Google's TPUv7 Outperforms Nvidia's GB300 NVL72 in Agentic Workloads

Google's new TPUv7 chip outperformed Nvidia's GB300 NVL72 on SemiAnalysis's Agentic Inference Benchmark, delivering up to 1.67x better performance per cost. The test used real agentic workloads with long conversations and tool calls, with Google running four chips against Nvidia's 72-GPU rack. Nvidia still leads in some low-interactivity workloads according to the data.

Key Takeaways

  • <ul><li>Google's TPUv7 chip beats Nvidia's GB300 NVL72 on SemiAnalysis benchmark, delivering 1.67x better performance per cost</li><li>Jabil's AI revenue grew from $5.4 billion to $14.4 billion in fiscal 2026 and is expected to reach $22.1 billion in fiscal 2027</li><li>Google One AI Pro costs $19.99 per month but free trial users miss out on Gemini 4 Argon and YouTube Premium Lite</li><li>SpaceX Senior Manager Gabriela Castillo Madrid will testify before NYC Council on AI safety after skipping an October hearing</li><li>Mozilla's Trustworthy AI Map features 79 companies across 15 countries building AI products with trust baked in</li><li>Two-thirds of estate agents use AI for property descriptions and marketing, but only 4% use it for sales progression</li><li>Moomoo's Global Paper Trading Competition offers $100,000 in virtual capital and $200,000 in trading coupons with new AI Companions</li><li>Gloucestershire College launches a 30-hour AI Strategy course for business leaders starting November 11, 2026</li><li>New Mexico State University partners with MVEDA to offer AI readiness training for small businesses in Southern New Mexico</li><li>VideoAmp reduced staff last summer and shifted operations to AI agents for live campaign optimization</li></ul>

Jabil stock could double by 2030 on AI hardware demand

Jabil makes server racks, cooling solutions, and power management systems for big tech companies. The company's earnings growth is accelerating, which could push its stock price higher over the next four years. A $10,000 investment in Jabil could be worth over $22,750 by the end of the decade if earnings grow by 15% in fiscal 2030.

Cerebras AI hardware stock could double by 2030

Cerebras is a leading developer of AI hardware that uses a massive 2.5D wafer-scale integration architecture to process large amounts of data in parallel. The company's chips are designed to be scalable and flexible for many applications. Cerebras focuses on partnerships to grow its business. The Motley Fool predicts a $10,000 investment in Cerebras could at least double by 2030.

AI hardware stock prediction page not found

This article link returned a page not found error. The original prediction article discussed a $10,000 investment in an AI hardware stock that could double by 2030, hinting it was not Nvidia or Micron. The content for this specific URL was unavailable.

Jabil stock could double as AI server demand grows

Jabil reported $36 billion in revenue for fiscal 2026, up 21% from the prior year. The company makes AI server racks, cooling systems, and power management solutions for Meta, Amazon, and Google. Jabil's AI revenue grew from $5.4 billion to $14.4 billion in fiscal 2026 and is expected to reach $22.1 billion in fiscal 2027. The company expects revenue to hit $44.5 billion this fiscal year.

AI hardware stock prediction page not found

This article link returned a page not found error. The original prediction article discussed a $10,000 investment in an AI hardware stock that could double by 2030, hinting it was not Nvidia or Micron. The content for this specific URL was unavailable.

Gloucestershire College launches AI training for businesses

Gloucestershire College is offering a new 30-hour AI Strategy & Opportunity course for business leaders. The training uses government apprenticeship funding for employees aged 19 and over. It starts November 11, 2026 at the Cheltenham campus. The course helps decision-makers plan how to use AI in their organizations. It is led by AI strategy specialist Lorrin White from Time For Strategy.

NMSU and MVEDA launch AI course for small businesses

New Mexico State University Global Campus partnered with the Mesilla Valley Economic Development Alliance to create an AI readiness course. The course helps small businesses in Southern New Mexico understand and use AI tools. It covers topics like customer communication, research and data analysis. The course also addresses limitations such as accuracy, privacy and security concerns. The partnership aims to strengthen the local business community through education.

