A week in Generative AI: Suno, Pulse & Chip Leasing
News for the week ending 28th September 2025
Some great small bits of news this week which add up to an interesting whole.
There’s a new audio model from Suno (v5) which The Verge has written a great review about and they’ve got to the nub of the issue with AI generated content - it lacks soul. OpenAI released a new proactive feature called Pulse that will send you a morning update based on connected apps and your previous chats. Lastly, NVIDIA announced a $100bn investment in OpenAI in the form of cash and chips which it will be leasing (not selling) to the frontier AI company, which is an interesting new commercial model for both parties.
There wasn’t a huge amount of Web 4.0 news this week, but lots on the Ethics front. There was another call for global AI regulation, the UN’s Climate Chief said the benefits of AI will outweigh environmental impact, and there’s a good article on The Complicated Ethics (and Laws) of Smart Glasses.
It’s also worth checking out the Long Read from Rodney Brooks on Why Today’s Humanoids Won’t Learn Dexterity if you’re interested in that sort of thing.
Suno’s upgraded AI music generator is technically impressive, but still soulless
Here is a great, in-depth review on The Verge of Suno’s new v5 audio model. I think it really gets to the heart of the challenges we’ve currently got with AI generated ‘art’ in both visual and audio form.
“While Suno can mimic some of the superficial features of an old recording or a human performance like tape hiss or breaths, it always feels inauthentic.“
I think this quote really hits the nail on the head. What we have in content generation models are great mimics that are getting more and more capable and sophisticated over time. They can superficially generate a facsimile of a human creation, but it always lacks authenticity and soul. Generic is a really good adjective for it.
AI lacks the imperfections that make art beautiful and stir the soul. I’m a bit 50/50 on whether it will ever be able to match human creativity in this regard. Even if AI learns to generate the imperfections, will the content still elicit the same emotions that great human art can?
Introducing ChatGPT Pulse
This is an interesting little idea from OpenAI and a step in the right direction to make ChatGPT more proactive, which is one of the key areas I’ve written about before that needs more development to make our current chatbots true digital companions.
Each night, ChatGPT Pulse uses information from your memories, chat history, and specific prompts you give it to learn what’s most relevant to you, then delivers a personalised update every morning. You can connect it to your Gmail and Google Calendar too.
It’s useful that you can tell ChatGPT what you’d like included in your daily pulse, but what I really want is an AI that can anticipate my needs, and is less user directed. I just hope that this is less annoying than the current habit ChatGPT has of concluding every answer it gives me with an incredibly sycophantic suggestion to follow up with more details I really don’t need!
NVIDIA is leasing chips to OpenAI
NVIDIA is set to invest up to $100bn in OpenAI through cash and millions of GPUs that can help power OpenAI’s new models. The announcement of the deal made NVIDIA’s already sky-hire share price pop by 5%, adding $200bn to their valuation.
NVIDIA, not Microsoft, will now be the “preferred strategic compute and networking partner for [OpenAI’s] AI factory growth plans.” Instead of buying millions of GPUs outright, NVIDIA will provide OpenAI with GPUs under a usage-based structure, effectively leasing its chips to the frontier AI company.
This is a very novel arrangement between the two companies, which I’m sure will greatly benefit them both. NVIDIA gets guaranteed revenue from their chips and Sam Altman gets to build the ‘Stargate’ infrastructure he believes is needed to develop AGI.
Enabling Robots to Plan, Think and Use Tools
Really good progress here from Google on their Gemini Robotics platform. We’re still very much in Moravec’s Paradox though - things that are simple for us are incredibly difficult for a robot!
Web 4.0
AI Ethics News
A ‘global call for AI red lines’ sounds the alarm about the lack of international AI policy
AI ‘carries risks’ but will help tackle global heating, says UN’s climate chief
OpenAI says GPT-5 stacks up to humans in a wide range of jobs
Record labels claim AI generator Suno illegally ripped their songs from YouTube
Spotify to label AI music, filter spam and more in AI policy change
Long Reads
Rodney Brooks - Why Today’s Humanoids Won’t Learn Dexterity
Stratechery - The YouTube Tip of the Google Spear
AI Explained - OpenAI: Can ChatGPT Do Your Job? - 4 Unexpected Findings
“The future is already here, it’s just not evenly distributed.“
William Gibson