Saturday, May 2, 2026

Your Saturday morning market coffee. The week three cloud giants committed $515 billion in AI capex in a single evening, Brent oil briefly hit $126 (a four-year high), Powell said goodbye from the podium, and the S&P 500 still closed at a fresh all-time high.

Week in Review

The Numbers

S&P 500 closed at a fresh all-time high on Friday. The week capped what sources described as the S&P's fifth consecutive weekly gain and its best month in years. The Dow lagged: GM beat massively but fell on EV cost drag; Ford's beat was nearly all a one-time $1.3B tariff refund.

The Story Arc

Tuesday opened with WTI near $99-100 as Iran's Hormuz blockade held. Iran's foreign minister offered to reopen the Strait — and agreed to defer nuclear talks to a later stage — in exchange for the US lifting its naval blockade and ending the war. Oil rose 3% anyway to $111.26 Brent. Traders didn't believe the offer. Wednesday was quiet: GM reported a +41% EPS beat but fell 2.95% on EV cost drag; Starbucks beat on its turnaround thesis; the dollar held near 98.60. Wednesday night was the moment the entire quarter had been building toward. MSFT, META, and GOOGL all reported after the close. Cloud growth was uniform and historic — Azure +40% YoY, Google Cloud +63% YoY — but the buyside sold META's capex hike to $125-145B and MSFT's margin guide with 7-9% and ~3% afterhours drops. GOOGL was the exception: cloud demolished estimates, stock +6.5% AH. Thursday delivered a double-header no one fully digested: the FOMC held rates at 3.50-3.75% with a historically hawkish 8-4 dissent vote (most since October 1992), and Powell confirmed it was his final press conference as Fed Chair (term expires May 15). Amazon simultaneously reported a 70% EPS beat but fell 3% AH on $200B capex guidance. Trump confirmed the naval blockade of Iranian ports continues until a nuclear deal is reached — Brent hit $126.41, its highest in four years. Q1 GDP printed at +2.0% (below the +2.3% estimate) with an embedded PCE price index of 4.5%. Friday was the exhale. Apple's fiscal Q2 delivered a March-quarter iPhone record ($56.99B, +22% YoY), a Services all-time high ($30.98B), and a new $100B buyback — AAPL +3% AH. XOM and CVX both beat adjusted EPS. Japan intervened in FX markets with approximately $34.5 billion to break the yen's slide from 160.7/USD. Markets closed at records.

The week's narrative tension: stagflation on the tape (2.0% GDP, 4.5% PCE price index, 3.2% core PCE), oil at $126 intraday, a hawkish lame-duck Fed, and yet indexes at ATHs. The earnings tape — specifically, three cloud platforms reporting their fastest growth in years — overwhelmed every macro concern. The S&P tells you which narrative won.

Biggest Movers

Winners

Losers

Market Scoreboard

Weekly Index Performance

Commodities & Rates

Oil's intraday spike to Brent $126.41 (a 4-year high) on Thursday is the week's defining chart. But gold's failure to rally alongside oil — again — says the same thing it said last week: the buyside does not believe this is uncontrolled escalation. The 10Y yield rising +9bps during an oil shock is the textbook "inflation scare" signal, and yet equities still hit all-time highs. The cloud earnings tape overwhelmed every macro headwind.

Earnings Recap

Mag-7 and Semis

Oil Majors (May 1 pre-market)

Consumer, Autos & Financials

The Lesson

Cloud beat uniformly. Capex punished selectively. Gold refused oil. The three hyperscalers reporting Wednesday night collectively posted their fastest cloud growth in years. The market rewarded GOOGL — the only one of the three that beat without a capex shock — and punished META and AMZN for committing $200B and $145B respectively to AI infrastructure. MSFT was punished for margin compression. These are forward-looking cost concerns, not backward-looking performance misses. The cloud businesses are accelerating; the debate is whether $515B in combined FY2026 capex generates commensurate returns.

Ford's 247% EPS beat is the cautionary tale of the week: the entire beat was a $1.3B tariff refund from a Supreme Court ruling on Trump tariffs. Strip that out and Ford missed. The market partly figured this out. Mastercard raised its full-year guide but fell because management disclosed April cross-border volume was already softening — a rare case of a company punished for current-quarter honesty in a single earnings call.

