Nvidia's $40B investment spree, Google's carbon cover-up, and the Musk v. Altman trial keeps delivering

Something about the Nvidia story stopped me cold this morning. Not the number itself - though $40 billion in a single year is genuinely staggering - but what it signals about where the real power in AI is quietly consolidating. Jensen Huang isn't just selling the picks and shovels. He's buying stakes in the mines.

Buckle up - today's edition covers Nvidia's investment empire, a carbon accounting scandal at Google, the courtroom drama keeping the AI world glued to their feeds, and Sony's bet on AI-assisted game development.

Nvidia's CEO Jensen Huang co-founded the company back in 1993. But what was Nvidia's very first product?

The answer is hiding near the bottom of today's newsletter... keep scrolling. 👇

Nvidia has already committed $40 billion to equity deals in the AI ecosystem in 2026 - and we're not even halfway through the year. That figure puts it firmly in the conversation alongside the world's largest sovereign wealth funds and most active VC firms.

This isn't a side hustle. Nvidia is systematically taking equity stakes in the companies building on top of its hardware - creating a flywheel where its own chips power the growth of companies it partly owns. It's a strategy that makes Intel and AMD's approaches look almost quaint by comparison. If you've been tracking AI investment trends, this is the story that reframes everything else.

For developers and founders building on Nvidia's stack, this raises a real question: is your infrastructure provider also becoming your investor and potential competitor?

Read the full story →

Planning documents reviewed by The Guardian reveal that developers working for Google significantly understated the carbon emissions of two proposed AI datacenters in the UK - by a factor of five. One facility covers 52 hectares (about 130 acres) in Essex. A similar error appeared in a second set of Lincolnshire plans from another developer.

Understating emissions by 5x in planning applications isn't a rounding error - it's a figure that fundamentally changes how local authorities and the public evaluate these projects. Environmental concerns around AI datacenters have been growing for months, and this kind of revelation makes those concerns significantly harder for the industry to dismiss.

The timing matters too. The UK is in the middle of navigating its own tech policy framework around AI infrastructure. Discoveries like this tend to harden regulatory positions - and could complicate approvals for datacenters that are genuinely needed to run the models everyone's building on.

Read the full story →

The second week of the Musk v. Altman trial has delivered more than anyone expected. Court documents have surfaced showing that Microsoft executives were genuinely worried OpenAI might walk away from the Azure partnership and publicly disparage the platform - with internal messages describing fears that OpenAI could 'shit-talk' Azure and jump to Amazon instead.

But the most striking detail came from Shivon Zilis, who testified that Elon Musk himself once tried to recruit Sam Altman away from OpenAI. That detail cuts directly against the narrative Musk has been constructing in court - that Altman and Brockman deceived him. Week 1 had Musk alleging he was misled into donating $38 million. Week 2 suggests his relationship with Altman was a lot more complicated.

If you want the deeper backstory on the OpenAI leadership drama, our breakdown of the OpenAI founding and early chaos is worth revisiting.

Read the full story →

In its latest earnings presentation, Sony shared its clearest statement yet on how AI fits into the future of PlayStation. The company called generative AI a 'powerful tool' for game development - while being careful to add that 'vision, design, and emotional impact' remain fundamentally human responsibilities.

Sony's framing is deliberate. The gaming industry is in a genuinely tense spot right now - many indie developers have explicitly rejected generative AI on ethical and creative grounds, while large publishers are quietly evaluating where it reduces cost and time. Sony is threading that needle by positioning AI as assistive rather than generative, which may help it avoid the backlash that other studios have faced.

The practical applications are still vague - Sony didn't detail specific tools or pipelines - but the public commitment signals this is moving from 'we're evaluating it' to 'we're building toward it.' Speaking of building fast: if you're prototyping product ideas in this space, 60sec.site is an AI-powered website builder that lets you spin up a landing page in under a minute - useful when you need to test an idea quickly without a dev queue.

Read the full story →

The answer is the NV1 - a graphics and multimedia card released in 1995. It was Nvidia's first commercial product and was built around a unique quadratic texture mapping approach that, frankly, didn't catch on. The company nearly went bankrupt before pivoting to the GPU architecture that eventually took over the world. From almost nothing to a $40B investment empire in a single year - not bad for a company that almost didn't make it past its first product.

Nvidia is now both the infrastructure provider AND an equity investor in hundreds of AI companies. Does that dynamic make you more or less comfortable building on their stack? Hit reply and let me know - I genuinely read every response and it shapes what we cover next.

That's all for today - see you tomorrow with more from the frontier. For deeper dives, audio breakdowns, and the full archive, head to dailyinference.com.