Most importantly, I’m upgrading all existing subscribers to the paid tier for the next 12 months.

Here’s what you should know:

We’re now sending once per week - every Monday morning. I’ll write a detailed, weekly market commentary that’ll make sure you’re completely up to speed on the previous week’s action, and you’re overprepared for the week to come.

This will always be free.

Behind the paywall (which you now have for free), readers will get some of the deepest analysis of the AI capex spend boom you’ll find anywhere, with a sharp focus on the companies that the hyperscalers are dumping hundreds of billions of dollars into.

Here’s what that content structure will look like:

-Owning the AI Supercycle: new beefy section of the newsletter detailing AI beneficiaries
-BRRR AI Rankings: Monthly updated rankings of the stocks most likely to 5x, 2x, +35% over the next 12 months. Lives on thebrrr.com
-AI Research Reports: We’ve compiled reports for 24 stocks set to benefit from the AI boom and will continue to add new ones. Lives on thebrrr.com
-Momentum Scanners: Product that programmatically scans technicals of hundreds of stocks and surfaces the best setups playing out. Lives on thebrrr.com

I’ve done a lot of work to thebrrr.com - you’ll just need to enter the email address you received this email from on the homepage, and you’ll have full access behind the paywall.

PS - I just gifted the entire 10k+ list premium subscriptions all at once, but beehiiv is applying the gifts very slowly. It’s about ½ way through as I send this.

The real story last week was not that CPI suddenly got benign or that the Fed got room to cut. It was that the market spent the week repricing war risk.

The headline numbers looked great on the surface. The S&P 500 rose 3.6% for the week to 6,816.89, its best week since November. The Nasdaq climbed 4.7% to 22,902.89. The Dow gained 3.0% to 47,916.57. By Friday, the VIX had fallen back to 19.23, below 20 again. That looks like a clean risk-on turn.

The move was driven first by ceasefire relief. On Tuesday alone, after the U.S.-Iran ceasefire headlines, the Dow jumped 2.85%, the S&P 500 rose 2.51%, and the Nasdaq rallied 2.80%. The S&P moved back above its 200-day moving average for the first time since mid-March.

Breadth was powerful. NYSE advancers beat decliners by 5.67 to 1, Nasdaq advancers beat decliners by 3.05 to 1, and the S&P logged 21 new 52-week highs versus just 5 new lows. This was not a random bounce. It was a broad relief rally in a market that had been leaning hard into war-premium pricing.

Oil told the story even more clearly. WTI Oil settled at $96.57 on Friday, down 13.4% for the week, its biggest weekly decline since April 2020. Once the market got a whiff of de-escalation, the energy shock started to come out of the tape. Energy stocks got hit immediately. On the ceasefire day, the S&P 500 energy sector fell 3.7%. Exxon and Chevron each dropped more than 5%.

The mirror image was just as important. As oil broke lower, the market piled into fuel-sensitive cyclicals. Commercial airlines rose 5.7% on the day, travel and leisure gained 5.2%, homebuilders rose 4.9%, and chips rallied 6.3%. The Russell 2000 outperformed. In other words, this was a classic “war premium unwinds, domestic cyclicals breathe again” move.

One other piece of the rally deserves a mention because it tells you exactly where this market still wants to hide. Alongside the ceasefire relief move, money ripped back into the AI trade. CoreWeave was the clearest example.

The stock surged as investors digested its new Anthropic cloud deal, just one day after Meta expanded its CoreWeave agreement to roughly $21 billion through 2032, reinforcing the idea that frontier labs and hyperscalers are still scrambling to lock up capacity.

Broadcom rallied too, for the same reason: the market continues to reward the companies sitting closest to compute scarcity, custom silicon, and the AI capex boom.

That matters because even in a week dominated by oil, geopolitics, and CPI, traders still snapped back into the one secular growth story they trust most. In other words, the macro panic eased, and investors immediately went back to chasing AI infrastructure winners.

The deeper question, and one we get into below, is whether that move was just a squeeze or another sign that the fight for compute, power, and capacity is only getting more intense.

