AI: Weekly Summary

Michael Parekh
5 min readSep 2, 2023

--

… week ending September 1, 2023

Welcome to the Saturday weekly AI summary here on ‘AI: Reset to Zero’. Trust everyone is preparing for a great Labor Day weekend. Here are this week’s AI highlights:

  1. Google Cloud Conference focused on new AI offerings, announcing the Nvidia DGX Cloud Partnership: Also, Meta plus other LLM AI Model services were announced:

Above events are significant for Google showing a renewed focus on enterprise AI services for Google Cloud. They are currently third behind Microsoft Azure and market leader Amazon AWS. Also notable was Nvidia cementing its DGX GPU data center compute services ACROSS ALL the major cloud vendors. Nvidia Founder/CEO Jensen Huang showed up on stage with Google Cloud head Thomas Kurian. Separately, Google matched the $30/user pricing by Microsoft a few weeks ago for AI-augmented Google Workspace productivity apps. Again, a more detailed dive here on broader implications, especially for Nvidia. The newest trillion which is turning into the new Nike. In my view, Nvidia is executing their version of Intel’s famous ‘Intel Inside Strategy’ in the PC Tech Wave. Developers around the world focused on AI services, are paying attention.

2. Google Tensorflow TPU advantages relative to Nvidia GPUs:

Google is in a new semiconductor industry report comes out in a pole position with their Tensorflow TPU AI chips today. That complemented with Google’s next-gen LLM AI Gemini puts the company on solid footing vs OpenAI and Microsoft with GPT 4 and ChatGPT, over 2024 and beyond. Technically, Google is leveraging its direct relationship with Taiwan Semiconductor (TSMC) fabs in Taiwan for scale production into 2024. All this is relevant for the gating of AI Augmented Google Search services next two years competitors. Note that Meta is also one of top two Nvidia GPU customers, along with Microsoft. So Google having independent scale is a key advantage,

3. OpenAI hits $1 Billion Revenue Run Rate; announces ChatGPT Enterprise vs Microsoft Bing Enterprise:

The story above validates the billion dollar run rate for OpenAI given the expected momentum for ChatGPT Plus $20/month subscription services, supplemented with the rising popularity of third party API services ramping off their GPT 3.5/4 LLM AI models. OpenAI ChatGPT Enterprise should build new growth in 2024, presumably in coordination with Microsoft Bing Enterprise efforts, announced earlier. Separately, Google matched the $30/user pricing by Microsoft for CoPilot Office 365 a few weeks ago, for AI-augmented Google Workspace productivity apps. My more detailed take here.

4. AI data center company CoreWeave builds strong momentum in AI data centers with Nvidia allocation and backing: WSJ has an ‘origin story’ feature highlighting how the NJ based company transitioned from Crypto mining to state of the art AI data centers, with Microsoft, Nvidia and other industry stalwarts as key customers and investors. Company also selling $500 million plus in stakeholder shares at a 3x valuation bump from May, at a $6 billion number. That would be double ‘forward revenue multiple among publicly traded cloud firms’. Shows continued appetite for AI investments by corporate strategic and financial venture investors at large.

5. On the Regulatory front, Meta moves to counter Canada’s new law to pay news outlets for content working for the company so far: Latest data shows Meta seeing no impact on Canadians’ usage of Facebook services. Independent tracking data shows ‘daily active users and time spent on app in Canada staying roughly unchanged”, since Meta started blocking news at the start of August. Bolsters both Meta and eventually Google’s resistance to Canada’s new rules. Those themselves were inspired by similar moves in Australia that were settled by both companies. Events here also have potential bearing on similar developments in EU markets later this fall. The broader issues of what the AI industry ultimately pays for data and content to creators remains an ongoing tussle.

Other items of note on this week’s AI related events:

  1. I did a piece on the US and China continuing to ‘thread the needle’ on their relationship on Trade, AI, and national security concerns. The week saw a visit by the third senior White House visitor to China by Commerce Secretary Gina Raimondo. The talks and engagements continue despite real issues and differences.
  2. Also did a piece on US youth focused on iPhones vs Android smartphones and a counter-trend to simpler flip-phones to counter ill effects of technology. Also on a related topic of near-sightedness (Myopia) in Asia that is spilling over to Europe and the US, again starting with the younger segments of the population. Important given the coming AI ‘Smart Agent’ apps and services by Meta and others, that will likely garner more smartphone time for tens of millions.

AI Papers and Developments:

  1. New AI Paper on ‘the Dark Side of LLM Models’, with a focus on security threats and vulnerabilities, by researchers from Tilburg University and University College London. Concerns of negative impacts of LLM AI about as we go into the second half of 2023 and beyond.
  2. Microsoft announced open-sourced VALLE-X multilingual text-to-speech synthesis and Voice cloning model. This followed a well received AI research paper on topic. This is on the heels of last week’s release of SeamlessM4T by Meta, also focused on multi-lingual LLM capabilities. I have a deeper dive on LLM translators here.

Thanks for joining this Saturday with your beverage of choice. Enjoy this long Holiday weekend, and the end of Summer. Happy Labor Day All.

For your Sunday tomorrow, this week’s ‘AI In my View’ is coming up. 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)

(You can also subscribe to my Newsletter “AI: A Reset to Zero” for free on Substack for more content like this.)

--

--

Michael Parekh

Investor in tech-driven business resets. Founded Goldman Sachs Internet Research franchise in 1994. https://twitter.com/MParekh