Jordi’s Sunday Market Report: The Fireworks Are Over, But the AI Cycle Is Not
Main takeoff points
- This looks more like a rotation than the start of a sustained bear market: after a nine-week S&P run and an eight-of-nine-week Nasdaq run, the tape finally punished crowded beta.
- The easy phase of the AI trade is over. The agentic AI buildout is still early, but the buy-anything memory/semi/infrastructure chase has moved from discovery to digestion.
- Capital did not leave the market; it rotated. Healthcare and financials caught bids as money came out of the most crowded AI-linked winners.
- The right question is not whether AI is a bubble. It is whether the macro data, credit spreads, earnings, and profit margins have actually broken. Jordi’s answer this week: not yet.
- The next AI opportunity is moving up the stack: from memory and hardware scarcity toward applications, edge compute, optical networking, power, healthcare, and “human software.”
- Inflation and energy remain the constraints that can make this cycle feel strange: compute turns electricity into tokens, and oil/geopolitical risk keeps policy from having a clean escape hatch.
The market finally made investors pay for crowding
The week looked ugly if you owned the most crowded part of the tape. The S&P 500 fell 2.6%. The Nasdaq dropped roughly 4.5%. Beta got hit hard enough that Jordi Visser compared the factor move to episodes that force you to look back toward the early-2000s period for anything similar.
But his main point was not panic. It was discipline. A market that had run for weeks without a real reset finally gave investors a reset. That is different from saying the structural story has failed.
Jordi’s phrase for this week is the right one: the fireworks show is over. The phase where investors could buy almost anything attached to memory, semiconductors, data-center infrastructure, or agentic AI and watch it gap higher is finished for now. The odds on the tote board are no longer mispriced in the same obvious way. The trade has been discovered. The next phase is harder.
That does not make it a bear market. It makes it a two-sided market.
Rotation is not the same as repudiation
The cleanest distinction in Jordi’s latest video is between a failed theme and a maturing theme. AI did not stop mattering because memory stocks sold off. A narrow, parabolic part of the AI complex simply stopped offering the same risk-reward.
That is why the rotation matters. On Friday, the pain was concentrated in the areas most directly tied to the prior AI chase, while healthcare and financials acted better. Capital did not disappear. It moved.
That is exactly what should happen when a powerful structural theme advances from discovery to digestion. Early in the move, investors under-own the theme and chase the obvious names. Later, everyone knows the story, the easy multiple expansion has already happened, and the market starts asking a more adult question: where is the next bottleneck, the next margin pool, the next underpriced beneficiary?
Jordi is not telling investors to abandon AI. He is telling them to stop treating AI as a single ticker basket.
The bubble analogy is too lazy
The dot-com comparison is useful only if it is used carefully. Jordi’s point is that even in 2000, the sustained bear market did not become obvious merely because valuations were high. The macro had to break. Claims had to rise. ISM had to weaken. Earnings had to roll. Profit margins had to come under pressure. Credit had to start telling the truth.
That is the cluster he is watching.
Right now, the evidence does not support a clean replay of the old bust template. Semiconductor sales, corporate profit margins, and forward earnings are not behaving like a system that has already rolled over. Valuation can create air pockets. Crowding can create violent corrections. But a durable bear market usually needs more than a scary chart and a good historical analogy.
The practical takeaway is simple: do not confuse a valuation reset in the most crowded leaders with proof that the economic cycle has broken.
From labor versus capital to compute versus energy
The deeper regime-shift argument is where Jordi’s worldview becomes most important. He is not treating AI as another software cycle. He is treating it as a change in the production function.
The old economy was organized around labor and capital. Companies hired people, leased offices, borrowed money, managed payroll, and scaled through human systems. The new AI economy increasingly consumes compute, memory, electricity, and proprietary data. The input is electrons. The output is tokens, decisions, code, analysis, discovery, and automation.
That changes what matters at the margin.
If the bottleneck is no longer simply labor, then the next scarcity is not captured by the old playbook. Power matters. Cooling matters. Optical networking matters. Edge compute matters. Specialized models matter. Application-layer ownership matters. Healthcare data and pharma compute matter. The businesses that understand this shift will not look like the last cycle’s software winners, even if the market keeps trying to value them with last cycle’s categories.
This is the line Jordi keeps drawing: regime shift, not replay.
Memory was the first fireworks. Applications are the next test.
Jordi was explicit that memory may have already delivered the easier part of the move. That is not the same as being bearish on every company involved in AI infrastructure. It means the asymmetric opportunity has changed.
The early agentic AI trade rewarded the most obvious scarcity: memory, semiconductors, and infrastructure. But once the crowd is positioned there, the market moves on to selectivity. Investors have to ask where recursive improvement shows up in revenue, productivity, and business design.
That is why Jordi keeps returning to the application layer. He highlighted healthcare and Eli Lilly because the next phase of AI is not just generic models trained on the internet. It is specialized models trained on valuable private data, aimed at difficult real-world problems. In that framing, pharma and longevity are not defensive side notes. They are part of the same AI regime shift.
The phrase “human software” sounds strange until you take it seriously. If AI accelerates biological discovery, clinical workflows, molecule design, and patient-specific insight, then the application layer may be where the market eventually stops talking about AI capex as a cost center and starts seeing it as an economic engine.
Energy is the hedge inside the AI story
The AI thesis is not weightless. It runs on electricity. It is built with copper, optics, chips, substations, cooling systems, gas turbines, batteries, and power contracts. That is why Jordi’s energy comments belong in the market report, not in a separate geopolitical footnote.
He was buying Chevron and Exxon, not because he sees them as the same kind of upside vehicle as the more explosive AI names, but because energy is now part of the portfolio constraint. If oil and shipping risks do not normalize quickly, the inflation backdrop gets harder. If rates remain pressured while AI leaders are correcting, the Fed’s policy room narrows. If the Strait of Hormuz or broader supply-risk premium stays elevated, the market cannot simply pretend the old disinflation script is back.
Energy is not the opposite of AI. In this cycle, energy is one of AI’s limiting reagents.
Positioning now matters as much as narrative
The most dangerous market habit is to keep arguing the theme after the trade has changed. Jordi’s framework this week is useful because it separates the structural thesis from the positioning setup.
Structurally, AI is still early. Agentic systems, specialized models, optical networking, edge devices, tokenization, and application-layer automation are not finished stories. They are just beginning.
Tactically, the easy chase is over. Investors who arrived after the October skepticism flipped into consensus are now playing a harder game. The reward is no longer in owning everything with the right label. It is in knowing which part of the stack has already discounted perfection and which part is just starting to be repriced.
That is why the next few months can be choppy without being fatal. Digestion is not death. It is the market forcing a better distinction between theme, price, and timing.
Bottom line
Jordi’s latest message is not “sell AI.” It is “stop being lazy about AI.”
The market just reminded investors that a regime shift does not move in a straight line. The first fireworks phase rewarded broad exposure to the obvious scarcity. The next phase will reward selectivity: application layers, healthcare, power, optical infrastructure, energy hedges, and companies that turn AI from a capex story into a production system.
A real bear market would require a broader break in macro data, credit, earnings, and margins. That is not the tape Jordi sees yet. What he sees is a rotation after an enormous run, a crowd forced to sober up, and an AI cycle moving from discovery to digestion.
The distinction matters. In the old cycle, investors could wait for labor, credit, and rate signals to explain the economy. In the new one, the market is already asking a different question: who controls the scarce inputs when intelligence becomes industrial output?
That is the report this week. The fireworks are over. The cycle is not.