But this is where the interpretation matters. The bounce was real. The resolution was not. By Friday, the market was being forced to confront the fact that physical normalization still lagged the optimism in asset prices. Shipping traffic through the Strait of Hormuz was reportedly still running at less than 10% of its historical average. So markets were effectively pricing ceasefire optimism before the underlying logistics picture was fully repaired.

That is why the CPI report matters, but only in the right order. March CPI rose 0.9% month over month and 3.3% year over year. Core CPI rose just 0.2% month over month and 2.6% year over year. If you only looked at core, you could talk yourself into a cleaner disinflation story. But the composition told a much messier truth.

Energy rose 10.9% month over month. Gasoline surged 21.2%, accounting for nearly three quarters of the monthly headline increase, according to the BLS. That was the largest monthly gasoline increase since the series began in 1967. Fuel oil jumped 30.7%. Shelter still rose 0.3% month over month and 3.0% year over year.

So no, the CPI report did not “save” the market. What it did was show that the inflation effect of the oil shock is already arriving in the hard data even as some of the broader underlying inflation measures remain calmer. That is a very different story. It points less toward a Fed pivot and more toward a Fed that is likely stuck on hold longer while it waits to see whether the energy spike proves temporary.

The bond market got that message. The 10-year Treasury yield finished at 4.317% on Friday, up 2.4 basis points on the day, even as equities still logged their best week in months. That is not the signature of a market that thinks the macro problem is gone. It is the signature of a market trying to balance relief on geopolitics against the reality that inflation has re-accelerated on the headline and the Fed cannot declare victory.

Household sentiment told a similarly uneasy story. The University of Michigan’s preliminary April consumer sentiment reading fell to 47.6 from 53.3 in March, a 10.7% monthly drop and an 8.8% decline from a year earlier.

One-year inflation expectations jumped to 4.8% from 3.8%, while long-run expectations rose to 3.4% from 3.2%. Even with markets bouncing, the real economy was still feeling the gas-pump version of stagflation-lite.

That leaves us with a much cleaner framework for the upcoming week. Stocks had their best week since November because the market was unwinding a war premium, not because macro suddenly turned easy.

The ceasefire mattered because it crushed oil and volatility. But by the end of the week, traders were already backing away from the idea that this was a full all-clear. The right read is not “cool CPI, risk-on, move along.” The right read is that the market got a de-escalation rally while the macro backdrop stayed messy.

The most important AI story of the week was not another generic “AI spending remains strong” update. It was that the trade is becoming less about model theater and more about exponential demand for capacity, power, custom silicon, and financing.

We’re covering some of our highest-conviction names in this space for this week’s breakdown.

Start with CoreWeave, because yes, it absolutely won the week soaring 25%. Reuters reported that CoreWeave struck a new multi-year cloud capacity deal with Anthropic, with capacity coming online later in 2026.

Terms were not disclosed, but the context is what matters. CoreWeave had already signed an $11.9 billion deal with OpenAI last year, announced an expanded $21 billion Meta agreement on April 9 that runs through December 2032, and had previously placed an initial $6.3 billion Nvidia order in September 2025. This is no longer a one-customer curiosity. It is a real capacity platform being validated by multiple frontier buyers.

The Meta number matters on its own. Reuters said the new $21 billion deal is in addition to a similar $14.2 billion agreement signed in September 2025. It also gives Meta early access to Nvidia’s Vera Rubin chips, which Reuters described as roughly twice as fast as the current-generation Blackwell platform.

CoreWeave’s response to that demand tells you how extreme the infrastructure buildout has become. The company said it plans up to $35 billion of capex in 2026, versus $14.9 billion in 2025. It also disclosed plans to sell $1.25 billion of bonds and $3 billion of convertible bonds. This is utility-scale capex, financed like an infrastructure business, wrapped in an AI equity story.

That is why the deeper story is not merely that CoreWeave won the week. It is that the independent compute merchant model is getting harder to dismiss. The market used to treat neoclouds as a fragile intermediary layer between hyperscalers and chip vendors.

