The Stargate has started dialing

What does the new Stargate project mean for us?

My favorite Stargate episode is “The Fifth Race”. That’s when Colonel Jack O’Neill stumbles upon an Ancient computer that downloads a huge database into his mind. The episode leads to him calling the 8th chevron the first time, introducing the viewers to the Asgardians and the origins of humanity.

What I love about this episode is that it not only changes the whole series for the better, while it also completely rewrites the power dynamics.

Up until now, humanity was a formidable but inferior adversary to the Goa’uld, but now we’re basically ants on a highway who are in way over their heads.

This is why I find Trump’s Stargate Project’s naming incredibly fitting. OpenAI announced it too.

That’s a lot of power in one photo. Source: BBC

So what does this mean? Here’s what OpenAI’s post says:

The initial equity funders in Stargate are SoftBank, OpenAI, Oracle, and MGX. SoftBank and OpenAI are the lead partners for Stargate, with SoftBank having financial responsibility and OpenAI having operational responsibility. Masayoshi Son will be the chairman.

One thing is clear: Altman’s getting more powerful. This move solidifies OpenAI’s position in the market and completely obliterates competition. The federal government decided to enter the AI race and OpenAI is gonna be running the show.

The announcement follows:

Arm, Microsoft, NVIDIA, Oracle, and OpenAI are the key initial technology partners. The buildout is currently underway, starting in Texas, and we are evaluating potential sites across the country for more campuses as we finalize definitive agreements.

This builds on the US CHIPS Act which passed in 2022. Since then, chip construction spending exploded in the US. Now Stargate will enter this game on both supply and demand side of the equation.

I’m envisioning massive R&D facilities popping up at an accelerated rate that will inevitably fuel OpenAI’s research ambitions. Altman will get access to basically unlimited compute to pursue AGI.

That’s an important bit, because compute has always been the hard limit to scaling. Altman claimed in his Intelligence Age blog post that scaling has no limit to intelligence. He said:

In three words: deep learning worked.

Whether or not that’s true is irrelevant now. Even if Marcus is right and deep learning will hit a wall (or is already), there are a few things that I think are now true:

1. Even the current level of technology hasn’t been utilized to 1%. Today’s AI is capable of so much more than coding todo list apps and sending you reminders to take out the trash. However we do not know how to make good use of it yet.

2. With concentrated R&D, we’ll see a mid-term drop in computing costs. As I explained in Part 2 of “We are not feeling the AGI”, gpt-4o-mini is already virtually free. We need o1 to be at that cost level, to make unit economics work for widespread AI adoption.

3. Even is there is a wall in deep learning, much of the problems can be obscured by brute forcing compute and still create value. That’s because (1) and (2) mean that even inefficiently designed AI systems with good constraints can generate more business value. I see that happening in my own projects.

4. The US government intends to become a big spender on AI systems, potentially upgrade everything from public infrastructure to transportation to military. I wouldn’t be surprised if there’d be an attempt to create a New Deal type situation with AI investments.

5. AI creates disproportionate power dynamics as we already see with OpenAI. Elevating this to federal level will widen this gap even more. Given how the White House and Silicon Valley are now a couple and in their honeymoon phase, I think we’re likely heading towards a world that looks like Continuum¹.

Of course this is not a done deal yet, but the rules of the game have been rewritten.

Ants on a highway

So we’re now ants on a highway. Humanity has now moved on to call the 8th chevron and go all in on AI. This will rapidly advance developments and possibly consolidate not just the US but also the US tech sector’s power beyond recognition.

We’re now dealing with technology we barely understand, but keep pushing forward. The e/acc crowd is having their heyday but we’re increasingly becoming like O’Neill in Stargate. All that matters now is calling the 8th chevron and not think about how it affects everything else.

The Stargate Project will likely give AI development the escape velocity it has yet to achieve.

Our society will struggle to adapt and our everyday lives will be bombarded by changes all around. It’s a big thing, but it won’t happen overnight.

Highways exist. Ants too.

What most people forget when discussing AGI is that even if we magically create an AI system that surpasses humans, we would still need to adopt it.

There are some parts of the planet that virtually live in the 18th century today. Yes, 86.3% of the world is literate. But that also means that over one billion people are alive today who can’t read or write.

Technology does not exist independently of human society. The way I use technology every day is vastly different from how my wife uses it, even though we live in the same household. There are stuff that I find trivial that she thinks are impossible.

So yes, technology will advance faster and we will soon see more magic in our lives. But that doesn’t mean our life as we know it is going to end with a snap. This is not the Avengers.

1. Plot from Wikipedia: City Protective Services (CPS) law enforcement officer Kiera Cameron lives with her husband and son in 2077-era Vancouver under the corporatocratic and oligarchic dystopia of the North American Union and its Corporate Congress, a technologically advanced high-surveillance police state.