intel, nvidia and openai Updates

The artificial intelligence sector continues to evolve rapidly, influencing market dynamics, product development, and regulatory discussions. Micron recently helped ease pressure on the broader AI trade market, while Robinhood's CIO, Stephanie Guild, anticipates strong earnings growth not only outside of technology but also within tech companies demonstrating clear utility. This comes as Intel navigates significant leadership changes in government affairs, marketing, and advanced technology strategy, amidst U.S. lawmaker scrutiny over its dealings with China-linked chipmaking equipment and rumors of a potential acquisition of AI startup SambaNova. In the competitive chip landscape, a report from the Council on Foreign Relations highlights that Huawei's AI chip capabilities, specifically its Ascend 910C, still lag significantly behind Nvidia's H100, delivering only about 60% of its performance. This gap stems from Huawei's inability to access advanced manufacturing processes and packaging from TSMC, which Nvidia utilizes. Experts estimate Huawei's next-generation chips might only reach H100-level performance by 2026 or 2027, by which time Nvidia is expected to be even further ahead, maintaining its lead in both performance and production volume. AI's impact extends into commerce and content creation. Klarna introduced its Agentic Product Protocol, an open standard designed to help AI agents efficiently find merchant products, compare prices, and track availability across various stores, leveraging Klarna's extensive database of over 100 million products. Meanwhile, generative AI presents a challenge to marketers by fostering a "just good enough" content economy. Disney's recent deal with OpenAI, licensing over 200 characters for systems like Sora and ChatGPT, exemplifies this shift towards producing average creative work at massive scale, potentially eroding the premium quality brands typically strive for. The integration of AI into everyday devices and industrial systems also brings new developments and concerns. Indian firms Brandworks Technologies and SandLogic are partnering to develop full-stack Edge AI hardware, aiming to launch AI-enabled devices for mobility, home-tech, and education solutions in 2026, supporting India's self-reliance goals. On the consumer front, LG TV owners expressed frustration after a firmware update added an unremovable Microsoft Copilot shortcut to their smart TVs. LG has since confirmed it will allow users to delete the icon, while planning deeper Copilot integration into its 2025 OLED TVs. Security and ethical considerations remain paramount. CISA, the NSA, and other agencies released new guidance on securing AI in operational technology, identifying risks like prompt injection and data poisoning in industrial control systems and emphasizing human oversight. Concurrently, Adobe faces a class action lawsuit from author Elizabeth Lyon, who alleges the company misused copyrighted books from the Books3 collection to train its SlimLM language models without permission or payment, a claim Adobe denies.

Key Takeaways

• Micron's market performance helped ease pressure on the AI trade sector.
• Intel is undergoing leadership changes and facing U.S. scrutiny over China dealings, with rumors of potentially acquiring AI startup SambaNova.
• Huawei's Ascend 910C AI chip delivers approximately 60% of Nvidia's H100 performance, with a significant gap expected to persist until at least 2026-2027 due to manufacturing limitations.
• Klarna launched its Agentic Product Protocol, an open standard enabling AI agents to find and compare over 100 million merchant products across different stores.
• Disney licensed over 200 characters to OpenAI's Sora and ChatGPT, indicating a shift towards high-volume, "just good enough" content production that lowers costs.
• Indian firms Brandworks Technologies and SandLogic are collaborating to produce full-stack Edge AI hardware, with initial products for mobility, home-tech, and education launching in 2026.
• LG TV owners reacted negatively to an unremovable Microsoft Copilot shortcut added via a firmware update, though LG plans to allow its deletion and deeper integration in 2025 OLED TVs.
• CISA and the NSA released new guidance on securing AI in operational technology, highlighting risks like prompt injection and the necessity of human oversight.
• Adobe faces a class action lawsuit alleging it used nearly 200,000 copyrighted books from the Books3 collection to train its SlimLM AI models without permission.

Micron eases AI trade market pressure

Michael Santoli, a senior markets commentator for CNBC, discussed his main points from the day's market activity. He noted that Micron helped reduce some of the pressure on the AI trade. This insight came during his analysis of market trends.

Robinhood CIO sees strong earnings outside tech

Stephanie Guild, Chief Investment Officer at Robinhood Markets, believes earnings growth could be stronger than expected in sectors outside of technology. She also thinks certain tech companies with clear usefulness may perform well. Guild shared these insights with Romaine Bostick and Katie Greifeld on "The Close" on December 19, 2025. This discussion highlighted her views on the current market landscape.

Indian firms partner for Edge AI hardware

Brandworks Technologies and SandLogic are partnering to create full-stack Edge AI hardware in India. This collaboration aims to develop AI-enabled devices for voice assistants, intelligent IoT systems, and mobility applications. These products will process information directly on the device, reducing reliance on cloud services and improving speed and data control. The first co-developed AI-enabled hardware, focusing on mobility, home-tech, and education solutions, will launch in 2026. This strategic partnership supports India's goal of self-reliance in advanced electronics and plans to export these AI capabilities globally. Brandworks will also showcase its automotive electronics at CES 2026 in Las Vegas.

