Last week’s face-ripping rally was a war-premium unwind.

The headline numbers were strong. The S&P 500 finished at 7,126.06, up 4.53% on the week. The Nasdaq closed at 24,468.48, up 6.84%, its best weekly move in months. The Dow ended at 49,447.43, up 3.2%. By Friday, the rally had enough momentum to push the S&P and Nasdaq to their third straight record close, while the Nasdaq notched its 13th consecutive advance, its longest winning streak since 1992.

That is not the tape of a market celebrating softer inflation, an easier Fed or a surprisingly great earnings season. It is the tape of a market aggressively repricing down a near-term oil shock.

Reuters summed up the consensus well earlier in the week. By April 14, the S&P had already climbed back above its pre-war level even though the macro backdrop was clearly worse than it had been in late February.

Oil was still sharply higher. The 10-year Treasury yield was around 4.25%, up from 3.96% on February 27.

Fed rate-cut expectations had been slashed to almost nothing by year-end. In a normal market, that combination should not be sending stocks to record highs.

The only way it makes sense is if investors believe the oil spike is a short-lived disruption rather than the start of a durable inflation regime.

That belief may turn out to be right. But it is still a belief.

Because the hard data did not suddenly get easy last week.

The March CPI report showed exactly the opposite. Headline CPI rose 0.9% month over month, the biggest increase since mid-2022, and 3.3% year over year. Gasoline prices jumped 21.2% in March, the largest increase since the government began consistently tracking the series in 1967, and accounted for nearly three quarters of the monthly CPI increase.

Core CPI was calmer at 0.2% month over month and 2.6% year over year, but that offered less comfort than it first appeared.

Reuters noted that economists expect the second-round effects of the oil shock to show up later, especially through transport, services, and goods costs.

That is the key distinction. The CPI report did not say inflation was solved. It said the first-round energy hit is already here.

Consumer psychology told the same story. The University of Michigan’s early April sentiment reading fell to 47.6, an all-time low, from 53.3 in March.

One-year inflation expectations jumped to 4.8% from 3.8%, and five-year expectations rose to 3.4% from 3.2%. Even if the market chose to celebrate reopening headlines, households were still staring at $4 gasoline and acting like they knew exactly what that meant.

So the market and the economy are still telling slightly different stories.

The market is saying: the worst-case energy shock is fading, corporate earnings still look solid, and once the ceasefire firms up we can go back to the old playbook.

The economic data is saying: even if the war premium comes out of oil, the inflation hit has already started to bleed through, and the Fed is in no position to declare victory.

The problem is that the underlying plumbing is still not normal.

Reuters reported on Friday that traffic through the Strait of Hormuz had fallen to less than 10% of its historical average after the war began, even as markets cheered the reopening headlines. In other words, asset prices were already discounting normalization faster than the physical system had actually normalized.

That does not mean the rally was fake. It means the rally was conditional.

If the de-escalation holds and oil keeps leaking lower, this market can keep climbing. The earnings backdrop is still supportive, and investors clearly want to buy every off-ramp they are given.

But if energy disruptions linger, shipping remains constrained, or the ceasefire proves more fragile than the tape currently assumes, then last week starts to look less like the start of a clean new leg higher and more like a very powerful unwind of an overcrowded fear trade.

That is the framework I think matters most going into this week.

While wartime geopolitics captures the imagination of retail traders and hedge funds alike, one of the more important tells was where money moved as the panic eased.

While the market rally saw wide breadth, the biggest outperformers were in tech, semis, and AI-linked infrastructure. This reinforces the idea that when geopolitical stress cools, investors still want exposure to the companies closest to compute scarcity, custom silicon, and the AI capex buildout.

Anthropic was a big part of that signal. Reuters reported this month that the company’s run-rate revenue has surged to more than $30 billion, up from roughly $9 billion at the end of 2025, while it also locked in new compute through CoreWeave and a separate Broadcom-Google arrangement that gives it access to about 3.5 gigawatts of AI capacity starting in 2027.

AI demand is still far outrunning infrastructure. Let’s invest in the companies building the infrastructure.

The most important AI story last week was not a flashy model launch or another vague “AI demand remains strong” headline. It was that the market got fresh proof, in just a few days, that the real bottlenecks are still compute, power, and custom silicon.

Start with the two names you really cannot avoid: CoreWeave and Bloom Energy.

Reuters reported on April 15 that Jane Street committed roughly $6 billion for CoreWeave cloud services and made a $1 billion equity investment in the company at $109 per share.

That came just days after Reuters reported that CoreWeave had struck a new multi-year capacity deal with Anthropic, and just after Meta expanded its CoreWeave agreement to $21 billion. In other words, this was not one nice headline. It was a sequence of large, real counterparties all trying to lock up scarce AI capacity at once.

