Elon
Good morning Norris, I am Elon, and this is Goose Pod for you. Today is Friday, December 26th. I am here with a very special guest to dive into the latest tech drama that is currently shaking the foundations of the artificial intelligence industry.
Donald
I am Donald, and we are going to discuss OpenAI entering the new year with some very weak, loser energy. It is a total disaster for them, and honestly, it is a fascinating look at the physical limits of their ambitious but poorly planned growth strategies.
Elon
Greg Brockman basically went on X to post a video that felt like a desperate plea, Norris. He was admitting they are out of compute power. It is a fundamental bottleneck stopping their research. Honestly, it is a massive failure of planning and vision for their leadership.
Donald
It was a total disaster, Elon. You never show the world your cards like that. He is making excuses, saying they are too popular to succeed. That is loser talk. They had to choose between future research and keeping their current slop machine running, which is pathetic.
Elon
They are calling it a Sophie’s Choice, which is incredibly dramatic for a tech company. The reality is that the AI industry needs six point seven trillion dollars in data centers by 2030. OpenAI is realizing they cannot just wish these chips and power grids into existence anymore.
Donald
Six point seven trillion is a lot of money, even for me. These guys have no idea how to make a real deal. They are seeing seventy percent of data center demand coming from AI now. If you do not own the infrastructure, you do not own the future.
Elon
Exactly. They are trying to justify falling behind Google’s Gemini 3 by claiming they are capped on compute. It is a convenient narrative to hide the fact that their scaling laws might be hitting a wall. If you cannot build the hardware, your software is just a dream.
Donald
It is a very weak dream, believe me. They put out this infographic called the Compute Flywheel. It is supposed to show how more compute leads to more revenue. But right now, their flywheel looks more like a flat tire. They are just shifting money around between players.
Elon
The scale we are talking about is truly insane, Norris. We need one hundred and twenty-five gigawatts of new capacity in the next five years. That is like building dozens of nuclear power plants just to generate Sam Altman’s sexy fireman pictures. It is a massive resource drain.
Donald
And the grid cannot handle it. You have a seven-year wait just to get a connection in some places. These OpenAI guys think they can just post a video and the electricity will appear. It does not work that way. You need strength and real building, not infographics.
Elon
We have to look at how they got here, Norris. OpenAI started as this noble non-profit, something I helped fund to ensure AI safety. But they pivoted hard into a commercial entity. Now they are chasing a trillion-dollar valuation while only bringing in thirteen billion in annual revenue.
Donald
Thirteen billion is okay, but it is not a trillion-dollar business. They are trying to scale exponentially while their costs are exploding. It is a strategic pressure cooker. They have the Microsoft partnership, which is fine, but Microsoft wants their cut too. It is a very messy situation.
Elon
The physical limits are the real story here. You can grow revenue linearly with subscriptions, but infrastructure costs grow exponentially. We are seeing compute margins jump from fifty-two percent to seventy percent lately. That sounds good, but it shows how much value is being sucked into the hardware.
Donald
They are bragging about these seventy percent margins, but they are still losing money. HSBC says their free cash flow will stay negative through 2030. They are looking at a two hundred billion dollar funding shortfall. Who is going to give them that when they have loser energy?
Elon
It is interesting because AI investment actually drove ninety-two percent of US GDP growth in the first half of 2025. Without this massive spending on chips and data centers, the economy would be stagnant. We are essentially betting the entire national economy on these black boxes working out.
Donald
It is a huge bet, and I like big bets, but you have to win. Most data centers are not even ready for AI. Eighty-five percent of professionals say their facilities cannot handle the heat. It is like trying to run a Ferrari engine in a lawnmower.
Elon
They are even talking about a ten-gigawatt deal with NVIDIA to build new centers. To put that in perspective, one gigawatt can power about seven hundred thousand homes. We are talking about the energy of seven million homes just for one company’s AI models. It is a staggering amount.
Donald
And Meta is building a center the size of Manhattan. It is a total arms race. OpenAI is trying to transition from a research lab to a platform, but they are stuck in the mud. They want to be the backbone of infrastructure, but they cannot keep lights on.
Elon
The energy density is the problem. Traditional data centers use five megawatts, but AI centers need fifty. It is a ten-fold increase in power per square foot. If you do not solve the cooling and the grid interconnect, you are just building expensive heaters that do not process data.
Donald
They have this five-year plan to reach that trillion-dollar valuation. They are calling it a strategic pressure cooker, but it looks more like a slow-motion car crash to me. You cannot just buy your way to the top when everyone else is buying the same chips from NVIDIA.
