AI: Weekly Summary. RTZ #779

1. xAI rolls out Grok 4: Elon Musk launched the latest major upgrade to xAI’s LLM AI model, Grok 4, in various sizes and capabilities. The model on first tests ranks well against comparable offerings from major peers including OpenAI, Google Gemini, Anthropic, Meta and others. Key features include multimodal capabilities and the ability to deploy ‘AI Agent Swarms’ in a collaborative fashion. The deep reasoning aspects of the model are also available in higher quantities with a $300/month subscription tier. Initial reviews, while generally positive, are also finding that Grok 4 tends to quote founder/CEO Elon Musk on controversial topics, as expressed on X/Twitter to his 220 million+ followers. And other mishaps. Nevertheless, xAI is planning its third funding round in two months, with a targeted valuation of almost two-thirds of OpenAI. More here.

   

2. Meta draws Apple ‘First Blood’: Meta founder/CEO Mark Zuckerberg continues his aggressive AI Talent recruitment strategy for Meta’s Superintelligence Labs, recruiting Apple’s top AI/ML research expert. Ruoming Pang headed up a 100 person strong team at Apple for many years, and it’s expected that some of his top folks may follow him to Meta. His compensation seems to be in Meta stock of $200 million plus over seven years, a figure that competes with the CEO compensation plans of most public companies. These sums seem comparable to what Meta has offered to create an ‘AI Dream Team’ from OpenAI, Anthropic and others. Expect the white hot AI Talent race to continue, both in the US and China. Of course, Meta continues to accelerate on its AI Data Center Infrastructure capex investments vs OpenAI, Meta, Google, Microsoft, Amazon and others. More here.

3. Apple Org Changes: Apple also saw its COO Jeff WIlliams, 62, an Apple veteran of 27 years stepping down. Only two years younger than CEO Tim Cook, the move marks a rare departure in Apple’s senior management ranks, and is not related to Apple’s perceived challenges on the AI front. The company has a deep bench of folks at the top, especially on the Supply chain side. Sabih Khan, a 30 year Apple veteran is stepping into that role, with the Apple Design team to now report to Tim Cook. That would be an arrangement similar to what prevailed under Steve Jobs over a decade ago. Of course the external debates over Apple’s AI strategy continue to fuel ‘Build vs Buy’ and external talent acquisition ruminations. More here.

   

4. Sakana AI shows AIs working better together: Two years after the two founders of Sakana Ai launched their LLM AI unicorn in Japan, they’re releasing an open source software framework that leverages multiple top LLM AI models to work better together. Founded by two of the eight original 2017 Google ‘AI Attention is Everything’ paper, Sakana, which means ‘fish’ in Japanese, has been closely watched by the AI industry globally. The performance metrics of the new software are impressive at launch, and seem to promise meaningful pathways for further improvements on their approach. This work points to additional ways Sakana AI may teach the world how to fish for better AI. More here.

5. Industry’s new AGI to Superintelligence Chase: Another impact of Meta founder/CEO Mark Zuckerberg’s latest ramped up retooling of Meta’s AI efforts under its new ‘Superintelligence Labs’, is an industry shift away from the oft-stated AGI (artificial general intelligence) goal. The new goal is Superintelligence, a term popularized last year with ex-OpenAI founder Ilya Sutskever’s Safe Superintelligence Unicorn, founded last year. As with AGI, the definitions and timelines for Superintelligence are also less defined, with a lot of top industry opinions on detailed descriptions, goals and timelines. But the term as with AGI, continues to be catnip for top and aspiring AI researchers worldwide from a recruitment incentive perspective, and that seems to be part of ‘secret sauce’ to draw them in. In addition of course to unprecedentedly generous sign-on and multi-year compensation. More here.

   

Other AI Readings for weekend:

1. China AI Unicorn, Moonshot AI releases globally competitive, updated open-source LLM AI models vs DeepSeek, Alibaba Qwen et al. More here on China’s AI position.

   

2. AI Data Centers ramp up on Debt. More here.

   

(Special Note: Doing a new podcast series on AI from a Gen Z and later perspective called AI Ramblings. Now 11 weekly Episodes and counting. More with the latest AI Ramblings Episode 11 here, on AI issues of the day, as well as our latest ‘Reads’ and ‘Obsessions’ of the Week. Co-hosted with my nephew Neal Makwana)

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)