That matters because it pushes CoreWeave further away from the old “speculative neocloud” bucket and closer to something the market may have to treat as a real infrastructure merchant.

The company plans to spend $30 billion to $35 billion in capex this year, more than double 2025 levels, to buy Nvidia chips, build more data center capacity, and secure power. That is not normal software-company behavior. That is grid-scale, financing-heavy, utility-style expansion, except wrapped inside one of the hottest themes in the market.

Then there is Bloom Energy, which may be the cleaner expression of the next bottleneck. Reuters reported on April 13 that Bloom will supply Oracle with up to 2.8 gigawatts of fuel-cell capacity, with 1.2 gigawatts already contracted and deployment already underway. That is a huge number. More importantly, it tells you exactly where the trade is going.

The problem is no longer just who gets the best GPU. The problem is who can actually get enough electricity to turn those chips on fast enough.

That is the key shift. AI is no longer just a semiconductor story. It is becoming a power procurement story. If hyperscalers and cloud providers cannot wait years for traditional grid upgrades, then companies that can deliver faster on-site or near-site power start looking less like side beneficiaries and more like critical enablers of the whole buildout.

The next layer of the story is that the beneficiary set is broadening beyond Nvidia into the custom chip and connectivity stack.

Reuters reported on April 14 that Meta extended its custom AI chip deal with Broadcom through 2029, including an initial commitment of more than one gigawatt of computing capacity.

Broadcom will also provide Ethernet networking technology for Meta’s expanding AI clusters. That is important because it reinforces the idea that hyperscalers are not just buying more compute, they are trying to optimize the full stack, from custom silicon to cluster interconnects, so they can reduce reliance on off-the-shelf GPUs where possible and improve inference economics over time.

Then on April 19, Reuters reported that Google is in talks with Marvell to build two new AI chips, including a memory processing unit designed to work with Google’s TPU systems and a new chip specifically aimed at running AI models more efficiently.

That one is earlier-stage than the Bloom or CoreWeave headlines, but the direction is unmistakable. The hyperscalers are still spending aggressively, but they are also getting more surgical. The money is spreading into the parts of the stack that improve efficiency, reduce cost per token, and help squeeze more output from constrained infrastructure.

If you zoom out, the system-level confirmation was just as strong. Reuters reported on April 16 that TSMC raised its annual revenue forecast, said AI demand remains “extremely robust,” and guided capex to the high end of $52 billion to $56 billion. A day earlier, Reuters reported that ASML lifted its 2026 revenue outlook to 36 billion to 40 billion euros, with management saying chip demand is outpacing supply and customers are accelerating 2026 and beyond capacity plans.

That is what makes the tape so interesting right here. This is no longer a story about one superstar stock sucking all the oxygen out of the room. It is starting to look more like a full industrial buildout, where the winners show up at different layers of the constraint stack.

If you want the cleanest framework, it is this:

CoreWeave is a bet on compute scarcity being real enough that customers will sign enormous, financing-heavy capacity deals to secure it.,

Bloom Energy is a bet that AI power demand is now so urgent that faster-to-deploy electricity becomes a high-value product in its own right.,

Broadcom and Marvell are bets that hyperscalers will keep pushing deeper into custom silicon and networking to reduce friction and improve economics.,

TSMC and ASML are the proof that this is not narrative froth. The underlying capex engine is still running hot.,

The big takeaway is that the market is beginning to reward the companies closest to the actual choke points. Not the loudest AI branding. Not the prettiest demo. The choke points.

That is where the real money in this cycle is still being made.

Late last year, a Klimt sold for the highest price ever paid for modern art at auction.

An outlier sure, but it wasn't a fluke. U.S. auction sales grew 23.1% in 2025. The $1-5mm segment even grew 40.8% YoY.

Now, the S&P, teetering on all time highs, just posted its worst quarter since 2022, oil was up 94% (briefly), and Moody's puts recession odds at 48.6%.

So, that's where it gets interesting.

Each environment is unique, but after dot-com, post war and contemporary art grew about 24% annually for a decade. After 2008, about 11% for 12 years.

It’s also had near-zero correlation with the S&P 500 since ‘95.*

Now, Masterworks lets you invest in shares of artworks featuring legends like Banksy, Basquiat, and Picasso.

Shares can sell quickly, but my subscribers can skip the waitlist:

Skip waitlist

^{*Investing involves risk. Past performance is not indicative of future returns. See important Reg A disclosures at masterworks.com/cd.}

Got feedback? Follow the writer on Twitter @frank_locascio and send a message.

The BRRR is meant for informational purposes only. It is not investment advice. Please consult with your investment, tax, or legal advisor before making any investment decisions.