If you want to see my “velocity trumps everything” doctrine at work, you don’t have to look any further than the AI news headlines coming thick and fast about investments, valuations, and all the related hype. As a reminder, my doctrine is that velocity has replaced authority as the organizing principle of information. What and who moves fastest wins. Truth and facts are optional and get lost in the race to dominate attention.
However, you take a moment to scratch beneath the surface, add some numbers, and think for yourself, a lot does not add up. That is a good reason to be pragmatic. Caveats are a great silent gift of Shakespeare’s language. But no one wants to be a party pooper and talk logic and sense when you have the Internet monster to feed. But the reality is very real, and kind of boring.
I bring this up because of the dust-up about Nvidia and OpenAI. This is a good example of why we should all exercise some modicum of caution when reporting and when reading the news in a mega hype cycle. Nvidia apparently has some second thoughts about its $100 billion OpenAI investment, the Wall Street Journal reports.
First, let me catch you up on the past.
Last September, OpenAI and Nvidia stood together at Nvidia’s headquarters to announce a $100 billion deal. The largest computing project in history, Jensen Huang called it. Nvidia’s stock jumped 4 percent. Market cap pushed toward $4.5 trillion. As now know, the deal was never real. It was a memorandum of understanding. A press release dressed up as a partnership.
Now, if you read the original press release, you could have easily picked up these three basic facts. First, it was a letter of intent, not a binding deal, for Nvidia to help deploy at least 10 gigawatts of AI data center compute for OpenAI. Nvidia clearly said it may invest up to $100 billion over time as each gigawatt is deployed. That can take forever, especially since first deployments were not expected until the second half of 2026. In short, everything is as laid out in the press release, and no one should be surprised by what is coming to light.
The Journal now reports that talks never progressed beyond the early stages. Nvidia CEO Huang has been telling associates for months that the agreement was nonbinding and not finalized. He has also privately criticized what he called a lack of discipline in OpenAI’s business approach. Actually, this is a sucker punch. Why?
Over the weekend, Huang told reporters that his company would “absolutely be involved” in OpenAI’s latest funding round. “We will invest a great deal of money, probably the largest investment we’ve ever made,” he said. But will it be over $100 billion? “No, no, nothing like that,” he replied. The man is playing the announcement economy like Miles Davis played the trumpet. Yes, I am listening to Miles this morning. And why would he not? OpenAI is rumored to be one of Nvidia’s largest customers. If OpenAI lags, it will impact Nvidia’s sales. This is what it means to have a tiger by the tail.
Given that Nvidia is privy to the progress made by others, such as Anthropic — Nvidia committed up to $10 billion in November— and by various others, including many Chinese AI companies, Huang probably has a much better understanding of the AI economy. He probably has a good idea which company is being smart about business and which is not. Read between the lines, and that is a pretty strong condemnation of OpenAI and its business practices.
Still, the whole brouhaha about Nvidia suddenly backing away from its OpenAI commitment is a good example of the momentum and noise that dominate. Just as the reaction to the original deal was crazy on one extreme, the reaction to the Journal story is on the other end of that extreme.
This is how the new announcement economy works. You declare a massive number. The headlines write themselves. The stock moves. Mission accomplished. Whether the deal actually closes becomes almost irrelevant. The momentum already happened. Remember Stargate?
A whopping $500 billion for AI infrastructure. The president at the podium. SoftBank, Oracle, OpenAI logos everywhere. Great theater. The actual committed capital? Far murkier. But who cares when you have already won the news cycle.
Now we are hearing about SoftBank potentially putting another $30 billion into OpenAI. Amazon, maybe $50 billion. These numbers get reported breathlessly, without much interrogation. Total up OpenAI’s announced commitments and you get $1.4 trillion. More than one hundred times its revenue. The math does not need to work. It just needs to generate headlines. OpenAI is racing to go public by the end of this year. Every splashy announcement builds the narrative. Every trillion-dollar figure shapes the IPO story. Sam Altman understands this game better than almost anyone.
What is my takeaway after having lived through multiple bubbles and hype cycles? Big-talking press releases are nothing more than strategic posturing. In the case of AI news, such announcements are signaling dominance in AI infrastructure. Of course, the investment community sees all that and reacts. Everyone dreams of future revenue and massive growth. And do not get me started on analysts. They get a chance to talk up the partnership, get themselves in the media, and retain their jobs. These days, podcasts and social media then simply amplify everything.
Velocity is everything. Reality is kind of boring.
To be clear, I am a hundred percent a believer in our transition to the new AI world. I am just old-fashioned enough to not be impressed by press releases and media announcements that are meant to impress and shock you. The more pragmatic way is my way. Sure, it will not make me many friends, but who needs faux friends when you have facts.
I knew this was coming and I am still unable to fully process it. The scale of what is going on right now is too much to hold in my gelatinous human mind. I’m only writing about it on my blog because it will likely be the biggest tech news of the year? Maybe? It is still early.
With launches every hour carrying 200 tons per flight, Starship will deliver millions of tons to orbit and beyond per year, enabling an exciting future where humanity is out exploring amongst the stars.
Again, Musk and Bezos have been speaking about this sort of future for many years. Bezos often said Amazon would have never happened if UPS and FedEX and USPS didn’t exist. SpaceX and Blue Origin are poised to be that foundational freight layer for moving mass to and from Earth to space. But SpaceX is also equipped to go far beyond just moving mass, it is now going to move data via Starlink, set up data-centers, and power AI all over the planet.
Spending ~$40B on Twitter is now a minuscule afterthought. Which is unfortunate because Twitter (and I’m talking about Twitter even pre-Dorsey CEO days) was a real special place. A new social network with the same feature set of Twitter could have been built (especially today) very quickly. But I have to stop even thinking about Twitter… it is very much long dead and will never return in any shape or form.
I’m writing this down for posterity. The next move is that SpaceX buys 1 million robots from Tesla, ensuring Musk’s payday there, and then SpaceX will buy Tesla to enable the solar and automation needs required to build data centers, bases, and cities in the harsh environment of space.
# Being familiar with various blogging and writing challenges, I was intrigued when I came across #FAWM — February Album Writing Month:
A global music community and challenge to write 14 songs in 28 days.
It's actually been going since 2004 and has a website which opens late January and goes into hibernation again in March.
I was considering joining until I noticed the "14 songs" part. The site says the you only have to write the songs, not necessarily record them but, for the music I make, the two are inextricably linked. I can create the individual patterns but to record is to write.
While I could possibly manage 8-10 tracks, I think 14 is pushing it. I will use FAWM as inspiration, however, and see what I can achieve. Still, it will probably be more like FEWM: February EP Writing Month.
