AI: Weekly Summary. RTZ #360

1. OpenAI’s GPT-4o ‘Omni’ rollout: OpenAI started and continued their expected fast-paced volley of product and feature rollouts, kicking off the week with GPT-4o or ‘Omni. Highlight was a multimodal Voice powered ChatGPT, matching Google’s Astra rolled out the next day. The demos were quite impressive, including third party demos by Khan Academy’s Sal Khan using GPT-4o to tutor his son in Math. Later in the week OpenAI rolled out features enabling ChatGPT integration with Google Drive and Microsoft OneDrive, increasing office AI functionality for millions. More here.

2. Google I/O 2024 applies AI across ecosystem: Google also had a well-laid rollout of AI-powered announcements at I/O 2024, highlighting ‘AI’ over 120 times. CEO Sundar Pichai neatly closed his keynote with that observation with Google Gemini running a live count of his ‘AI’ mentions. Highlights included an upcoming integration of Google multimodal Gemini AI into Google Search in the US, and across many of its flagship properties. ‘AI’ count expectations are now on for Apple CEO Tim Cook at WWDC on June 10. More here.

3. Voice a key AI UI innovation ahead: Both OpenAI and Google are leading with multimodal voice user interface and user experience (UI/UX) elements across their services. This is also a focus by others like Meta with Meta AI, and a spate of startups including ElevenLabs and others. With Apple’s expected integration of OpenAI GPT into its Siri voice assistant at its Developer WWDC on June 10, expect billions of mainstream users experiencing Generative AI in full force going into the second half of this year. More on this trend here and here. 

4. AI Industry transition to AI Inference chips: The next rush for AI chips is on for lower-cost and more power efficient AI inference chips, both in cloud data centers, and more ‘on-prem’ local and ‘edge’ applications on PCs and devices. Nvidia is very much focused on this Inference trend. Apple’s initiatives on ACDC, a data center optimized chip with TSMC is a case in point. As billions of mainstream users start to on-ramp on multimodal and voice driven LLM AI applications, this transfer of AI processing loads onto Inference optimized chips may mitigate the current unrelenting pressure on bigger AI data center and power infrastructure demands. Both for ‘Big and Small AI’. More on trends here and here.

5. Elon Musk’s xAI nears $10 billion Oracle Data Center deal: News of xAI nearing a multi-billion dollar deal for Oracle Cloud data center servers, continues the ongoing trend of big tech investing tens of billions on AI data center and power Nvidia ‘accelerated computing’ infrastructure in 2024. The deal would make xAI one of Oracle Cloud’s largest customers, and points to a key use of proceeds for its upcoming $6 billion funding round. 

Other AI Readings for weekend:

1. OpenAI’s Reddit partnership continues trend in content LLM AI data deals

   

2. Transition for OpenAI’s Superalignment team with founder Ilya Sustkever’s departure

Thanks for joining this Saturday with your beverage of choice. 

Up next, the Sunday ‘The Bigger Picture’ tomorrow. Stay tuned. 

(NOTE: The discussions here are for information purposes only, and not meant as investment advice at any time. Thanks for joining us here)