AI: Weekly Summary. RTZ #1025
-
Microsoft Copilot brands & bundles Anthropic CoWork: Microsoft is leveraging its new Anthropic investment by bundling its virally successful Claude Code based Cowork into ‘Copilot Cowork’. It’ll now be sold as a super $99/month/seat bundle to users of its 450+ million strong global user base of Office 365, itself a subscription product now. It’s a super-bundle because it’s not just Copilot Office, but Copilot Cowork. Which means it’s designed to allow users to ‘safely’ deploy Claude Cowork’s ability to do local tasks on a user computer, in Microsoft’s Azure Cloud. Complete with guardrails around the enterprise customer’s data and corporate IP privacy ‘guardrailed’ and secured. It’s a direction that is being emulated by OpenAI, Perplexity, Google and others. And a raise to charge premium prices for AI Inference Compute for intelligence tokens, that is still a daunting variable cost for users. Especially as they get to do more AI Reasoning and AI Agents to do truly useful things sooner than later. The Compute for all that while coming down in price, is still likely to balloon ahead of the organization’s ability to recoup its AI investments, and remains a challenge for Microsoft and its peers. Here’s More.
-
Market assesses OpenAI’s year-end IPO: OpenAI remains the seminal AI IPO in Q42026, being anticipated by Wall Street investors small and large. Along with of course SpaceX/xAI by Elon Musk, and Anthropic, OpenAI’s ‘Pepsi’ to its ‘Coke’. The first two are expected to trade above a trillion dollars in market cap each if launched successfully. With Anthropic not too far behind. The street is already handicapping OpenAI’s prospects given its global pole position with the ChatGPT brand name and over 900 million monthly users. And its fleet of AI Data Centers being ramped up now beyond just Microsoft Azure, to Softbank, Amazon, Google, and other clouds. Not to mention its own multi gigawatt data centers being built with the help of Oracle, Broadcom, CoreWeave and others. In addition, OpenAI has a ramping portfolio of AI Applications incorporating AI Reasoning and AI Agents, not least of which is its recent open source OpenClaw acquisition. Again all that is the ramping costs of all this against the lagging revenues and presumed profits. Not to mention the Elon Musk lawsuit against OpenAI kicking off next month. That’s the core balancing act that investors will have to assess six months from now. All again of course in a global macro context of trade and tariff curbs, supply chain dislocations, and of course War in the middle east. Here’s More.
-
Meta leans into AI Agents and considers Google Gemini: Meta is considering using Google Gemini for its Meta AI efforts, while continuing to invest in its own LLM AI models. A big possible distribution coup for Google Gemini following its similar deal with Apple/Siri recently. Also, Meta remains in a huge hurry on all things AI, on everything from next gen LLM AI models beyond open source Llama to more closed ‘Avocado’, ‘Mango’ and eventually ‘Watermelon’, being developed by its recently built Meta Superintelligence Labs (MSL). And of course its $135+ billion AI Data Center compute spend this year alone, ramping to higher amounts the year ahead. And now it’s also in a hurry to deploy AI Agents across its core properties touching 3.5+ billion global daily users. Its recent acquisition of formerly Chinese AI Agents company Manus, is being supplemented with the acquisition of ‘Moltbook’, a recently launched ‘social network’ for AI Agents. All based on open source OpenClaw’s platform, recently acquired by OpenAI, where Meta founder/CEO Mark Zuckerberg was unsuccessful in bringing on OpenClaw founder Peter Steinberger. Meta doesn’t plan to keep Moltbook in its current form. But its two founders are being added to the MSL team, to aggressively develop their ideas around Meta’s core properties and advertiser platforms. Augmenting them with AI Agent innovations to come. All of this continues to highlight Mark Zuckerberg’s relentless drive not to miss any innovation rock underway in the AI Tech Wave to date.. Here’s More.
-
AI Agents need their own Internet: The above explosion of AI Agents, be it from Meta, OpenAI, Anthropic and others, is increasingly highlighting the need for a parallel set of Internet Infrastructure designed explicitly to accelerate AI Agents to work better and faster for their billions of users to come. Agents in the trillions soon interact with each other on behalf of their billions of users. And doing it all with security and enhanced privacy for the underlying users and their data. While adhering to global regulations around ‘machine 2 machine’ agentic interactions with guardrails and safety. This while there is a rush of newer AI ‘World Models’ that increasingly go beyond traditional LLM AI and their data sources today on the internet. Into the physical world with their messy systems of self driving cars, robots, drones, digital twin installations and far more. All the major tech companies are focused on these efforts, even while making sure today’s AI Agents can just use the internet as it’s designed for humans. AI Bot protection systems like Captchas and their more advanced iterations guarding the current internet as ‘for humans’ only. Here’s More.
-
New War Risks for US/Mideast AI Data Centers: The hard fought race just a few months ago by leading US tech companies to lay out multi-billion dollar plans for AI Data Centers in the middle east, with the help of the Trump administration, is running into headwinds with the current US/Israel War with Iran. Amazon saw its AWS data centers attacked by Iranian forces using $20,000 ‘Shaheed’ drones and missiles several times in the less than two weeks since the war was initiated by the US and Israel. The attacks highlight a new vulnerability to these investments that was mostly theoretical a few days ago. And its an issue that could have an overarching impact on the viability of these investments, even if hostilities ceased in the short term. The vulnerabilities have been highlighted as being not too different from the ones feared in the theoretical context of a future China/Taiwan altercation. And the possible impact on AI Data Center Compute infrastructure, being built out in the hundreds of billions and multi-gigawatts of power around the world. With the big showcase installations having already started in the Middle East. Here’s More.
Other AI Readings for weekend:
-
Elon’s SpaceX/xAI now ramps up Grok Coding team while pruning teams and rebuilding xAI. More here.
(Additional Note: For more weekend AI listening, here’s the latest AI Ramblings Episode 45 on topical items. This week, Apple’s new ‘Value’ Gauntlet & More):
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)
