Intelligence Crashing Down...
OpenAI | Compute free-for-all | Galaxy Digital
The AI Bubble’s Final Draw:
OpenAI’s All-in Bet on the Singularity
Reality is now crashing down on OpenAI, and things aren’t looking too great. The numbers came in and the internal goals that were set? Missed. Both on user growth (aiming for 1B+ weekly users) and revenue.
They’re still growing fast in absolute terms, however the rules haven’t changed. With the forever rising compute costs and fierce competition closing in on all sides, investors are becoming nervous.
The Wall Street Journal went straight for the “impending doom” narrative. And it’s hard to blame them as OpenAI’s own Chief Financial Officer pulled the plug on future compute contracts not being funded if revenue continues to stall.1
This Company was touted to be “Man’s final invention”. So where will this end?
The Burn Rate Reckoning…
OpenAI’s mistake was trying to serve everyone at once instead of focusing on a clear niche early on. Anthropic did the reverse, focusing on coding and enterprise customers that were willing to pay, and that strategy proved effective.
Either way, the economics stopped making sense ages ago. With many arguing that it’s been a massive con from the start, calling Sam Altman the biggest scam artist of all time for a brief period.
But here’s the thing: While progress hasn’t been exponential, scaling hasn’t failed either. Profits may be the lagging indicator for now, but after GPT-5.5? No-one is laughing at Altman anymore.
So while the CFO raises concerns about the future compute contracts? There’s no turning back. Compute is the only realistic edge they have? Google has the fresh data, Grok has the video and X live feed, and Anthropic owns the trust…
Compute is Their Only Realistic Edge
What if it all comes down to this? Scaling.
We don’t actually know the path to “Superintelligence”…
I believe that’s why Google invested $40B into Anthropic2, it’s the OpenAI-lite on the same brute-force compute strategy. Since wherever this goes? Google wants a slice.
Now OpenAI is pushed against the wall, and that’s exactly why they’re so dangerous. They can’t back down so they will need to own the full stack and out-scale everyone on both inference and compute.
They’ve locked in partnerships for gigawatts from literally everyone and are committing up to $1.4 trillion in infrastructure over the next ~7 years (Project Stargate alone is $500B by 2029).
Altman has been crystal clear: “it’s always been about the compute bro”. And directing it at hard problems3. I think that’s the masterplan….
Last Round of Scaling: Compute at all Costs
I think OpenAI plans to solve both profitability and intelligence by doubling down on compute. Making a model so powerful that it can revolutionize entire businesses across drug discovery to energy optimization.
I think it’s the final draw but compute isn’t infinite. Stargate is officially behind schedule. Numerous sites are delayed or even cancelled.
And yet one just came online: Galaxy Digital’s Helios campus in West Texas4. First data hall delivered to CoreWeave this month. Power capacity already doubled past 1.6 GW.
The lights are finally on and I think demand is inevitable.
The Grand Scheme for OpenAI?
These models are already getting capable, even now we don’t know the true reach of its capability as most people aren’t concentrating their AI queries onto hard physics problems.
Imagine if OpenAI manages to actually make ChatGPT capable of running mathematical problems for hours? With virtually limitless compute, there’s no telling what limits there will be to its intelligence.
I think they know trying to sell AI related services to people is a losing battle. Instead they will seek to hone the real intelligence beneath to just solve the remaining bottlenecks across the board.
I mean… What if AI gets smart enough to solve its own data and compute problems? That’s the real inflection point nobody’s pricing in! And eventually scaling that toward other businesses harvesting their data and so on…
Singularity flywheel unlocked.
My Move?
What am I thinking? I’m thinking OpenAI is out of options, so either this crashes and burns or we’re officially entering the singularity through some mystical LLM breakthrough… We’re past the point of no return? So be it.
Here is what I’m doing:
GLXY 0.00%↑ Demand: Their time has finally come, I am already invested for their crypto business but this timing is just too great. Could spike during the summer.
I’m overexposed to GOOGL 0.00%↑: Through them, I get a safe vessel into AI alongside a small stake in both Anthropic and SpaceX. Huang doesn’t believe AI will ever be sentient? Larry Page is betting on DeepMind.
OpenAI Loophole?: IPO is planned somewhere this year and the only way to actually get exposure is through BMNR 0.00%↑. Their $80M investment into Eightco just bought them a part of their OpenAI stake5. With BMNR 0.00%↑now trading on the NYSE, the nature of this trade is evaporating fast.
The Shovels: With the rise of Photonics? I don’t know what to think anymore... Immediate GPU demand will remain. Regardless, I’m still counting on future copper demands ERO 0.00%↑ .
Investors are currently paying a high premium on the assumption that AI will eventually pay off and that the infrastructure costs will inevitably cool. My overall message? Invest where the economics
STILL MAKE SENSE
🛰 Signal LOST: End of transmission — INVADER out…
Disclaimer: This post is for informational and entertainment purposes only. It reflects personal opinions and is not intended as financial advice. Always do your own research before making investment decisions