Playground Games released the PC hardware requirements for Fable, listing an RTX 2060 as the minimum GPU and RX 6700 XT as recommended. Steam page mentions a potential 60 FPS cap.
ASRock launched the B550M Pro-A WiFi 6E and increased AM4 production, driven by current memory pricing. Bearish: Capacity expansion may exacerbate DDR4 motherboard oversupply.
Architect Labs unveiled Redwood, an AI accelerator verified by AI in under two weeks, targeting edge devices. While projections suggest 1.75x throughput of Nvidia's Jetson Orin Nano on Samsung 8nm, the chip remains an FPGA prototype.
Samsung introduced the industry's first LPDDR5X-PIM memory, achieving 3.01x faster AI inference speeds compared to standard LPDDR5X. This technology addresses the high cost of HBM by offering a cheaper in-memory processing alternative for AI workloads. Watch: Potential demand shift from HBM to LPDDR5X.
AMD will introduce native UCIe 1.1 support in select Versal adaptive SoCs to enable low-power, high-bandwidth communication within a package. This modular chiplet approach aims to reduce latency, power, and development costs. Currently, this news has no direct impact on spot prices or supply of existing components.
Nvidia acquired Cloverleaf to secure land and power for datacenters, addressing infrastructure bottlenecks for its liquid-cooled Blackwell and Rubin GPUs. This move supports GPU demand but highlights the critical need for liquid cooling infrastructure. Bullish for liquid cooling and power equipment suppliers.
Signaloid CEO announced a new ASIC tape-out in May 2026, leveraging distribution-extended compute hardware (UxHw) to achieve 10-100x performance improvements in robotics and physical AI applications.
Samsung's zHBM reduces thermal resistance by 50% and increases density 10x, while liquid cooling penetration in AI chips is projected to rise to 60% by 2027. As 3D stacking exacerbates thermal challenges, the industry is shifting toward CPO and STCO solutions. Bullish for advanced packaging and cooling technologies.
SK Hynix is establishing a HBM design team in Silicon Valley to deepen co-design with US chipmakers like Nvidia and AMD. This strategic move signals a shift towards customized, jointly developed memory solutions, which is likely to tighten supply and secure premium pricing for HBM4.
Google expands its TPU roadmap, adding Marvell to its supplier base alongside Broadcom and MediaTek, with AMD potentially joining the TPU v10 race. This shift marks a move towards multi-source supply for custom silicon.
SK hynix expands its San Jose HBM team to deepen base-die co-design with major customers like NVIDIA and AMD, signaling a strategic push for HBM4 early engagement.
Cerebras Systems launched the CS-4 AI accelerator, achieving up to 30x faster performance than GPU solutions. Built on three WSE-3 Turbo chips, it delivers 750 PFLOPs of compute and 10x throughput per watt. Bullish: The launch signals a significant competitive threat to incumbent GPU providers in the AI inference market.
Samsung Display plans to expand OLED production capacity in Asan (Plant 2) starting 2027, targeting completion in early 2028, driven by foldable demand and Apple's rumored entry. Concurrently, rising memory costs are expected to keep mobile manufacturing expenses elevated in 2H26.
Gigabyte launched two new B450 motherboards targeting budget builds and AM4 upgrades. These boards support Ryzen 5000 but lack PCIe 4.0 and DDR5. No pricing or availability data provided.
Delta Electronics and Lite-On Technology are increasing capital expenditure to 9% of revenue to meet the surging demand for power and cooling solutions in AI data centers. Chipmakers like Nvidia and AMD are refreshing platforms annually, driving up power density and thermal requirements. This capacity expansion is necessary to avoid missing design cycles amid intense competition.
China's high-end AI chip market is projected to reach nearly 90% domestic share by 2026, with Huawei expected to surpass Nvidia, leaving overseas suppliers with only about 10% of the market.
AMD CEO Lisa Su unveiled a full-stack AI strategy focusing on rack-scale systems for agentic workloads. The plan integrates GPUs, CPUs, networking, and software. This strategic shift indicates future demand but provides no immediate supply or pricing data.
AMD CEO Lisa Su outlined a full-stack AI strategy focusing on rack-scale systems for agentic workloads. This strategy encompasses GPUs, CPUs, networking, and software. It signals a strategic shift but lacks immediate supply chain data.
Samsung and SK hynix reported significant ASP increases for DRAM and NAND in 2Q26, with HBM4 ASPs reaching $31-32/GB. Driven by customer diversification and general memory price recovery, ASP growth is becoming a key earnings lever. Bullish: Supports strong 2H earnings outlook for Samsung and SK hynix.
Intel upsized its common stock offering to $20 billion, aiming for 'general corporate purposes' including potential capital expenditures for its foundry business. Analysts debate the adequacy of this amount for significant expansion, with some suggesting it targets wafer fabrication equipment (WFE) to meet AI demand. The move signals an intent to boost capacity but raises questions about the scale of future supply growth.
Whose chips are really inside the hottest new products? We open them up and show you, one part at a time.
iPhone DuoFirst-pass BOM: 37 parts, 12 brands and 16 public MPNs; A20 Pro, C2, N1 and the power/RF chains are located, while the NAND brand stays unconfirmed.See the full teardownAll teardowns| Analog | $0.68 | −5.9% |
|---|---|---|
| Connector | $0.75 | +0.2% |
| CPU | $386.74 | −21.3% |
| DDR4 | $88.40 | −0.2% |
| DDR52026-W13 | $14.89 | −11.2% |
| GPU | $478.20 | +8.2% |
| MCU | $0.68 | −61.7%Unusual |
| MLCC | $0.01 | +54.0%Unusual |
| MOSFET | $0.12 | −82.8%Unusual |
| PMIC2026-W13 | $0.44 | −12.5% |
| SSD/HDD | $288.43 | n/a |