AI: Weekly Summary RTZ #465

Aug 31, 2024

Nvidia’s driven Q2 results & culture: Nvidia again reported strong Q2 results, exceeding on most metrics. Gross margins were a bit softer due to Blackwell delays and fixes. Guidance for Q3 is strong across product lines from AI GPUs to Datacenter. Blackwell shipments expected in the ‘billions’ in Q4 vs Q3 earlier, with a strong ramp in 2026. Stock sold off about 5% following a 150% increase year to date. Nvidia, led by Jensen Huang’s long-term execution focus on cutting edge AI performance, continues to drive over 90% share on AI chips. Meanwhile, WSJ had a timely glimpse into Nvidia culture, with employees taking recent success in stride. More here and here.

OpenAI’s potential fundraise with Apple & Nvidia: OpenAI is expected to raise a multi-billion dollar round lead by VC Thrive with a billion, with corporate investors Apple, and Nvidia joining the cap table with existing investor Microsoft. Echoes of Anthropic’s recent raises with both Amazon and Google as investors. Now both top LLM AI hyperscalers will have their own camps of corporate strategic investors. The partnerships are of course also about AI Infrastructure partnerships to ramp up on AI GPU gear and data centers to echo ramping investments by Meta, Microsoft, Google and others. OpenAI can also use the Compute infrastructure for its AGI ramp, and rollout of its ‘Strawberry’ Reasoning AI and other product efforts like SearchGPT and others. More here and here.

Meta and OpenAI releases new ‘chatbot’ metrics: This end of summer sees new metrics for the chatbots from OpenAI and Meta, giving a glimpse into their growth at the halfway point in 2024. OpenAI announced 200 million weekly active users, double the number from last November. That was of course the first anniversary of OpenAI’s ChatGPT launch in 2022 that kicked off the current AI Tech Wave in earnest. Meta’s Mark Zuckerberg also announced that its MetaAI now has over 400 million monthly active users and 185 million weekly users. This of course is only a few months after it launch across Meta’s three main global platforms with over 3 billion users. All this foots with Microsoft highlighting its Copilot AI growth in July, up over 150% year to date. All these numbers are from a small base of course, and the quality of the user growth and engagements need to mature into sustained usage. But it’s all good progress towards mainstream AI adoption. More on that here.

AI benefits for Amazon Developers: Amazon AWS Andy Jassy had an informative piece highlighting the AI driven Developer productivity across Amazon and Amazon AWS. Demonstrates that the cobbler’s children have shoes too, as Amazon ‘cobblers’ work hard at providing state of the art AI models and compute to enterprises large and small around the world. AI of course has seen its earliest ‘sticky’ applications with Developers, with Microsoft CoPilot Github being Exhibit A. More here.

Synthetic Data critical for next gen AI models: The debate over the pros and cons of Synthetic Data being created with LLM AI models continues. With commensurate improvements in software innovations to mitigate the risks of ‘Model Collapse’ by using sufficient amounts of ‘real’ data in the mix. Given the importance of massive amounts of new Data to fit the next generations of LLM AI models to Scale, these innovations are critical to keep the Data flowing in ever larger quantities. More here.

AI: Reset to Zero