But if OpenAI, Meta, and now Anthropic are all willing to lock up capacity there, the more serious question becomes whether CoreWeave is evolving into a durable toll road for frontier compute, or just a temporary release valve until the hyperscalers and labs internalize more of the stack themselves. We’re betting on the former.

Even so, the sharper angle may be Anthropic. Reuters reported that Anthropic’s revenue run rate surpassed $30 billion in early April, up from roughly $9 billion at the end of 2025.

That is a staggering acceleration in a very short period of time. More importantly, Anthropic is not just booking revenue. It is translating that revenue into infrastructure commitments. Reuters also reported that Anthropic said a new Broadcom-Google arrangement builds on its commitment to invest $50 billion in strengthening U.S. computing infrastructure.

Anthopic also announced the existence of a new model called Mythos, and it was not just another benchmark flex or incremental model release.

The model was reportedly powerful enough to uncover vulnerabilities across multiple core layers of internet infrastructure, which is exactly the kind of capability jump that forces people to rethink what these systems are actually becoming.

When a frontier model can do that, it stops being a parlor trick and starts looking like a genuine force multiplier for both cyber offense and cyber defense. That is why the market paid attention. This was a reminder that frontier AI is not only getting better at writing text or code, it is getting better at probing the real plumbing of the digital world.

And once that becomes credible, the implications spread quickly, from national security to enterprise software to the valuation of any company whose moat depends on humans staying comfortably ahead of the machines.

Elsewhere, Broadcom signed a long-term deal with Google to develop and supply future generations of Google custom AI chips and other components through 2031. Broadcom also signed a deal with Anthropic to provide roughly 3.5 gigawatts of AI computing capacity based on Google processors, starting in 2027.

Broadcom’s CEO had previously said the company expects to deliver 1 gigawatt of TPU capacity for Anthropic in 2026, rising to 3 gigawatts in 2027. That is a huge shift in the AI trade. Labs are no longer just buying generic GPU exposure. They are actively diversifying their compute stack, pushing into TPUs, custom silicon, and alternative capacity paths.

Broadcom looks increasingly like a picks-and-shovels winner in the post-Nvidia architecture. The company projected AI chip revenue in excess of $100 billion in 2027. It reported first-quarter AI revenue of $8.4 billion, up 106% year over year, on total quarterly revenue of $19.31 billion. It then guided second-quarter revenue to about $22.0 billion, including $10.7 billion of AI chip revenue. That is what a market shift looks like when custom silicon stops being a side show and starts becoming central to the buildout.

And none of this sits in a vacuum. Bridgewater estimated that Alphabet, Amazon, Meta, and Microsoft would collectively invest about $650 billion in AI infrastructure in 2026, up from $410 billion in 2025.

Reuters separately reported a planned $600 billion-plus AI spending surge by big tech in 2026, including Amazon at $200 billion. The scale here matters because it changes the bottleneck. The question is no longer simply whether demand exists. The question is whether the industry can finance, power, build, and network the capacity fast enough.

AI is no longer just a model story. It is a contracted-capacity story. Anthropic’s revenue is exploding from roughly $9 billion to above $30 billion run rate in one quarter, and that revenue is being translated directly into gigawatts, TPU commitments, own-chip exploration, and new cloud dependencies.

CoreWeave is stacking customer commitments across OpenAI, Meta, and Anthropic while trying to finance a $35 billion capex program. Broadcom is turning custom silicon and rack-level integration into a multi-year growth engine. The real trade is shifting from chatbot hype toward infrastructure scarcity, inference economics, and financing capacity.

If there is a risk in that thesis, it is that the financing side can become the story as quickly as the technology side. Utility-scale capex, debt issuance, power constraints, and customer concentration are not side notes. They are part of the valuation framework now. Reuters noted that Microsoft still accounted for about 67% of CoreWeave’s 2025 revenue. That concentration may improve at the margin as Meta and Anthropic deepen their ties, but it also shows how fast this market is being built around a small handful of counterparties.

The takeaway is simple: yes, CoreWeave won the week, but the deeper and more investable point is that frontier labs are locking up and diversifying compute so aggressively that the next phase of the AI trade will be won by whoever controls power, silicon optionality, networked capacity, and the balance sheet to fund it.

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.