The Neuron
Posts
😺 $7 trillion

😺 $7 trillion

PLUS: investing in AI

Pete Huang
February 12, 2024

Welcome, humans.

Congrats to the Kansas City Chiefs on their back-to-back Super Bowl win!

Overheard in San Francisco: “We sacrificed a self-driving car the night before, and they still wouldn’t give us the W?!”

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

OpenAI’s CEO is courting Earth’s richest for the biggest deal ever.
There are a bunch of ways to “invest in AI” (we share what stands out).
There are 46 AI startups we’re watching for updates from.
A new pair of futuristic AI glasses is on preorder.

*disclaimer: This newsletter is not financial advice; we don’t ever give financial advice!

OpenAI CEO Sam Altman aims for the biggest deal of all time.

Source: Inc / Getty Images

$7 trillion United States Dollars is enough to buy you…

The combined GDP of Japan and the UK.
14,000 of Jeff Bezos' latest superyacht Koru.
A week's groceries at NYC’s Whole Foods (inflation, am I right?!?).

This figure is mind-boggling—spending just 1 trillion is hard enough, let alone 7!

And yet, OpenAI CEO Sam Altman wants to raise $7 trillion to 100x the world’s capacity to build more chips that power AI.

To do it, he’s sliding into the DMs of high rollers like SoftBank CEO Masayoshi Son (who invested/lost $14 billion on WeWork) and the UAE (whose sovereign wealth funds total ~$4 trillion).

But it’s not just about the cash. Even though $7 trillion could build hundreds of new chip plants, it’ll also take:

manufacturing know-how.
specialized machinery.
skilled workers.

All of these are in short supply, given how quickly AI has become the new hot thing.

Why it matters: As Big Tech floats the idea of AI agents and superhuman AI, this type of deal shows we might be in just the opening drive of this blockbuster game.

…and why it matters for your portfolio after a quick halftime break…

FROM OUR PARTNERS

Summarize bulky documents for half the cost with Mixtral.

One of our favorite use cases for AI is trimming lengthy PDFs into a quick TL;DR.

But text summarization is expensive using LLMs like GPT.

Instead, use Mixtral on OctoML, the fastest place to run open-source models (simply replace any GPT with Mixtral in just three lines of code).

OctoML’s benchmarks show over 50% savings in cost vs. GPT-3.5 without sacrificing quality.

What his $7 trillion bet might suggest you look into.

If you want to “invest in AI”, you’ve got tons of options:

Companies making AI models (public: Meta/Google; private: OpenAI/Anthropic).
Chipmakers (Nvidia/AMD).
Those supplying to chipmakers (ASML).
Server builders (SuperMicro).
Cloud providers (Microsoft/Google/Amazon).
Creators of AI software (tons of startups & big tech companies).
Developer tooling and infrastructure (even more startups).

And while finance gurus are always selling you on the “one AI stock poised for a 20x rise”, the reality is that deep-pocketed investors don’t know who benefits the most from AI.

So they’ve proceeded to invest in all of it.

This explains why Nvidia is up 231% over 1 year, SuperMicro is up 717%, OpenAI is worth $100 billion, and any new AI startup (seriously, any) can raise mouth-watering rounds.

One way to read Sam Altman’s $7 trillion bet: out of that long list, chips are the most crucial to building more powerful AI systems (think ink for a printer). You need chips not just to develop ChatGPT but also to use it. And where we’re going, we’re gonna need way more chips than we have.

Are we saying “buy Nvidia stock”? No—again, we don’t give financial advice (even when bombarded by our great aunts on Thanksgiving!).

Around the Horn.

Following a spat of AI-related phone scams, the FCC banned AI voices in robocalls.
AI made a splash in Super Bowl ads this year, but it was nothing like crypto’s blitz of 2022.
Former Pakistan PM sent a victory speech from prison using an AI-generated voice.
Check out the 46 startups at Y Combinator working on AI products.