intel will continue to develop its own GPU products despite Nvidia's contribution of GPU chiplets for Intel CPUs, with the collaboration seen as complementary rather than replacing Intel's graphics efforts. However, skepticism remains about Intel's commitment to graphics development given Nvidia's dominance and Intel's cost-cutting strategies.
Asus has launched the NUC 15 Performance mini-PCs featuring Nvidia GeForce RTX 5060 and RTX 5070 GPUs, with various ports including DisplayPort 2.1 and HDMI 2.1, and configurable hardware options. The units are available in the Eurozone starting at €1,632, but global pricing is not yet disclosed.
China has launched its first 6nm GPUs, the Lisuan 7G106 and 7G105, with up to 24 GB memory, outperforming the RTX 4060 in synthetic benchmarks and capable of running demanding games at 4K with high FPS, marking a significant advancement in China's GPU segment.