AI: Weekly Summary. RTZ #801

Michael Parekh AI: Reset to Zero

4 days ago

5 MIN READ

Big Tech ‘AI Investment’ Acceptance: Earnings season for big tech has kicked off with a bang. Four of the Mag 7s have reported better than expected results, especially Microsoft and Meta. As well as Apple and Amazon a day later. The focus on the latter of course being on AWS. And investors mostly like what they see. And for now they seem fine with the ‘pedal to the metal’ stance on AI Infrastructure and Talent investments by most of them. Investors for now are leaning into the AI Investment opportunity story, even accepting unprecedented VERTICAL INTEGRATION of Infrastructure into traditional, high margin, software business models. Even for non Mag 7 AI companies like OpenAI, Anthropic and others. This despite their ongoing questions on when the investments will see returns at ‘AI Scale’. For now mid-year 2025, the music in the AI market musical chairs game is getting louder and speeding up. And the chairs for ultimate winners are diminishing rapidly. Exhibit 1 is Meta founder/CEO Mark Zuckerberg, who penned a piece highlighting Meta’s AI Superintelligence strategy in detail. And which will take hundreds of billions in AI Infrastructure investment. Not to mention years to deliver consumer/enterprise AI products and services at scale. More here.

OpenAI momentum accelerates ahead of GPT-5: OpenAI has its own strong AI momentum, just like its bigger tech peers and partners. The company apparently hit $12 billion in annualized revenue, while jumping from 500 to over 700 million ChatGPT Weekly Users. This is a clip of a billion dollars a month, vs half a billion in January 2025. The ramp represents both consumer and business customers, across both its ChatGPT subscription tiers, and APIs. The ‘bad news’ of course is the variable AI costs associated with the ramp, projected to be roughly $8 billion this year. That compares to a billion dollars projected earlier this year. Expect more of the same as OpenAI ramps GPT-5 next month. More here.

US Global Magnet for AI Capital: Global investors are zoning in on the US as the key market for AI investments, both private and public. This despite the growing global opportunity around ‘Sovereign AI’. The prime attraction of course remains US AI leadership up and down the AI tech stack. From AI Infrastructure all the way to Ai Applications and Services. And of course the rising ability of these companies to absorb and deploy capital at scale. This despite China being the second largest global AI market, with half the world’s AI Talent. Given current US/China geopolitical trends accelerating, China is a less accessible area for foreign investments. For now, the US remains unique in its ability to deploy hundreds of billions in AI capital a year More here.

Enterprise AI Software battles accelerate: Competition in enterprise software ‘SaaS’ companies is accelerating with AI products and services. Microsoft, Salesforce and a wide array of software companies, along with Google, OpenAI, Amazon and others, are in competition to change entrenched business software, habits and practices. Companies large and small while accelerating AI deployments, are also rapidly picking and choosing from a growing plethora of options at scale. This is in contrast to the relative slow ramp that had been noted in earlier months. .Additionally enterprise pricing is also moving away from per-seat to per task, especially on AI reasoning and agentic applications. Much of this remains to be developed and deployed at scale, but the core companies are fielding larger sales resources into the effort. More here.

Why AI Robot Hands are Hard: New reports on AI Robotic hands illustrate why they’re so hard to develop and build at scale. It’s an important gating factor to humanoid AI robots at scale, an area that has seen billions in new investments of late. The problems are both hardware and software, starting with the utter paucity of AI training data for robots. This is relative to the almost three decades of data on the internet that LLM AIs had an opportunity to feast on to date. Despite legal issues and complications. The other issue of note of course is the formidable manufacturing scale for AI robots exists mostly in China. And it would have to be built from scratch in this ‘Made in America’ phase of US Tech. It’s a challenge for everyone from Tesla to scores of recent AI robot startups.. More here.

More like this