😸 OpenAI + NVIDIA's next play

PLUS: Meta and ChatGPT user updates!

Welcome, humans.

If we’ve learned anything scrolling the AI video forums over the last few months, it’s that AI is REALLY good (and we mean, SCARY good) at generating nightmare fuel.

Please don’t click the links below if you get creeped out easily…

For those of you who like horror movies: how about treewalkers lurking in your suburban neighborhood? Or legit night terrors come to life? Or the most horrifying of all, yearbook photos from 1995?! The horror! The memories!!

Now here’s a link for everyone not trying to have trouble sleeping tonight:

Is it just us, or does Dwight look 100% like Jack White of the White Stripes in this pic?!

Here’s what you need to know about AI today:

We dive into the subtext behind OpenAI and NVIDIA’s money moves.
Meta’s new Llama model has seen wide adoption from big companies.
ChatGPT’s monthly usage has 2x’d over the last year.
The U.S. government will vet new models from OpenAI and Anthropic.

Advertise in The Neuron here.

What NVIDIA’s latest earnings call and OpenAI’s new funding round mean for the future of AI.

OpenAI and NVIDIA are pretty much the two companies people think of first when it comes to generative AI.

Call it being the market leaders, call it first mover advantage, call it whatever you want—when these two companies do something, everybody pays attention.

That’s why their latest money moves are worth analyzing a little closer than usual.

First, NVIDIA: The reason investors were spooked by NVIDIA's recent earnings report is because everybody’s still concerned that the big spenders in AI will start to slow down if the ROI isn't there.

We’ve been there, said that.

Sure, the company’s profit margin shrank a bit due to operational expenses rising, but that’s to be expected when you’re making the world’s most powerful GPUs.

Meanwhile, OpenAI: …is raising a new round of funding, and everybody wants a piece. Both Apple and NVIDIA are considering investing this time (Microsoft will likely chip some more in, too).

OpenAI’s most recent valuation in secondary sales was $103B, so this new round will be priced that high (or even higher).

Everybody needs something here: OpenAI needs runway to keep operating, Apple needs access to OpenAI’s models for its AI launch, and NVIDIA needs to keep its sales of GPUs up.

The subtext: The question on everyone’s mind is, “what are the AI use-cases that will actually create the value everyone is looking for?”

Y’know, the magic $600B in revenue everyone needs to generate* to make this AI spending worth it?

While image generators and code assistants are getting all the hype right now, NVIDIA CEO Jensen Huang thinks the real value-creating use-cases are coming next.

He thinks those are…

Large-scale recommender systems (a.k.a better algorithms).
Ad targeting.
Internet search.
User-generated content (improving creation + moderation at scale).
The “Enterprise AI wave”, which Jensen claims is just beginning.

Oh yeah, and general robotics, which Jensen sees as the “next era of AI.”

What 'high-value' use-case for generative AI would you actually pay for, but isn't being offered right now?

Write in.

Write in, pt. 2

*Pun sincerely intended.

FROM OUR PARTNERS

Cerebras created the world’s fastest AI.

With Cerebras, you can chat directly with Llama 3.1 (Meta’s powerful new open-source model) about 20x faster than the big cloud providers:

Query Llama3.1-8B and get 1,800 tokens per second (10¢/million tokens).
Query Llama3.1-70B and get 450 tokens per second (60¢/million tokens).

Already have an app built with OpenAI? No problem—Cerebras uses the same Completions format, so switching is as easy as swapping your API key.

Available via chat and API. Try it here.

Around the Horn.

Meta’s Llama 3.1 has been downloaded ~350M times, and is actively being used by Goldman Sachs, AT&T, Shopify, Spotify and more for customer service, doc review, and code generation.
Two AI coding startups raised huge rounds—Magic just raised $320M, and Codeium just raised $150M.
OpenAI says ChatGPT has 200M weekly active users (2x from a year ago) and said 92% of Fortune 500 companies use its products.
OpenAI and Anthropic signed a deal with the U.S. government to let it review, test, and evaluate all new AI models the companies release.

Treats To Try.

*The biggest gathering of open source AI leaders is coming to San Francisco this fall! Join OpenAI’s CTO and many others at Ray Summit 2024. Register now with code “Newsletter15” for an exclusive discount for readers of The Neuron.
CompanyGPT screens companies based on employee count, growth rate, funding, founder background, web traffic, location, and industry focus.
Podcraftr converts written content into podcasts.
Playground helps you design various types of graphics like art, stickers, and social media posts.
Creatify 2.0 generates viral videos in “one click” (unless you wanna change stuff), transforming scripts or existing videos into short videos with various styles and editing features.
Lummi offers royalty-free, AI-generated stock images for commercial use.

See our top 51 AI Tools for Business here!

*This is sponsored content. Advertise in The Neuron here.

Intelligent Insights

Interview here

Check out this 14 min explainer of everything you can do w/ Claude Artifacts.
Watch + learn how to work w/ Claude to make an app in this short <2 min video from Allie K. Miller (the screenshot + “Fix this” trick is the move!)
Here’s what happens when you spam “hi” at every large language model (related: what happens when Claude draws a self portrait).
Good insight on how AI SaaS tools might echo the trend of dropshipping (original convo starts here).
One of the world’s top AI experts, Stephen Wolfram, thinks we may never be able to explain machine learning because of “wild computation” (full source here).
Google DeepMind CEO thinks we should create a collaborative, private hub where the world’s top researchers build AGI in a vacuum… what could go wrong?