AI: Weekly Summary. RTZ #535
...week ending November 8, 2024
Election Implications for Tech/AI: The seismic changes post election days will of course have big implications for the US and world on almost every front. Technology and AI landscape will assuredly be impacted, on a wide range of fronts including antitrust, AI regulations (open source etc.), US/China AI geopolitics, AI infrastructure reshoring in the US and many other areas. A new twist this time versus President-elect Trump’s previous term is of course the role of Elon Musk on a wide range of government, and tech related issues. Big amongst them is Tesla’s presence in China, and the potential overlaps between its priorities there, vs China’s Xi Jinping's priorities vis a vis US tariff actions. Lots to continue to monitor here pre and post inauguration in 70+ days. More here.
Shifting Enterprise budgets for AI Software & Compute: New report on enterprise budgets balancing traditional and new AI related applications and services. Some indications that businesses are boosting their AI spend without increasing their IT budgets. This is leading to concerns on ‘zero sum’ pressure on software companies. On the one hand companies like Micrsoft, OpenAI, Amazon, Anthropic and many others are making good headway with businesses on AI deployments. An example is this week’s Amazon AWS/Anthropic deal with Palantir to use Anthropic’s Claude models for Palantir’s defense and commercial customers. On the other hand, investors are concerned that traditional software vendors like Salesforce and other SaaS companies have a more challenging transition ahead. New AI native companies like OpenAI are making meaningful headway on AI applications in enterprises as illustrated in this $100 million, 3 year deal with T-Mobile. More here.
New Study on the value of publisher data to train LLM AI models: Ziff David has a new study on how AI companies rely on content from ‘premium publishers’ to train their ever scaling AI models. This of course meant to add more data on one side of the content/copyright tug of war between publishers and AI companies. Meanwhile, deals between LLM AI companies and content/copyright owners and platforms are accelerating, with Reddit being one of the most recent examples. This follows Meta’s recent deal with Reuters, and of course the ongoing litigation/negotiations between OpenAI and the New York Times. All this is important of course as LLM AI models go beyond internet content for training and inference, to synthetic data and content. And content regulatory issues overseas, as Wikipedia is finding out in India. More here.
Industry implications of Amazon Alexa AI revamp delays : Updated reports on Amazon’s continued difficulties updating its significant Alexa/Echo platform with LLM/Generative AI capabilities. This has been an ongoing issue, and has implications for companies like Apple and Google. Apple of course in the context of updating its industry pioneer voice assistant Siri with AI capabilities as well. Apple’s Software chief Craig Federighi had comments on those efforts and how it may take more time to get it done and rolled out at scale. In the meantime, Apple’s iOS 18.2, which has the first AI enabled elements of Siri, and the much-awaited integration with ChatGPT, is undergoing wider beta tests. Give it all a year to be ready for primetime, and scale deployments. More here.
AI Compute Infrastructure ramp accelerates: The industry debate continues on how many trillions will it take in AI capex to update existing and build new AI data center Compute at Scale, complete with the requisitely ramped Power, Cooling and other critical components. New statements by Softbank’s Masayoshi Son put an upper bound of $9 trillion over the decade vs the $1-2 trillion dollar estimates by Nvidia founder/CEO Jensen Huang over the last year or so. In the meantime, investors from PE firms like Blackstone/Blackrock to sovereign funds in Saudi Arabia, UAE and elsewhere are leaning into the AI compute infrastructure ramp, with Nvidia in pole position in the near term. More here.
Other AI Readings for weekend:
Apple also won’t offer AI features in Europe/EU on regulatory concerns, like Meta earlier this year. Broader context here .
Taiwan’s TSMC suspended the latest AI chips nodes for China AI companies, following US concerns. 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)