Jordi Visser Sunday Market Report: The Edge Awakens, and the Physical AI Trade Is Still in the First Inning
Source: Jordi Visser’s latest personal YouTube video, “The Edge Awakens: Why the Physical AI Upgrade Is Still in the First Inning,” published April 26, 2026.
Main takeoff points
- Jordi’s message this week is that the AI trade has moved out of the cloud and into the physical world. The next constraint is not ideas. It is semiconductors, power, optical fiber, cooling, freight capacity, chemicals, and the hardware needed to push AI to the edge.
- The market is still too tempted to treat this as a familiar software cycle. Jordi’s frame is the opposite: software is the funding source and the vulnerable business model; the physical stack is where scarcity and pricing power are migrating.
- The industrial data are not confirming a standard slowdown story. Freight, capital-goods imports, machinery, chemicals, semis, and power demand all point to a capex boom that is being driven by AI paranoia across governments and corporations.
- Inflation risk is not only an oil story. The second-order problem is bottlenecks: chips, diesel, fertilizer, petrochemicals, power components, and grid capacity. Jordi expects CPI pressure to return and policy to stay constrained.
- This is a 1970s-style market in one important sense: parabolic winners can still suffer violent drawdowns. A 20% or 30% decline inside the right theme does not automatically invalidate the theme.
- Private credit and software remain the fragile underside. If seat-based software revenue gets hit while lenders are exposed to those businesses, the credit cycle can keep deleveraging even while the industrial side of the economy stays hot.
- Bitcoin is not the whole report, but Jordi’s setup is clear: if inflation pushes real yields negative while software breaks correlation with the old growth basket, Bitcoin becomes a cleaner macro asset than many institutions are prepared for.
Weekly market report
Jordi Visser’s latest market note is not a week-in-review. It is a regime note.
The headline is not that AI is still important. Everyone says that now. The useful distinction is where the AI trade has migrated. Jordi’s argument is that investors are still too anchored to the cloud era—software margins, hyperscaler leadership, familiar mega-cap concentration, and a neat digital-growth playbook. But the next phase is not neat. It is physical. It is semi-heavy, power-heavy, fiber-heavy, cooling-heavy, transport-heavy, and increasingly edge-heavy.
That is why the title of the week matters: the edge is awakening. The artificial-intelligence story is leaving the screen and moving into the world of racks, transformers, capacitors, freight lanes, gas turbines, packaging, batteries, CPUs, memory, optical transceivers, and eventually robots. If the last cycle rewarded companies that could scale software without much marginal cost, this cycle rewards the companies that can relieve bottlenecks in a world where demand is outrunning physical supply.
Jordi keeps coming back to Jensen Huang’s physical-world upgrade frame because it changes the scale of the problem. A $90 trillion upgrade of the physical economy is not a normal enterprise-software adoption cycle. It is closer to an industrial mobilization. The cloud concentrated the profits for roughly seventeen years. Pre-training concentrated the narrative around GPUs. But the next layer is broader: inference, edge devices, power semiconductors, optical networking, memory, CPUs, and real-world automation.
That is also why the old question—when does software bounce?—is the wrong question. Jordi is not saying every software company goes to zero. He is saying the market is still overpaying for business models built on code at the same time code itself is becoming easier to generate, replace, and compress. In that world, the attractive side is not abundance. It is scarcity.
The industrial economy is not behaving like the recession story
The most important part of the video is Jordi’s insistence that the real economy is giving a different signal from the standard bearish script.
He points to the turn in freight commentary, stronger rail and chemical shipments, truck tonnage at a three-year high, flatbed rejection rates at record levels, and capital-goods imports rising relative to consumer goods. That mix matters. It says the incremental dollar is not going into furniture, housing-cycle goods, or the old consumer discretionary complex. It is going into the physical inputs of an AI buildout: semiconductors, data-center parts, turbines, power equipment, and imported industrial components.
This is the regime-shift point. A lot of investors are still looking for the economy in the same places they looked during the last cycle—housing, autos, consumer goods, software spend. Jordi is looking at machinery, semis, chemicals, energy, materials, optical, and freight. The difference is not cosmetic. It changes what “growth” means.
If AI capex is being driven by the fear of falling behind—by corporate and national-security paranoia—then it is less sensitive to the usual consumer-cycle markers. The hyperscalers, OpenAI, Anthropic, Tesla, SpaceX, and governments are not spending because a discretionary consumer survey improved. They are spending because the strategic cost of not spending looks existential.
That is why Jordi’s line that “AI is the only story” should not be read as hype. It is a prioritization rule. Oil can matter. War can matter. The consumer can matter. But the physical AI buildout is the force reorganizing the market’s supply chains, capital allocation, and sector leadership.
Scarcity is the trade; abundance is the trap
The cleanest Jordi frame this week is scarcity versus abundance.
Ideas are now abundant. Software ideas can be generated, copied, tested, and shipped faster than the market is used to. Model progress compresses time. The useful bottleneck moves away from imagination and toward execution capacity: chips, memory, power, network equipment, cooling, component availability, construction capacity, logistics, and electricity.
That is why he spends so much time on power semis, optical components, memory constraints, PCB materials, copper foil, capacitors, helium, petrochemicals, and grid capacity. Those are not side details. They are the market structure.
The market keeps trying to flatten AI into a single ticker or a narrow Mag 7 story. Jordi’s version is more uncomfortable. The spenders may not be the winners. The hyperscalers can keep funding the buildout while their own returns on invested capital arrive slower than investors expect. At the same time, the companies supplying the bottlenecks can experience violent earnings revisions, pricing power, and parabolic moves.