Klarna launches new AI product protocol

Klarna introduced its Agentic Product Protocol, an open standard that helps AI agents find merchant products. This protocol makes products accessible and allows agents to compare items across different stores, check prices, and track availability. It builds on Klarna's vast product data, including over 100 million products and 400 million prices. Merchants can join by sending product feeds or implementing the protocol themselves, needing a relationship with Klarna for API use. Experts like Aaron McPherson and Sam Boro see this as a step towards standardizing AI commerce and helping smaller merchants compete, though large retailers might limit access.

Intel shifts leaders amid AI and China scrutiny

Intel recently announced major leadership changes in government affairs, marketing, and advanced technology strategy. These changes come as U.S. lawmakers scrutinize Intel's dealings with China-linked chipmaking equipment. The company is also facing rumors about potentially acquiring the AI startup SambaNova. These developments suggest Intel is working to increase its influence in Washington and speed up its AI goals. The company aims to navigate complex challenges while strengthening its position in the tech industry.

Generative AI threatens marketing with "just good enough" content

Generative AI poses a significant threat to marketers by creating a "just good enough" economy. Disney's recent deal with OpenAI highlights this shift, as Disney licensed over 200 characters to OpenAI's systems like Sora and ChatGPT. This partnership aims to produce average creative work at a massive scale, not necessarily better creative. James Kirkham, co-founder of Iconic, notes that this normalizes high-volume, "just good enough" content, which could erode the premium quality where brands typically compete. The real change is economic, as generative AI dramatically lowers the cost of producing content.

Agencies release AI security guide for industrial systems

CISA, the NSA, and other partners released new guidance on securing artificial intelligence in operational technology. They state that integrating AI into industrial control systems, like SCADA, brings new security risks requiring special protections. The guidance identifies AI threats such as prompt injection, data poisoning, and model drift, along with concerns about data sovereignty. It outlines four key principles for managing these risks, including understanding AI risks and building safety mechanisms. The agencies emphasize the need for human oversight and backup plans when AI is used in systems affecting physical operations.

Report says Huawei AI chips lag far behind Nvidia

A new report from the Council on Foreign Relations finds that Huawei's AI chip capabilities are still far behind Nvidia's. Despite U.S. export controls, Huawei's flagship Ascend 910C chip only delivers about 60% of the performance of Nvidia's H100. Nvidia uses advanced manufacturing processes and packaging from TSMC, which Huawei cannot access, leading to slower memory configurations. The report estimates Huawei's next-gen chips might only reach H100-level performance by 2026 or 2027, by which time Nvidia will be much further ahead. Huawei also faces significant limits in production volume, shipping only a fraction of the millions of chips Nvidia produces annually.

Adobe sued for using copyrighted books in AI training

Oregon author Elizabeth Lyon filed a class action lawsuit against Adobe, claiming the company misused authors' work to train its AI models. The lawsuit states that Adobe's SlimLM language models, used for document assistance, were trained on a dataset called SlimPajama-627B. Lyon alleges that SlimPajama is based on RedPajama, which supposedly includes Books3, a collection of nearly 200,000 pirated books. She argues that Adobe used copyrighted material without permission, credit, or payment. Lyon seeks damages and a declaration of willful infringement from Adobe, which has denied the allegations.

LG TVs get Copilot shortcut sparking user anger

LG TV owners were upset after a firmware update added an unremovable Microsoft Copilot shortcut to their smart TVs. LG confirmed the webOS update installed a shortcut to the Copilot web app, not the full application, to improve customer convenience. Following user complaints, LG announced it will allow users to delete the shortcut icon soon. LG plans to integrate Copilot deeper into its webOS for 2025 OLED TVs, similar to Samsung, which has included Copilot on its TVs since August. This trend raises concerns about privacy, data collection, and unwanted software on smart TVs.

Sources

• Micron relieves some pressure on the AI trade: Michael Santoli
• Robinhood’s Guild: AI Trade Under Pressure
• Brandworks Technologies and SandLogic Partner to Develop Full-Stack Edge AI Hardware in India
• Inside Klarna's agentic AI protocol
• Are Intel's (INTC) Leadership Shifts And AI Rumors Quietly Rewriting Its Washington Strategy?
• Future of Marketing Briefing: Why 'just good enough' is generative AI's real threat to marketers
• CISA, NSA and partners publish AI security principles for operational technology
• Huawei's AI chip capabilities still pale in comparison to American silicon — report from U.S. council details that despite fears, Nvidia continues to lead by a wide margin
• Adobe faces class action lawsuit after allegedly misusing authors work in AI training
• LG TVs’ unremovable Copilot shortcut is the least of smart TVs’ AI problems