Three Forces, One Framework: Navigating War, AI, and the Rotation You Can't See
Published March 8, 2026 | Protocol Wealth LLC
WTI crude hit $91 on Friday — biggest weekly move in futures history going back to 1983. Anthropic's revenue run rate doubled in three months. The median stock in the S&P 500 fell 3% this week while the index barely moved. Geopolitics, AI, and a massive internal rotation are hitting simultaneously, and flat index numbers are hiding all of it.
The Geopolitical Shock: Oil and the Strait of Hormuz
Oil posted its largest weekly gain in the history of the WTI futures contract (since 1983); West Texas Intermediate jumped over 35% to roughly $91 a barrel.¹ Tanker traffic through the Strait of Hormuz dropped more than 80%.² The U.S. government announced a $20 billion reinsurance facility to backstop Gulf shipping; President Trump's "unconditional surrender" rhetoric toward Iran suggests this isn't resolving overnight.
The knock-on effects go further than the pump. Elevated energy prices are an inflation input, a consumer spending headwind, and a potential obstacle to the rate cuts the market has been counting on. Historically, oil spikes of this magnitude have reversed — but "eventually" and "soon" are not the same word.
Our portfolio construction is built for exactly this environment. The framework starts with durable, cash-generative businesses — energy infrastructure, essential grid and power assets, companies that perform when the world gets noisy. We're structurally tilted away from the fragile corners of the market that suffer most when oil stays elevated and uncertainty persists.
The AI Question: Real Technology, Fragile Financial Architecture
Alright. That's the energy picture. Now the harder one.
One camp says AI will rapidly displace most knowledge-economy jobs. The other says it will solve everything. Both are compelling. Both miss something important.
The technology is real. A lot of the money chasing it is not.
Anthropic's revenue run rate went from roughly $10 million in 2022 to more than $19 billion by March 2026, according to Bloomberg reporting following the company's Series G announcement.³ Nvidia posted record results. Broadcom's AI revenue doubled. These aren't projections — they're real numbers.
At the same time, Oracle and OpenAI just abandoned plans to expand their flagship Stargate data center in Texas after financing negotiations broke down and OpenAI's demand forecasts shifted.⁴ OpenAI originally committed $1.4 trillion in infrastructure obligations — a figure recently revised down to roughly $600 billion by 2030 — representing significant revenue expectations on their suppliers' books.⁵ Private credit funds are gating at structural limits. Industrial stocks that rode the AI capex wave are breaking down.
The technology works. The financial bets layered on top of it — fragile, dependent on counterparties that can't fund their commitments — are where the risk lives. Half of what's being called "AI infrastructure" right now is just leverage wearing a hard hat.
We don't know how long the capex unwind takes. Nobody does. The timing is unknowable; capital cycles rarely unwind on a schedule. But we know what we're not doing: chasing the AI buildout on faith, adding leverage into uncertainty, or pretending the index tells you anything useful right now.
The Capex Complexity Problem
A lot of what passes for "AI strategy" in the market right now is a Rube Goldberg machine: spend billions on data centers to train models to generate code to replace workers to justify the spending on data centers. The chain is elaborate. At each link, value leaks out.
We don't invest in the machine. We invest in the electricity that powers it.
Power plants last 40 to 60 years. Chips last 2 to 3. Software moats erode in months. Every investment gets scored on one question: how fast does this business lose its competitive edge? We call it the decay constant — lambda. Nuclear and grid infrastructure? Lambda around 0.02; decades of useful life. AI chips? Closer to 0.25. That gap is the whole thesis. The slower the decay, the larger the position.
Human + AI: The Productivity Story
Researcher Ethan Mollick has been among the clearest voices on this: the future isn't AI replacing humans or humans ignoring AI. It's humans working with AI — and that's where the real productivity gains show up.
A recent Anthropic study confirmed it — AI can theoretically perform many white-collar tasks, but adoption lags capability dramatically. Workflows are embedded and switching costs are real. Integration takes time.
