This week on Tom's Hardware Premium: October 10, 2026
- ID
- 34047
- Status
- summarized
- Published
- 10 Oct 2026, 11:00 PM
- Fetched
- 10 Oct 2026, 11:54 PM
- Provider
- Tom's Hardware
- Category
- technology
- Original URL
- https://www.tomshardware.com/tech-industry/this-week-on-toms-hardware-premium-october-10-2026-tencent-grabs-bargain-chips-qualcomm-huawei-ip-deal-groq-lawsuit-and-more
- Source URL
- https://www.tomshardware.com/feeds/all
Summary
- Score
- 4.5
- Created
- 10 Oct 2026, 11:54 PM
- Tags
- Audience
- developersai_ml_learnerssaas_founders
What happened
Tom's Hardware's weekly Premium roundup for October 10, 2026 leads with a report that Tencent has agreed to lease high-performance AI chips from Oracle in a deal valued at $7 billion over five years, alongside items on a Qualcomm-Huawei IP deal and a Groq lawsuit. It also covers research arguing software optimization could be the easiest fix for hyperscalers' AI power squeeze, with data center demand projected to rival Japan's electricity usage by 2030. The excerpt is truncated at the Tencent item, so the Qualcomm, Huawei, and Groq details are headline-only.
Why it matters
The one usable number here is the shape of the deal, not the vendor: $7B over five years to rent compute rather than buy it. If you are planning GPU capacity or cloud spend for an AI product, that framing — lease/rent over capex — is now being used at ten-figure scale by a major buyer, and the power-constraint angle (data center demand approaching Japan's national electricity usage by 2030) suggests capacity and power, not silicon, are the binding constraint you should model into your assumptions. Beyond that, this is a paywalled index of articles; you cannot act on the Qualcomm-Huawei or Groq items without reading the originals.
Discussion angle
If compute is increasingly rented at multi-billion-dollar scale rather than owned, does that make regional AI capacity cheaper and more available over time, or does it just concentrate pricing power in a few hyperscalers? Worth asking what this means for anyone in Southeast Asia buying GPU time today.