AI-GENERATIVE.ORG runs on the AI-Generative Platform.See how it works →

Market · Infrastructure

AI Chips & Compute

GPUs, accelerators and the silicon that trains and runs AI.

  • 9 companies tracked
  • 246 stories this week
  • 114 outlets

This week at a glance

Updated Oct 19
  1. Intel launched its 480GB GPU D-series for AI inference.

  2. IBM landed a $240M Together AI deal to build an Nvidia-powered neocloud cluster, while Giga Computing powers Scaleway’s new AMD Instinct GPU services.

  3. Samsung reportedly plans to use TSMC technology for Nvidia’s next-generation memory chips.

  4. NVIDIA and Nebius expanded the AI Builder Program in Europe.

Where the market stands

Week of Oct 19 · reviewed by our editor

Nvidia has made RTX Spark PCs available and published guidance on using them to code, test and deploy on a single machine. NVIDIA NewsroomNVIDIA Developer

Scaleway’s new AMD Instinct GPU services use Giga Computing systems, while AMD has unveiled Helion, an open rack blueprint for agentic AI. HPCwireData Centre Magazine

TSMC says AI chip demand remains far above its supply and is exploring expansion in Texas as demand outpaces production. Yahoo FinanceThe American Bazaar

Intel and Qualcomm are courting Tenstorrent, while AMD’s proposed acquisition of World Labs would bring its research team into the chipmaker, subject to closing conditions. Startup FortuneAI-Generative

Reports flag a possible shakeout among AI-focused cloud providers, with exposure for Nvidia, and a capital warning from Broadcom’s chief executive. The Information24/7 Wall St.

Execution remains unproven in specific areas: cuPhoton users still need to adapt and integrate reference code, while the benefits of AMD’s proposed World Labs acquisition have not yet been demonstrated. AI-GenerativeAI-Generative

Outlook Our take

Scenario 1

More routes to AI compute

Analysis: New AMD cloud services and Nvidia desktop systems could broaden how businesses access AI compute, with adoption depending on whether these offerings support useful workloads without excessive integration work.

Watch for
  • More cloud providers announce AMD Instinct services.
  • Businesses publish production deployments using RTX Spark.
  • Vendors publish workload results that include integration requirements.
Scenario 2

Supply and capital constraints slow deployment

Analysis: Chip shortages and financial pressure on AI-focused cloud providers could keep deployment uneven, even with strong demand. Expansion plans may take longer to translate into usable capacity.

Watch for
  • TSMC continues to report demand exceeding available supply.
  • AI-focused cloud providers announce consolidation or capacity cuts.
  • Announced chip production expansions move into construction or production.
Scenario 3

Chip vendors deepen their software and research ties

Analysis: AMD’s proposed World Labs acquisition and interest in Tenstorrent could lead to closer combinations of silicon, software and research teams, although announced deals may precede demonstrable customer benefits.

Watch for
  • AMD announces completion or revised terms for the World Labs acquisition.
  • Intel or Qualcomm announces a formal transaction or partnership with Tenstorrent.
  • Combined teams release products with documented customer deployments.

Companies

Busiest in the news this week first

Latest news

Refreshed at 6am and 3pm ET · headlines link to the publishers

From AI-Generative