The companies that figure out how to integrate AI into existing platforms — adding value to what they already do well — will thrive. HubSpot just proved this with higher retention and revenue per user after integrating AI features. The companies with shallow moats, seat-based pricing, and discretionary use cases? They're getting disrupted.
We've seen this pattern before. Carlota Perez mapped it across five technological revolutions. The installation phase is always messy — speculative, full of casualties. The deployment phase rewards whatever infrastructure survived the chaos. We're deep in AI's installation phase. The constraint has already shifted from chips to megawatts, and our portfolio reflects that.
The Rotation You Can't See
Most investors are missing it: while the S&P 500 looks flat, there's an aggressive rotation happening underneath.
Breadth this week: roughly 6,500 declining stocks versus 3,100 advancers. Median stock fell nearly 3%. Single-stock dispersion hit the 97th to 99th percentile — individual stocks moving in wildly opposite directions while the index masks all of it. Money is rotating out of last cycle's winners fast and quiet.
The playbook here is straightforward: own businesses that generate real cash, score well on fundamental quality checks, and don't depend on the continuation of a single narrative.
What We're Doing About It
None of this changes how we build portfolios. Headlines get loud; the framework stays the same.
Every investment we hold has to clear a set of quantitative and fundamental checks — cash generation, balance sheet durability, competitive positioning, macro fit. Businesses that score well get full allocations. Businesses that don't get sized down or excluded entirely. The goal is not prediction; it's avoiding fragile businesses and weighting toward durability. Positions are then layered from the most durable assets at the foundation through progressively more cyclical or speculative opportunities at the top.
For clients whose risk profiles align, our current positioning reflects these themes:
Stay grounded in durable infrastructure. Power, grid, and energy assets sit at the foundation of our portfolio because they benefit whether the AI narrative accelerates or stalls. They also hold up in an inflationary, oil-driven environment. Holdings like GRID and nuclear-exposed names carry decay constants measured in decades, not quarters.
Be selective in technology. The best software businesses — deeply embedded, mission-critical, consumption-priced — are not the speculative AI capex plays. Don't throw out the category because of the casualties.
Cash generation matters more than usual right now. Free cash flow yield is the valuation metric that still works as a reliable anchor. The businesses generating real cash at attractive yields are where we're finding value.
Be aware that index-level calm masks significant underlying dispersion. The rotation underneath is one of the most aggressive in recent memory. Passive index exposure means full exposure to both sides — winners and losers — with no ability to tilt toward quality.
The Bottom Line
Look — Iran matters. AI matters. The rotation matters. None of them exist in isolation; the right response to all three is the same: own durable businesses, size them based on quality, don't let headlines dictate strategy.
We're monitoring all of it and will adjust as conditions warrant. The foundation doesn't change because the news is loud. Invest in what lasts. Avoid what decays.
Questions about how any of this maps to your situation — reach out.
¹ WTI closed at $90.90/barrel on March 7, 2026; 35.63% weekly gain per CNBC/Bloomberg. ² Tanker transits dropped from ~24/day average to 4 on March 1 per Vortexa energy intelligence data. ³ Anthropic revenue run rate per Bloomberg, March 3, 2026; 2022 revenue per company disclosures and SaaStr. ⁴ Oracle and OpenAI scrapped Abilene, TX data center expansion per Bloomberg, March 6, 2026. The broader 4.5GW Oracle-OpenAI agreement remains in effect. ⁵ OpenAI's original $1.4T commitment per HSBC and multiple outlets; revised to ~$600B per CNBC, February 20, 2026.
Protocol Wealth LLC is a registered investment adviser (CRD #335298). This material is for informational purposes only and does not constitute investment advice or an offer to buy or sell securities. All investing involves risk, including the possible loss of principal. Past performance does not guarantee future results. Information herein is based on sources believed to be reliable but not guaranteed. This material may contain forward-looking statements that are subject to uncertainty and changes in circumstances.