Moomoo launches third global paper trading competition with AI

Moomoo is holding its third Global Paper Trading Competition from October 12 to November 20, 2026. The contest partners with Nasdaq TotalView to give participants professional market data. This year introduces AI Companions, offering four types of assistants for different trading strategies. Participants receive US$100,000 in virtual capital to trade stocks, ETFs and options. Competitors can win a share of US$200,000 in trading coupons.

VideoAmp uses AI to connect ad planning and measurement

VideoAmp is using AI to improve TV advertising planning and measurement. The company reduced staff last summer to shift more operations to AI. It also hired new leaders including a SVP of agency partnerships. VideoAmp uses AI agents to help buyers optimize campaigns during live runs. The company relies on ACR data and partnerships with media sellers to access viewer data for better ad targeting.

Google TPUv7 chip beats Nvidia on AI benchmark

Google's TPUv7 AI chip outperformed Nvidia's GB300 NVL72 on SemiAnalysis's Agentic Inference Benchmark. TPUv7 delivered up to 1.67x better performance per cost than Nvidia's top system. The test used real agentic workloads with long conversations and tool calls. Google ran the test using four chips against Nvidia's 72-GPU rack. Nvidia still leads in some low-interactivity workloads according to the data.

SpaceX agrees to testify on AI safety after NYC Council subpoena

SpaceX's Senior Manager Gabriela Castillo Madrid will appear before the New York City Council's Committee on Technology to discuss AI safety. The hearing follows a subpoena from Council Speaker Julie Menin after SpaceX skipped an Oct. 5 hearing. That earlier hearing included OpenAI, Anthropic, Meta, Google, and whistleblowers. Menin stressed that self-regulation is not enough and that companies must explain their safeguards.

Google limits free AI Pro trial access compared to paid plans

Google gives free trials of its AI Pro subscription with new Pixel phone purchases and other promotions. However, trial users get less access than paid subscribers. Trial users miss out on Gemini 4 Argon and YouTube Premium Lite until they start paying. Google One AI Pro costs $19.99 per month and includes 5TB of storage and other subscription bundles.

Carroll Schools adopts AI policy for students and staff

Carroll County Public Schools has implemented guidelines on AI use for students, families, and staff. The policy addresses data privacy and security concerns. Officials say it meets almost every requirement of a new state law. The law requires schools to have an AI policy in place by the end of the year. The full policy document has not been released.

Estate agents use AI for listings but not for closing sales

Two-thirds of estate agents use AI tools regularly, but mostly for simple tasks. About 69% use AI for property descriptions, emails, administration, or marketing. Only 4% use AI for sales progression, which manages transactions from offer to completion. CEO Riccardo Iannucci-Dawson of Alto said agents need to use AI in ways that help sell homes faster.

Mozilla launches Trustworthy AI Map to build public confidence

Mozilla and seven partner organizations created the Trustworthy AI Map featuring 79 companies across 15 countries. The map shows companies building AI products across seven layers of the AI stack, including identity, data control, and interoperability. To be featured, companies must bake trust into their product or business model. Mozilla Ventures and its partner Otari support this effort to help investors and businesses explore AI opportunities.

Sources

NOTE:

This news brief was generated using AI technology (including, but not limited to, Google Gemini API, Llama, Grok, and Mistral) from aggregated news articles, with minimal to no human editing/review. It is provided for informational purposes only and may contain inaccuracies or biases. This is not financial, investment, or professional advice. If you have any questions or concerns, please verify all information with the linked original articles in the Sources section below.

Google TPUv7 Nvidia GB300 NVL72 Performance per cost Agentic Inference Benchmark AI revenue growth Jabil Google One AI Pro AI safety SpaceX Mozilla Trustworthy AI Map AI in estate agency Moomoo Global Paper Trading Competition AI Strategy course AI readiness training AI agents in optimization

Comments

Loading...