AI: Weekly Summary. RTZ #836

Michael Parekh AI: Reset to Zero

7 days ago

4 MIN READ

Google dodges ‘Remedies’ bullet: Google got some good news in the ‘Remedies’ phase of its antitrust trial targeting its Search and Chrome businesses. Both those businesses were relatively untouched, with a requirement to share some Search data with rivals. The judgment likely allows Google to continue its annual $20+ billion Search distribution deal with Apple, and potentially expand it with Gemini for Apple Siri and beyond. Also takes off the table, Google needing to sell its global market leading Chrome browser to OpenAi, Perplexity or others. It’s also notable that Judge Mehta cited AI as a mitigating factor for increasing competition in the Search space since the trial started a few years ago. As CEO Sundar Pichai said at a White House dinner after the decision, “I’m glad it’s over”. More here.

OpenAI Organizes AI Applications: OpenAI’s new CEO of Applications from Instacart, Fidji Simo is busy organizing the App organization at the company. Moves include a new ‘B2B Applications CTO from within, as well as the appointment of a new CTO of Applications from its $1.1 billion acquisition of Statsig, an AI product testing AI startup. In addition, OpenAI is moving Chief Product Officer Kevin Weil, formerly from Meta, to the research side as its new VP of AI for Science. The AI product management responsibilities remain under Fidji Simo. Overall, the moves ramp up OpenAI’s robust applications product pipeline going into the Fall and beyond. Next up possibly, an ‘AI Browser’ as Chrome is off the table, as that space heats up (see item 2 below). More here.

AI Industry Grapples with Variable AI Compute Costs & Pricing: While the leading LLM AI companies innovate rapidly and make their core products more capable and price efficient, the overall costs for customers remain relatively expensive. Part of this is related of course to the variable costs of rising AI inference costs that scale up with AI, as users deploy more AI capabilities in their end products and services. This remains a key characteristic of the AI business both on the enterprise and consumer sides. The leading companies are all testing a range of tiered pricing models, with subscriptions continuing to edge up from $20/month to $200/month and far higher. The next couple of years will continue to see this experimentation, as end customers find the right product-market-fit for AI applications and services. More here.

China AI focuses on Applications: Recent developments from China indicate a marked focus on AI Applications vs the race for AI AGI/AI Superintelligence that drives the agenda for US LLM AI companies. While the Xi government continues to help its tech/AI industry ramp up the AI Infrastructure, and Power requirements, the private sector there continues to focus on AI applications and services at the upper levels of the AI tech stack. And as noted, a majority of the AI industry in China is focused on open source LLM AIs post the initiatives by DeepSeek and others in that market. These developments mark a differentiated strategy in China vs the US, amidst ongoing geopolitical trade and tariff tussles. More here.

Google leads with TPUs with AI Cloud partners: Google is expanding its efforts to compete against Nvidia AI GPU infrastructure with its TPU chips. These have until now been mostly deployed for Google’s own products and services. They’re now being deployed more proactively with AI Cloud partners like Amazon AWS, Microsoft Azure and others for end customer deployments. This marks a notable strategy shift in the ‘Frenemies’ environment for AI data center infrastructure. Google’s TPU chips for now are matching Nvidia’s roadmap with Blackwell to Rubin and beyond. Google Cloud is also being more proactive with TPU AI data center deployments, with notable recent deals with OpenAI, Apple and others. More here.