That is why the report’s practical market lesson is not “buy the same AI trade as last year.” It is to understand who owns the scarce input and who merely has to buy it.
Inflation is the policy constraint, not background noise
Jordi’s inflation point is not just that oil and geopolitics can lift headline CPI. The stronger point is that the bottleneck economy is inflationary by design.
If data-center demand forces more power equipment, if semis require more specialty inputs, if fertilizer and diesel costs rise, if shipping costs reset higher, if power grids are run hot, and if companies are forced to pay up for limited components, then the inflation impulse does not stay neatly contained inside one commodity chart.
He expects CPI pressure to move back above 4% if these conditions hold. The policy implication is more important than the exact number. If inflation expectations rise while the labor market remains stable, the Fed has less room to rescue asset prices. Real yields can go negative again without a clean return to the disinflationary growth regime investors remember.
That is the trap. Markets want the reward of a capex boom and the comfort of an easing cycle at the same time. Jordi is warning that this mix may not allow both.
A 1970s analogy is useful only if it is used carefully. This is not a demand-led replay with the same industrial base, the same demographics, or the same geopolitical structure. But it does rhyme in one crucial respect: bottlenecks can create a market where the secular winners move violently higher while policy stays boxed in and drawdowns remain brutal.
The edge has barely started
The most forward-looking piece of Jordi’s note is the edge.
So far, the AI cycle has been dominated by training, cloud compute, and hyperscaler capex. But Jordi is focused on the next shift: inference and edge deployment. Once AI moves from chatbots and centralized compute into devices, factories, robotics, vehicles, logistics systems, and local decision-making, the bottleneck broadens again.
That is where power semiconductors, CPUs, optical networking, memory, and energy efficiency matter. Nvidia’s reported 800-volt DC architecture work, rising demand for power conversion efficiency, and pressure on optical and semiconductor components all fit the same pattern. The point is not one announcement. The point is that the whole stack tightens when AI leaves the server rack and starts moving toward the physical world.
Jordi also highlights the robotics timeline. If meaningful robotics breakthroughs arrive over the next six to eighteen months, the market will have to price another layer of physical AI demand before the current one is even solved. That is why he calls it the first inning. Not because the charts are cheap. Many are not. But because the earnings base, supply response, and device deployment cycle are still early relative to the size of the transition.
This is the part investors will struggle with most. The right theme can be early and expensive at the same time. Parabolic charts can be correct and dangerous at the same time. Jordi’s warning is that a 30% drawdown in the middle of a secular shortage does not mean the shortage disappeared. It means volatility is part of the regime.
Software and private credit are the weak underside
The report is bullish on the physical AI buildout, but it is not complacent.
Jordi remains negative on software and private credit because the two risks can connect. Seat-based software companies are vulnerable if AI tools reduce headcount, compress workflows, or make legacy platforms less valuable. A company can beat earnings and still fail the market’s new test if it cannot show that AI revenue offsets pressure on the old model.
That pressure then matters for credit. Jordi points to Medallia’s default as an example of the kind of event that can sit below the market’s radar while equity investors focus on the index. The issue is not that one default breaks the system. It is that private credit exposure to software and software-adjacent businesses creates a slow deleveraging channel.
That is why he does not frame this as a clean systemic crash. Too much capital is still flying around. Too many industrial sectors are doing fine. The economy can have an AI-driven industrial boom and a credit/software unwind at the same time.
That is the second-order market. The index can hide it. Theme baskets can reveal it. Gross exposure can make it more violent.
Bitcoin belongs to the endgame, not the headline
Jordi’s Bitcoin view fits the same framework, but it should not be turned into a crypto wrapper around the report.
His setup is macro. If inflation rises while policy cannot tighten aggressively, real yields can turn negative. If software loses its old role as the default growth asset, institutions will have to search for another form of growth and store-of-value exposure. If the dollar regime looks less certain and the physical economy becomes more constrained, Bitcoin can become more interesting precisely because it is not another software company with a seat-based revenue model.
He also notes that Bitcoin has not yet gone parabolic in the same way many AI-infrastructure names have. That relative setup matters. In a market where investors hate buying the end of a parabolic chart, an asset that is still closer to its 2024 level can become a cleaner reallocation candidate if the macro quadrant turns favorable.
The important distinction is that Jordi is not using Bitcoin as a substitute for the AI-infrastructure thesis. He is using it as the later-stage expression of the same regime: inflation pressure, policy constraint, software disruption, and the search for assets that survive the transition.
Bottom line
This week’s Jordi report is a warning against lazy pattern matching.
It is not 2021 again. It is not simply 2007 again. It is not just an oil shock, not just a consumer slowdown, not just a Mag 7 story, and not just a crypto story. It is a physical AI buildout colliding with an economy that does not have enough slack in the right places.
The investment implication is uncomfortable but clear: scarcity matters more than narrative. The market will keep rewarding companies that relieve physical constraints and punishing businesses whose old margins were built on code, inertia, and abundance. But the path will not be smooth. Parabolic markets do not move in straight lines. Bottleneck regimes produce both huge winners and violent air pockets.
Jordi’s strongest point is that the edge has barely started. If he is right, investors are still underestimating how much of the next market cycle will be decided not by who has the best AI demo, but by who can supply the power, chips, components, transport, and infrastructure required to make AI physically real.