AI chips news today centres on three announcements from 9 and 10 November 2026: Google introduced two new TPU systems, AMD started shipping sixth-generation EPYC processors, and Nvidia joined a Korean deal to connect local AI chips to its platform. The announcements landed in the same 48 hours and covered training hardware, server processors and sovereign AI infrastructure.
This article reflects reporting available as of 10 November 2026. Figures come from company announcements and market reporting, and vendor performance claims are labelled as claims.
Key Takeaways
- Google announced TPU 8tx and TPU 8i Max on 9 November 2026, expanding the eighth-generation TPU family to four systems.
- AMD started general availability of sixth-generation EPYC server processors on 10 November 2026, with up to 256 cores per socket.
- Nvidia, Rebellions and SK Telecom agreed to link Korean R200 NPUs to Nvidia racks, with deployment expected in 2028.
- AMD processors power the host of Google’s confidential TPU 8tx system, a notable shift from Google’s own Axion chip.
- Chip stocks split on 10 November 2026: AMD rose sharply while Nvidia fell after its board opened an independent investigation.
Table of Contents
The Short Answer
The short answer is that the AI hardware news from 9–10 November 2026 shows competition spreading beyond accelerators. Google expanded its custom silicon, AMD shipped a new server CPU line aimed at agent workloads, and a Korean partnership showed that national AI plans now mix local chips with Nvidia networking. Memory supply, security and export rules shaped every announcement.
AI Chips News Today: The 9–10 November Headlines
Four stories defined the two days. Google led with new systems for secure inference and large-context work. AMD followed with server processors built for the agent era. Korea added a sovereign AI deal. Nvidia closed the period with a board investigation and falling shares.
| Story | Company | What happened | Why it matters |
|---|---|---|---|
| TPU 8tx and TPU 8i Max | Two new eighth-generation TPU systems announced 9 November 2026 | Targets confidential inference and the memory wall | |
| 6th Gen EPYC general availability | AMD | Zen 6 server processors shipped 10 November 2026 | CPUs take a bigger role in agent workloads |
| Korea sovereign AI deal | Nvidia, Rebellions, SK Telecom | R200 NPUs to join Nvidia racks via NVLink Fusion | Local silicon plus Nvidia networking for national AI |
| Board investigation | Nvidia | Independent investigation opened 8 November 2026 | Shares fell while the wider chip index rose |
What Did Google Announce on 9 November?
Google announced TPU 8tx and TPU 8i Max on 9 November 2026, according to SiliconANGLE and eWeek. The two systems join TPU 8t and TPU 8i, which Google introduced in April 2026, to complete a four-system eighth-generation TPU family. Google described the new systems as purpose-built for the agentic era, when AI agents run long tasks across large amounts of data.
TPU 8i Max pairs two TPU 8i chips in one package. Google says the design doubles compute capacity, on-chip SRAM and high-bandwidth memory bandwidth compared with TPU 8i. The target is the memory wall: models with very large context windows need more memory bandwidth than a single chip provides. TPU 8tx takes the other path. It is built for confidential computing, so regulated enterprises and governments can run frontier models on sensitive data with hardware-backed protection during processing.
Why Did Google Choose AMD Processors for the TPU 8tx?
Google paired TPU 8tx with fifth-generation AMD EPYC processors as the host CPU, according to AMD and TheStreet. The choice stands out because Google runs its own Axion processors as hosts on TPU 8t and TPU 8i. TPU 8tx uses the same silicon as TPU 8t, yet Google selected AMD silicon for the host role in the confidential system.
Security explains the switch. AMD says its Secure Encrypted Virtualization technology extends confidential-computing protection from the host CPU to the connected accelerators. AMD also says Google security experts reviewed its security processor firmware in depth. Google Cloud said the setup brings hardware-enforced confidential computing to cloud TPUs for the first time. The system supports private model fine-tuning, distillation and reinforcement learning without moving sensitive data between separate clusters, according to eWeek. Agentic AI workloads in banks and government agencies are the stated audience.
What Did AMD Launch on 10 November?
The AI chip news from AMD on 10 November 2026 was general availability of sixth-generation EPYC server processors, codenamed Venice and built on the Zen 6 architecture. AMD announced the launch in a press release carried by GlobeNewswire on 10 November 2026. Systems from leading server vendors are shipping now, while cloud instances from Microsoft Azure and Oracle Cloud Infrastructure are expected soon.
The lineup scales to 256 cores and 512 threads per socket. The flagship EPYC 9996 runs at up to 4.1 GHz boost with 1,024 MB of L3 cache, while a high-frequency variant reaches 5.3 GHz. The chips use the SP7 socket, 16 channels of DDR5 memory and PCIe Gen 6. AMD positions the line for three jobs: general enterprise services, high-frequency host nodes for GPU systems, and dense sandboxes for AI agents. AMD claims up to 3.3 times more agents per watt than Nvidia’s Vera CPU. That figure is a vendor claim with no independent testing published as of 10 November 2026.
What Is the Nvidia, Rebellions and SK Telecom Deal?
Nvidia, Rebellions and SK Telecom announced a collaboration on 9 November 2026 in Seoul to build Korean sovereign AI infrastructure, according to a GlobeNewswire release and Aju Press. Rebellions will connect its next-generation R200 neural processing units to Nvidia’s platform through NVLink Fusion. The combined systems will use Nvidia’s MGX rack architecture, and SK Telecom plans to deploy them in its AI data centres.
