How are Chinese companies accessing Nvidia chips despite US export restrictions?
Cover Photo Major News from Nvidia, Apple, Google's Prompt Gallery, xAI's Grok-2 and Raidio.FYI

Chinese AI Developers Circumvent U.S. Chip Export Restrictions

Chinese AI engineers are bypassing U.S. export restrictions on advanced Nvidia chips by accessing computing power overseas through brokers and decentralized platforms. These platforms, often utilizing blockchain technology for anonymity, allow Chinese companies to rent Nvidia’s powerful GPUs without importing them directly. Entrepreneurs like Derek Aw facilitate this by setting up data centers abroad, such as in Australia, to meet the high demand. Despite U.S. sanctions, the use of decentralized GPU services continues to rise, with some companies building larger clusters to support intensive AI training.

Apple’s Ambitious AI Plans: A New ‘Personality’ for Robotics

Apple is reportedly developing a new AI ‘personality’ using generative AI to enhance its future robotic devices, potentially replacing Siri. According to Bloomberg’s Mark Gurman, this AI could be integrated into a tabletop robot, similar to Amazon’s Astro, with a target price below $1,000. While these robotics projects are in early development stages, Apple is also expanding its generative AI capabilities across its devices. Initially, these AI features will be powered by ChatGPT, indicating Apple’s ongoing efforts to advance its AI technologies.

Google’s AI Studio Enhances Developer Experience with New Prompt Gallery

Google has launched a new Prompt Gallery in its AI Studio platform, enhancing the tools available to developers using the Gemini API. Announced by Logan Kilpatrick on X.com, this feature offers a variety of pre-built prompts, showcasing the Gemini models’ capabilities across technical and creative applications, from math tutoring to trip recommendations. The free Prompt Gallery aims to democratize AI access, lowering entry barriers and promoting innovation. This strategic move by Google could significantly impact the AI tools market, challenging competitors and accelerating AI adoption.

xAI’s Grok-2 Models Gain Speed and Accuracy with Code Overhaul

Elon Musk’s xAI has enhanced its Grok-2 and Grok-2 mini large language models by rewriting their inference code using SGLang, significantly boosting their speed and accuracy. This update, executed by developers Lianmin Zheng and Saeed Maleki, positions Grok-2 as a leading AI model, ranking second on the Lmsys Chatbot Arena leaderboard. Grok-2 excels in mathematical tasks and maintains strong performance across various categories. xAI’s commitment to innovation promises further advancements in AI speed and efficiency.

Will.i.am Launches AI Radio App, Raidio.FYI, with Interactive Digital Personas

Rapper will.i.am has introduced Raidio.FYI, an AI-powered app that simulates a radio experience with digital personas and curated content. The free platform allows users to engage in real-time, interactive conversations on various topics, offering a creative co-pilot to refine ideas and strategies. Raidio.FYI features AI co-hosts, curated playlists, and project management tools, all secured with advanced encryption. Will.i.am encourages users to actively engage with the AI, promoting thoughtful interaction while ensuring factual reporting.

Frequently asked questions

Chinese companies are accessing Nvidia’s advanced chips through overseas brokers and decentralized platforms. They rent GPU computing power from data centers located in countries like Australia, using blockchain technology for anonymity. This approach allows them to utilize Nvidia’s powerful GPUs for AI development without directly importing restricted hardware, effectively circumventing US sanctions while meeting their computing needs.
Google’s Prompt Gallery is a new feature within the AI Studio platform that provides developers with pre-built prompts for the Gemini API. It offers a variety of ready-to-use prompts across technical and creative applications, from math tutoring to travel recommendations. This tool helps developers better understand and utilize Gemini models’ capabilities while lowering entry barriers for AI development.
xAI’s Grok-2 and Grok-2 mini models have been enhanced through a complete code overhaul using SGLang, resulting in significant speed and accuracy improvements. The update has elevated Grok-2 to second place on the Lmsys Chatbot Arena leaderboard, with particular excellence in mathematical tasks and strong overall performance across various categories.
Apple is developing a new AI ‘personality’ using generative AI technology to enhance future robotic devices, potentially replacing Siri. The company plans to integrate this AI into a tabletop robot priced under $1,000, similar to Amazon’s Astro. Initially, these AI features will be powered by ChatGPT while Apple continues to develop its own AI technologies.
Raidio.FYI is an AI-powered radio app launched by will.i.am that creates an interactive radio experience using digital personas. The platform enables real-time conversations on various topics, features AI co-hosts, curated playlists, and project management tools. It’s designed to provide users with a creative co-pilot while maintaining security through advanced encryption.
Nvidia chips are crucial for AI development, providing the powerful GPU computing capabilities needed for training and running advanced AI models. Companies worldwide are seeking access to these chips through various means, including cloud services and data centers, due to their superior performance in handling intensive AI workloads.
Google’s Prompt Gallery stands out by offering a free, comprehensive collection of pre-built prompts specifically designed for the Gemini API. It democratizes AI access by providing practical examples and use cases, making it easier for developers of all skill levels to implement AI solutions. The gallery’s integration with Google’s AI Studio platform creates a seamless development experience.
Picture of Gor Gasparyan

Gor Gasparyan

Optimizing digital experiences for growth-stage & enterprise brands through research-driven design, automation, and AI