AI: Weekly Summary. RTZ #722

1. Trump Mideast Trip with Tech Cos: The Trump Administration kicked off its first overseas trade mission to the Mideast, accompanied by a host of US companies and CEOs. Many from the US tech/AI industry like OpenAI, Nvidia, Amazon, Microsoft, AMD, Tesla and others. The trip through Saudi Arabia, UAE, and Qatar resulted in hundreds of billions committed to US AI data center and AI GPU chip purchases, both for deployment in the region and in the US. Nvidia in particular got approval to sell hundreds of thousands of its latest Blackwell AI GPUs in the region, sidestepping the Biden AI Diffusion Rules limits for Tier 2 countries. The visit also resulted in US contracts for defense and drone sales in the region. My video take on the trip here. More here.

   

2. Anthropic’s Claude AI Reasoning innovations: Anthropic continues the AI Reasoning race with its next iterations of its Claude LLM AIs, keeping apace with OpenAI, Google, and others like DeepSeek and Manus from China. The new versions of Claude Sonnet and Opus, expected in a few weeks, are different in their ability to go back and forth between thinking, or exploring different ways to solve a problem and tool use externally”. The focus on AI coding is also present here, a white hot area for LLM AIs, with OpenAI’s recent $3 billion purchase of Windsurf, and Cursor’s market leadership with its latest $9 billion valuation. Fusing AI Reasoning and Agents with ‘thinking’ remains the industry holy grail, with AI coding as the core, current use case. More here.

   

3. Meta’s Behemoth Llama difficulties: Latest reports highlight Meta’s difficulties with its largest open source Llama model ‘Behemoth’, and the likely delays in release schedules to later this year or beyond. The delays come at a time when the company is increasing its AI infrastructure spend to over $64 billion, and its plans to roll out Meta AI via its platforms and for its advertisers. And while competitors like OpenAI, Anthropic, Google and others are racing ahead. Google is likely to have more details on Gemini AI models at Google I/O in a few days. The episode highlights the near term ups and downs in developing state of the art LLM AI models, something we’ve seen at OpenAI, Google and other companies. More here.

   

4. Apple’s Trump India/China Dilemma: Apple saw an unexpected turn in its plans to make more US bound iPhones in India, to satisfy previous statements by the administration. President Trump from his mideast trip in Qatar is now asking Apple to stop moving iPhone production to India. He made a reference to a specific conversation with Apple CEO Tim Cook. This puts a monkey wrench in Apple’s plans since the recent tariffs to move more of the US bound iphones to be produced in India. Consensus continues on the impracticalities of Apple making its iPhones in the US at a reasonable cost and price for end consumers. And Apple’s massive investment and entrenchment that is still in China. This also has implications for the administration’s ongoing trade talks with India, as well as the current renewed negotiations with China. It’s unclear if these are bargaining and negotiating moves or something else. More here.

   

5. Wikipedia points to AI content data pros & cons: Wikipedia content and data has been a mainstay of today’s LLM AIs, and a recent report illustrates how that is going both positively and negatively. A core issue of this AI Tech Wave continues to be the use of internet content to train LLM AI models, and use the same on ‘Fair Use’ grounds as well for AI inference to drive better results for billions of users. So far, the LLM AI companies have been able to work with Wikipedia and have in many ways strengthened the organization and its prospects. The case study offers lessons for the content industry to navigate the difficult shifts in AI tech and politics, particularly in the charged environment of regulatory initiatives at the federal and state levels. Lots of lessons here for all stakeholders for AI’s incessant appetite for more data. More here.

   

Other AI Readings for weekend:

1. Google AlphaEvolve a step forward in AI Coding and other advances. More here.

   

2. AI Agent Security Issues just beginning. More here.

   

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)