ConnectingAI #100: The Future of AI Processing and more

Top ConnectingAI articles of the week 

1- The Future of AI Processing: AI computing is shifting toward specialized hardware and new architectures, with companies racing to overcome bottlenecks in data movement and energy consumption. Innovations like photonics and neuromorphic chips could dramatically speed up AI performance while reducing costs.

2- Shining a Light on the Future of AI and Optical Networking: Broadcom highlighted new breakthroughs at OFC 2025, showcasing optical networking technologies critical to scaling AI workloads efficiently. The innovations promise faster, more energy-efficient data movement across massive AI infrastructures.

3- Lumen Building the Trusted Network for AI: Lumen is investing heavily in building a secure, low-latency network designed specifically for AI workloads, combining edge computing, private networks, and cybersecurity. Their goal is to become the backbone of trusted AI-powered business operations.

4- Penn Engineers First to Train AI at Light Speed: Penn engineers have developed a new system that uses optical hardware to train AI models at the speed of light, dramatically improving speed and energy efficiency. This breakthrough could reshape the future of machine learning and AI hardware.

5- AI Drives Re-Engineering of Nearly Everything in Chips: The demands of AI are forcing a complete rethink of chip design, from materials to architectures, to handle exponential growth in model complexity and data volumes. Companies are embracing domain-specific architectures, 3D integration, and new interconnect technologies.

6- Lam Research Shines with Strong AI Chip Demand: Lam Research is seeing a surge in demand for its chipmaking equipment, fueled by the booming AI sector's need for advanced semiconductor manufacturing. The trend highlights how the AI revolution is reshaping not just tech giants, but also their critical supply chains.

7- Cerence AI Revolutionizing Automotive Interaction With New Partnership: Demand for AI chips continues to surge, benefiting AI infrastructure players. Investors are closely watching firms that enable AI applications across industries, from automotive to enterprise software.

AI piqued my interest

1- Machine-Learning "Periodic Table" Could Fuel AI Discovery: MIT researchers developed a "periodic table" for machine learning, organizing AI models based on their traits to make it easier to innovate and combine methods. This framework could accelerate the discovery of new, more powerful AI systems.

2- AI and Interior Design: How Technology is Shaping Home Spaces: AI is increasingly shaping the world of interior design, offering tools that generate personalized room layouts, color schemes, and furnishings. These AI-driven platforms are helping consumers and designers move faster from vision to reality.

3- Want to Make Music Like Imogen Heap? JEN, Her New AI Venture, Can Help: Musician Imogen Heap launched JEN, an AI-powered tool that helps anyone create professional-grade music compositions. The platform blends Heap's artistry with cutting-edge AI to democratize music production.

4- Anthropic CEO Wants to Open the Black Box of AI Models by 2027: Anthropic’s CEO Dario Amodei announced plans to make AI models more interpretable and transparent within two years. The initiative aims to build safer, more understandable AI systems amid growing concerns about AI’s unpredictability.

5- Elon Musk’s xAI Releases Grok-3 to Intensify AI Competition: Elon Musk’s xAI launched Grok-3, a new AI model designed to compete directly with ChatGPT and other leading chatbots. Grok-3 promises broader knowledge, sharper reasoning, and a stronger integration with X (formerly Twitter).

6- Liquid AI is Revolutionizing LLMs for Edge Devices with Hyena Edge: Liquid AI introduced the Hyena Edge model, designed to bring powerful large LLM capabilities to smartphones and other edge devices. This advance could dramatically reduce AI's dependence on cloud computing.

7- Everything to Know About ChatGPT in 2025: ChatGPT has evolved significantly, with new features like memory, more personalization, and multimodal capabilities blending text, image, and audio interactions. OpenAI continues expanding ChatGPT’s role as an all-purpose AI assistant for work, learning, and creativity.

More on ConnectingAI, Connecting Everything next week. 

Note: A version of this newsletter is published at ConnectingAI.substack.com. I have made it easier for LinkedIn readers to access the newsletter now. You can subscribe to this newsletter if it is easier.