The deal is a development plan, not a product launch. R200 systems are expected to be ready for deployment in 2028. No performance, power or price figures were published. The collaboration is supported by Korea’s Ministry of Science and ICT. SK Telecom plans to connect the hardware with its own models and public services under Korea’s AI for All project. AI agent use cases in Korean language services and public-sector work are among the stated goals.
How Did Chip Stocks and Suppliers React?
Markets split on Monday 10 November 2026. AMD shares climbed more than 10 percent, according to Longbridge, and closed at $508.58, according to TheStreet. Nvidia fell roughly 6 percent on the day, according to Longbridge, after its board said on Sunday 8 November 2026 that it was opening an independent investigation. Morningstar reported Nvidia shares down 4.77 percent at $177.74 in morning trading, while the PHLX Semiconductor Sector Index rose 4.33 percent to 9,796.62 at the same time.
The wider semiconductor AI news included quarterly results from four suppliers, reported in a Morningstar roundup on 9 November 2026. Onsemi reported third-quarter revenue of $1,722 million, up 11 percent year over year, and raised its AI data centre 2030 revenue target by $500 million to more than $3 billion. Lattice reported third-quarter revenue of $291.0 million, with FPGA revenue up about 90 percent on data centre AI demand. NXP reported third-quarter revenue of $3.79 billion, up 19 percent year over year. Astera Labs said its Scorpio fabric switches became its largest product family in the third quarter. A supply-chain report from Industrial Times, carried by KuCoin on 10 November 2026, said Nvidia may reduce chip orders amid stricter export reviews. That report is unconfirmed, and Nvidia has not confirmed any order cut.
What Comes Next for AI Hardware?
The next AI hardware news to watch is Nvidia earnings on 17 November 2026, when investors expect comment on custom-chip competition. Pricing and availability for the new Google TPU systems also remain unannounced, and independent benchmarks for AMD’s EPYC 6 line have not yet appeared. Memory supply is the third open question, with HBM demand outpacing supply across the industry.
- Nvidia reports earnings on 17 November 2026, the first formal update since the board investigation.
- Applied Materials reports later in the week, giving a read on equipment demand for chip fabrication.
- Google is expected to confirm TPU 8tx and TPU 8i Max pricing, availability and early customers.
- Independent tests should show whether AMD’s agent-per-watt claims hold outside vendor benchmarks.
- Rebellions, Nvidia and SK Telecom move from agreement to hardware development ahead of the 2028 target.
Key Terms at a Glance
- TPU: A Tensor Processing Unit is Google’s custom chip for training and running AI models, offered through Google Cloud.
- NPU: A neural processing unit is a chip designed for AI inference work, such as running a trained model on new data.
- Confidential computing: A hardware-backed method that protects data and model weights while AI models process them.
- HBM: High-bandwidth memory is fast memory stacked close to a chip to feed AI workloads with data.
- NVLink Fusion: Nvidia interconnect technology that lets non-Nvidia chips join Nvidia rack-scale systems.
Frequently Asked Questions
What is a TPU?
A TPU is Google’s custom Tensor Processing Unit for training and running AI models. Google offers TPUs through Google Cloud as an alternative to Nvidia GPUs for AI workloads.
What is the difference between TPU 8tx and TPU 8i Max?
TPU 8tx is built for secure, confidential inference on sensitive data, with AMD EPYC processors as hosts. TPU 8i Max pairs two TPU 8i chips in one package to double compute and memory bandwidth for large-context workloads.
When will the R200 systems from the Korea deal be ready?
R200 systems are expected to be ready for deployment in 2028, according to the companies. The announcement covers development and deployment plans, not shipping hardware.
What is the AMD EPYC 6 processor line?
EPYC 6 is AMD’s sixth-generation server processor line, codenamed Venice and built on the Zen 6 architecture. The AI chip news from AMD puts the line at up to 256 cores per socket, aimed at enterprise, cloud and agent workloads.
Why did Nvidia shares fall on 10 November 2026?
Nvidia shares fell after its board announced an independent investigation on 8 November 2026. The wider chip index rose the same day, so the fall was specific to Nvidia rather than the sector.
Related Reading on Future AI Wire
- Agentic AI vs Generative AI: Key Differences Explained
- AI Stock Sell-Off November 2026
- OpenAI Dots, Grok Bot and Compute Costs
Sources
- SiliconANGLE — Google advances agentic AI custom silicon expanding TPU family with 8tx and 8i Max (9 Nov 2026)
- eWeek — Google’s New AI Chips Take On the Memory Wall (10 Nov 2026)
- AMD via GlobeNewswire / Morningstar — AMD Showcases 6th Gen AMD EPYC Processors (10 Nov 2026)
- Nvidia, Rebellions and SK Telecom via GlobeNewswire / Morningstar — Collaboration to Advance Sovereign AI Infrastructure (9 Nov 2026)
- Morningstar / PR Newswire — Quarterly Reports and an Independent Board Inquiry Frame the Chip Sector’s Week (9 Nov 2026)
The Future AI Wire Editorial Team covers artificial intelligence, AI agents, agentic AI, and emerging technologies. We publish clear, research-informed explanations and practical insights to help readers understand AI developments, real-world applications, and the technologies shaping the future.