The AMD Delusion
Wall Street loves a comeback story. The narrative surrounding AMD’s recent "big day" centers on the idea that they are finally breathing down Nvidia’s neck. They aren't. While the Morning Squawk and every major finance rag celebrate AMD’s Instinct MI325X launch as a direct threat to the H100 hegemony, they are ignoring the cold, hard reality of the software moat.
Hardware specs are for hobbyists. Ecosystems are for incumbents. If you enjoyed this piece, you might want to check out: this related article.
I have watched dozens of hardware startups and mid-caps throw billions at "Nvidia killers" only to realize that engineers do not want to rewrite their entire codebase for ROCm just because the chips are 15% cheaper or boast better memory bandwidth on a spreadsheet. Nvidia’s dominance isn't about silicon; it’s about the millions of developer hours poured into CUDA.
AMD’s "win" is actually a consolation prize. They are becoming the budget-friendly alternative for cloud providers who need leverage during negotiations with Jensen Huang. Being used as a bargaining chip is not the same as being a market leader. If you are betting on AMD to flip the script in the data center, you are fundamentally misunderstanding why companies buy GPUs. They buy certainty. They buy compatibility. AMD is selling hope and a slightly lower bill. For another look on this story, check out the latest coverage from Financial Times.
The Anthropic-SpaceX Pipe Dream
The whispers of an Anthropic and SpaceX collaboration represent the peak of "deal-mania" where the logic of the business is sacrificed at the altar of the brand. SpaceX is a logistics and manufacturing powerhouse that happens to go to space. Anthropic is a research lab trying to build a digital god that doesn't offend anyone.
The "lazy consensus" suggests this is a move to bring AI to the stars—autonomous satellites, self-repairing hulls, or Mars-colony governance.
Here is the truth: SpaceX needs money to fund Starship, and Anthropic needs compute that isn't tethered to the whims of Google or Amazon. This isn't a "synergy" of vision. It is a desperate swap of resources. SpaceX possesses a massive, private power and cooling infrastructure through its launch facilities and Starlink ground stations. Anthropic has a valuation built on the assumption that they can scale faster than OpenAI.
When two high-burn companies swap favors, it’s usually because the traditional capital markets are starting to tighten. Don't look at this as a technological leap. Look at it as a defensive moat being built against a looming venture capital drought. If this deal goes through, it’s a signal that the cost of training models has officially outpaced the ability of Silicon Valley banks to fund them.
The Jet Fuel Crisis is a Logistics Lie
The "jet fuel crisis" mentioned in every morning briefing is being framed as a scarcity problem. It isn't. Global refining capacity for middle distillates is actually within historical norms when you account for the recent pivots in East Asian markets.
The real problem is the "Green Premium" and the complete collapse of the regional delivery infrastructure.
We aren't running out of fuel. We are running out of the political will to transport it efficiently. Airlines are screaming about prices because they failed to hedge their fuel costs during the post-pandemic lull, and now they want a bailout via public sympathy.
Why the Supply Chain is the Real Villain
- Refinery Optimization: Modern refineries are being tuned for Sustainable Aviation Fuel (SAF). This sounds great in a PR packet, but the yield is abysmal. We are trading 10 barrels of conventional efficiency for 1 barrel of "clean" optics.
- The Jones Act Hangover: In the U.S., moving fuel from the Gulf Coast to the Northeast is an exercise in bureaucratic masochism. You cannot blame "global shortages" when the fuel is sitting in Texas and you just don't have the right flagged ships to move it.
- Ghost Demand: The "crisis" is being inflated by speculators who see the airline industry’s desperation as a signal to hoard.
I’ve seen this play out in the shipping industry in 2021. You create a bottleneck, call it a crisis, and then double your margins while the consumer blames "the world state." If you’re an investor, don't buy the "scarcity" story. Buy the companies that own the storage tanks and the pipelines. They are the ones laughing at the Morning Squawk.
The Myth of the Enterprise AI Adoption
Every business analyst is currently obsessed with how companies are "integrating" AI. They point to Anthropic’s enterprise deals or OpenAI’s corporate tiers as proof of a revolution.
It’s a lie.
I’ve sat in the boardrooms. "Integration" currently means paying for 5,000 seats of a chatbot that 90% of the employees use to write slightly better emails or summarize meetings they didn't attend. This is not a productivity boom; it’s a massive transfer of wealth from corporate balance sheets to San Francisco server farms.
To actually disrupt a business, AI needs to touch the core logic—the supply chain, the proprietary R&D, the legal risk assessment. Most companies are terrified to let a black-box model anywhere near those assets. So they buy the "Enterprise" version to show the shareholders they are "innovating."
The real "big day" for AI isn't when a new chip drops or a new deal is signed. It’s the day a Fortune 500 company fires its entire middle management layer because a local LLM can actually handle the logic. We are years, maybe a decade, away from that. Everything else is just expensive theater.
Stop Asking About "The Future"
People keep asking: "What is the next big thing in the AI-Space-Energy nexus?"
That is the wrong question. It assumes these fields are moving toward a unified, better world. They aren't. They are colliding. The energy required to run the AI models that companies like Anthropic build is exactly what is driving the volatility in the energy markets that airlines are complaining about.
You cannot have "infinite intelligence" and "cheap travel" in the same physical economy. The power grids can’t handle it. The cooling requirements for the next generation of AMD chips will require more water and electricity than entire mid-sized cities.
The Brutal Math of Progress
The cost of computation is the new price of oil.
$$C = (P \times T) + L$$
In this equation, $C$ is the cost of innovation, $P$ is the power price, $T$ is the training time, and $L$ is the legacy infrastructure tax. As $P$ rises due to the "fuel crisis" and $T$ increases as models get larger, the $L$—the existing systems we rely on—starts to crumble.
We are watching a cannibalistic cycle where the tech industry consumes the energy sector’s stability to fund a digital expansion that hasn't yet proven its ROI.
The SpaceX Pivot
Watch SpaceX carefully. They aren't interested in Anthropic because they want a "smart" rocket. They want to be the utility provider for the next century. If you can put data centers in orbit, you solve the cooling problem (it's cold in space) and the land-use problem.
That is the contrarian play. The "jet fuel crisis" becomes irrelevant if the most valuable "freight" being moved isn't people in planes, but data through Starlink.
The Morning Squawk focuses on the day's headlines. They see AMD up 3% and think "success." They see a fuel spike and think "trouble." They are looking at the ripples while the tide is going out.
AMD will remain a distant second. The fuel crisis will vanish the moment we stop prioritizing "green" optics over engineering reality. And the Anthropic-SpaceX deal will be remembered as the moment we realized that AI is just a very expensive way to burn electricity.
Build your strategy on the friction, not the hype. The friction is where the money is. Stop reading the headlines and start looking at the bill for the electricity. That’s the only number that doesn't lie.
