The cost of PC memory has risen to unprecedented levels as the artificial intelligence boom continues to absorb huge amounts of the world’s DRAM production. The situation has become so severe that major technology companies and hyperscalers are reportedly locking up memory supplies well into 2028. That leaves very little leftover inventory for things like […]
Read Full Article →
Nvidia has built one of the strongest competitive positions the technology industry has ever seen. It dominates the market for artificial intelligence accelerators, owns the CUDA software ecosystem that developers have spent years learning and has expanded beyond individual GPUs into networking, rack-scale systems and increasingly complete AI infrastructure. OpenAI’s first custom processor is not […]
Read the Full Report →
AI was expected to drive unprecedented demand for GPUs, accelerators and memory. It is now creating a second semiconductor market in the transformers, power supplies, cooling systems and electrical infrastructure surrounding those processors. The transformer does not usually receive much attention in discussions about artificial intelligence. It lacks the glamour of a new GPU, the […]
NVIDIA has moved its Groq 3 LPX inference accelerator into full production, adding specialized hardware to its Vera Rubin platform as the chipmaker addresses a key performance problem facing agentic AI. The new accelerator is built to generate tokens at high speed after an AI model has processed its input. This capability is particularly important […]
Samsung and SK hynix are pursuing decidedly different strategies to overcome the bandwidth, power, and thermal limitations facing high bandwidth memory (HBM) as AI accelerators become increasingly powerful. At Hot Chips 2026, Samsung outlined a roadmap that moves from conventional HBM to custom HBM and ultimately to a more radical architecture called zHBM. SK hynix, […]
Leads Futurum's Semiconductors Practice covering edge AI, generative AI, and hyperscale silicon architectures.
25+ years covering AI and enterprise software. Co-founded 451 Research. London-based analyst covering AI platforms and infrastructure.
Covers intelligent devices, edge AI, sustainability, and the convergence of silicon and software at the edge.
Covers enterprise applications, digital workflows, and the intersection of AI with enterprise software strategies.
Marvell Technology Inc. shares spiked nearly 10% following a regulatory filing revealing that Google has secured the option to buy up to $12.2 billion in Marvell stock. The agreement signals a major expansion in their custom silicon partnership as tech titans rush to build cheaper, viable alternatives to NVIDIA Corp.’s dominant artificial intelligence (AI) processors. […]
Nokia and NVIDIA launch an AI-RAN platform targeting 2x spectral efficiency by 2028 — here’s how the claims compare with what’s shipping today.
Intel Xeon expansion gets €5 billion boost as Intel invests heavily in Ireland fab capacity for next-generation CPUs with EUV-compatible capacity.
FOXTRON’s adoption of Dimensity AX C-X1 highlights MediaTek’s push into AI-defined vehicles through integrated AI, connectivity, and cockpit platforms.
Institutional-grade research, proprietary data, and analyst access — reserved for premium subscribers
Proprietary capacity projections for TSMC's advanced packaging, including customer allocation estimates and pricing trajectory analysis.
PREMIUM ONLYQuarterly market share data for AI training and inference chips across cloud hyperscalers, enterprise, and edge deployments.
PREMIUM ONLYA 40-page report covering chiplet architectures, advanced packaging technologies, and the companies positioned to dominate the next era of semiconductor manufacturing.
PREMIUM ONLYSiFive’s IP vendor perspective on RISC-V expanding beyond embedded into the Performance CPU segment — including the P550 family, the RVA23 profile, and out-of-order CPU IP for AI workloads.
CoreWeave-sponsored webinar examining why operationalizing AI inference at scale — not training new models — is the real challenge facing enterprises in 2026.
Watch On-Demand →On-demand webinar with CharacterAI’s Chief Architect on how AMD MI325X accelerators with 256GB HBM and DigitalOcean enabled production-scale inference on half the hardware.
Watch On-Demand →RackN webinar on building AI infrastructure automation for bare metal at scale — covering operator challenges and high-speed, large-scale bare metal management with AIOps.
Watch On-Demand →RISC-V is commonly seen as an ISA for embedded products; for smaller devices that require low power consumption but aren’t very complex and can tolerate lower performance. SiFive, a leading RISC-V IP vendor, is proving that incorrect with its Performance family of products. They recently launched the latest product in the series, the P550, that supports the RVA23 profile: a crucial enabler for SW developers and the industry as a whole. In this webinar, we discuss how SiFive has been able to deli
Watch On-Demand →There has never been a megatrend as transformative as AI. 2023 unlocked new possibilities, 2024 brought new proof of concepts and 2025 was the year to prove the ROI and business value generation of AI. 2026 is the year of the AI reality check and organizations that successfully operationalize AI to drive concrete efficiency and measurable impact will be the ultimate winners.
Watch On-Demand →As AI adoption accelerates, organizations are discovering that the real challenge is no longer building models—it’s running them reliably at scale. This webinar explores the growing “production AI gap,” where promising prototypes fail to translate into consistent, real-world outcomes. Attendees will gain a clearer understanding of why inference—not training—is becoming the defining layer of AI, and what it takes to operationalize AI systems in environments shaped by unpredictable demand, latency
Watch On-Demand →
