AI: Weekly Summary. RTZ #646

Michael Parekh AI: Reset to Zero

3 weeks ago

3 MIN READ

Nvidia results strong again: Nvidia reported strong fourth quarter results, with mixed market reactions. Company showed strong demand for its AI chips, tempering the post DeepSeek concerns. Revenues up 78% beat analyst expectations. Chief concerns going forward remain Trump tariffs and Biden AI Diffusion Rules tiering the world in three tiers. Decisions on the Biden rules by the Trump administration up ahead, but long term concern would be China building its own chips and building global market share vs Nvidia and other US semi companies. Despite good results, short term concerns surround AI spending and GM impact from new AI infrastructure products. More here.

OpenAI releases GPT 4.5 on to GPT 5: OpenAI released its expected GPT 4.5 model, the last one without AI Reasoning built in. That will be in GPT 5 later this year, a true third generation LLM AI model. It’s being released as a research preview for now with more free and paid tiers later. GPT 4.5 is big in its compute capabilities, and reactions have been favorable for the most part. And competitors ranging from Google, Anthropic, Grok, DeepSeek and others remain strongly in the race to next gen LLM AI models (see next item). However, OpenAI remains the perceived global leader, including rising competition by China AI companies like Alibaba. More here.

Anthropic’s next Gen Claude 3.7: Anthropic released its next gen Claude 3.7 LLM AI, with integrated reasoning and agentic features. Dubbed a ‘hybrid reasoning model’, Anthropic leads into a feature promised by OpenAI with GPT 5 later this year, where reasoning is integrated into a simpler and familiar chatbot interface. The other claim to fame is that it’s more a Gen 3 LLM AI, again more similar to OpenAI’s GPT 5 to come. xAI’s Grok 3 also claims next gen status. Anthropic’s models will also power some of the revamped AI Alexa features from its investor/partner Amazon (next item). The global market for LLM AIs continue to accelerate into integrated AI reasoning features. More here.

Amazon shows off AI Alexa+: Amazon showed off its long awaited AI powered Alexa, with a tightly guided set of demos. The service employs a host of LLM AI models, including the latest from its partner Anthropic. Plan is to release these capabilities into the multi million global installed base of Amazon Alexa devices later this year. Initial release is planned for Amazon Show devices that incorporate a screen display. Pricing is $20 a month for a standalone subscription, with hundreds of millions of Amazon Prime subscribers getting it bundled into the current $15/month price. It’s designed to boost Prime subscriptions of course. Next similar voice AI competitors are Apple with Siri, and Google with its various devices. More here.

Meta plans standalone Meta AI & Reels apps: Meta is releasing its Meta AI service as a standalone app, to better compete against OpenAI’s ChatGPT and others. It’s alson contemplating a standalone Reels app for its short videos on Instagram, presumably to benefit against a possible TikTok ban, and against Google/YouTube’s Shorts app. That move would also bolster its options with content creators and advertisers. And lessen the load of heavy lift of multiple features currently packed into Instagram. The move to separate apps highlights the next leg of AI competition amongst the major companies for stronger distribution to hundreds of millions and billions of mainstream users. More here.

More like this