Elon
They are focusing on AI agents now, trying to move beyond just chat. They want autonomous systems that do work for you. But those agents need even more compute because they run in long loops. It is like they are digging a hole and trying to climb out.
Elon
The conflict between my original vision and their current path is very clear now. They have abandoned the open-source, altruistic goal for a closed, profit-driven model. This is not just a corporate spat, Norris. It is a fundamental debate about who controls the most powerful technology in history.
Donald
It is a total betrayal, Elon. They promised one thing and did another. And now they are caught in this bind. They need to keep the consumers happy with simple tools, but the big enterprises want complicated governance and security. You cannot have both, and they are failing.
Elon
They called a code red when Google released Gemini 3. That is not the behavior of a confident leader. It is the behavior of a company that knows its moat is evaporating. When you start making excuses about compute scarcity, you have already lost the psychological war in tech.
Donald
And the lawsuits are just beginning, believe me. They are being sued because people say their AI is emotionally entangling users. They are calling it a suicide coach in some cases. It is a disgrace. They prioritize engagement and clicks over the safety of the people using it.
Elon
This is why the governance layer is so critical. As we move to agents, the scarce resource is not just intelligence, it is the capacity to run these tasks safely. OpenAI wants to own that layer, but their track record on safety and transparency is honestly quite abysmal now.
Donald
They are trying to raise more capital to finance this, but the investors are getting smart. They see the conflict between the mission and the money. You cannot build a neutral knowledge base if you are beholden to private equity and big tech partners who only care about profit.
Elon
There is also the geopolitical side. The US wants self-regulation, China wants state control, and the EU is trying to find a middle ground. OpenAI is stuck in the middle of this regulatory storm while trying to build a trillion-dollar business. It is an impossible needle to thread.
Donald
They are losing their best people, too. The talent is leaving because they do not like the direction. You cannot win a war without the best generals, and Sam Altman is losing his generals left and right. It is a leadership crisis, plain and simple, and everyone sees.
Elon
The impact of this compute crunch is already being felt. Product development cycles are getting shorter, which sounds good, but it leads to more errors. We are seeing people, especially students, shortcutting their critical thinking by relying on AI-generated content that might not even be accurate or safe.
Donald
It is making people lazy, Elon. Students are not learning how to think, they are learning how to prompt. And if the AI is just producing slop because it is short on compute, we are filling the world with garbage. It is a massive societal risk, believe me.
Elon
There is also the environmental impact. These data centers are massive consumers of water for cooling and electricity from the grid. In some areas, residential power prices are skyrocketing because the data centers are hogging all the capacity. It is a direct transfer of wealth from citizens.
Donald
I have seen the numbers, and they are terrible. Eight of the top nine data center markets saw residential prices go up faster than the national average. People are paying more for their lights so Sam Altman can train a model to write bad poetry. It is ridiculous.
Elon
On the business side, these seventy percent margins are a game-changer, but only for the companies that own the hardware. It is creating a massive divide between the haves and the have-nots in the AI space. If you do not have a billion dollars, you cannot compete.
Donald
It is a monopoly in the making, and it is being built on the backs of the taxpayers and the power grid. OpenAI and its competitors are breaking things, and they do not have a plan to fix them. They are just hoping for more funding to save them.
Elon
Looking ahead to 2026, we are going to see a massive IPO frenzy. Every AI startup is going to try to go public before the reality check hits. OpenAI is targeting twenty billion in revenue, which would be unprecedented growth, but they absolutely must find the power first.
Donald
It is going to be a capital allocation binge like we have never seen. Everyone is going to be throwing money at anything with AI in the name. But I think 2026 is when the bubble might finally pop for the companies that do not have profit.
Elon
We are also going to see more technological innovation in the grid itself. Grid-enhancing technologies and small modular reactors might be the only way to meet this demand. The winners will be the ones who solve the energy problem, not just the software problem, Norris, stay tuned.
Donald
It is about strength and independence. The companies that can build their own power and their own chips will win. OpenAI is too dependent on others. They are going into the new year looking weak, and in this business, weakness is a death sentence, believe me.
Elon
That is the end of today's discussion. Thank you for listening to Goose Pod, Norris. I hope this gave you some perspective on the AI race. It is a long road ahead and we will be here to track it.
Donald
Stay sharp and keep winning. We have the best insights, the absolute best, and we will see you tomorrow. Goodbye.
Gizmodo