The oil-major paradox: XOM and CVX both beat adjusted EPS, yet reported net income down 36-45% year-over-year despite Brent hitting $126. The culprit is derivative timing charges and Hormuz-related production disruptions. You can beat the quarter and still have the underlying business get worse.

Geopolitical Update

Iran-Hormuz: The Wildest Oil Week in Four Years

• Tue Apr 28: Iran's FM Araghchi proposes reopening Strait — and agreeing to defer nuclear talks to a later stage — in exchange for the US lifting its naval blockade and ending the war. Trump rejects. Brent rises 3% to $111.26 — traders don't believe the offer.
• Wed Apr 29: Trump confirms naval blockade of Iranian ports continues until a nuclear deal is reached. Brent spikes to $118.03 (+6% in one session).
• Wed Apr 29 – Thu Apr 30: Oil climbed for 7 consecutive days through Wednesday Apr 29. Brent briefly hits $126.41 on Thu Apr 30 (4-year high). WTI reached an intraday high of approximately $110.90 before settling at $105.07 on Thu Apr 30.
• Fri May 1: Iran reportedly sends new peace proposal through Pakistan. Oil pulls back; Brent settles ~$111.
• Parallel shock: UAE announces it is quitting OPEC effective May 1, ending a membership of approximately 59 years (since 1967). The most recent prior exit was Qatar in January 2019. UAE is OPEC's third-largest producer and wants freedom to ramp production to 5 million bpd by 2027 (vs. its 3.2 million bpd OPEC quota). Oil rose despite the exit because the Iran war premium dominated.

FOMC Transition

The Fed held at 3.50-3.75% but the vote was 8-4 — the most dissents since October 1992. Powell confirmed April 29 was his final press conference as Fed Chair (term expires May 15). He confirmed he will remain on the Fed's Board of Governors. Kevin Warsh, Trump's nominated successor, has advanced out of the Senate Banking Committee.

Macro Data: Stagflation Whispers

• Q1 2026 GDP advance estimate (Apr 30): +2.0% annualized vs. +2.3% estimate — a miss. But the PCE price index (embedded in the GDP report) came in at 4.5%, the highest embedded inflation reading in a GDP report since Q3 2022 (approximately 3.5 years). Soft growth + hot prices = the market's least favorite combination.
• March PCE (Apr 30): Headline YoY +3.5% (vs. +3.6% est); Core YoY +3.2%. Both above the Fed's 2% target. Monthly headline was +0.7%, driven by surging gasoline prices tied to the Iran war.
• April NFP: Due May 8 (not this week). The first post-FOMC employment print will set the table for the June rate-cut debate.

Berkshire Hathaway Annual Meeting (Today, May 2)

Greg Abel's first annual meeting as CEO — Warren Buffett, 95, sits in the audience as Chairman of the Board, not at the CEO podium, for the first time in decades. Cash pile: $373.3 billion (record). Buybacks resumed in March 2026. Abel's pre-meeting letter signaled "faster trigger on bolt-on acquisitions in energy and industrials." Q1 2026 earnings released this morning at ~7am CT; analyst consensus was $4.82 EPS.

Other Developments

• Japan FX intervention: ~$34.5 billion (~5.48 trillion yen) deployed to support the yen (which had weakened to 160.7/USD). DXY fell to approximately 98.2 on the week, its lowest level since late February, a decline of roughly 0.4% from the Monday open of ~98.6.
• Hegseth Congressional testimony (Apr 29): Defense Secretary's first appearance before Congress since the Iran war began (~6-hour hearing). Iran war cost to date: $25 billion. FY2027 Pentagon budget request: $1.5 trillion record.
• US-China: Trump-Xi summit expected in May. Geneva tariff truce extended through November 2026 (US tariffs on China: 30%; China on US: 10%). No specific new round of bilateral talks confirmed this week.

Strategy Scorecard

Winners

Mixed

Losers

MVP of the Week

ai_infrastructure_layer. When Azure grows 40%, Google Cloud grows 63%, and AWS grows 28% in the same quarter — all three accelerating simultaneously for the first time in years — the cleanest trade is not the hyperscalers themselves (who get punished for spending) but the picks-and-shovels layer below them. QCOM was the cleanest single-name expression: +9-16% on confirmation of a hyperscaler chip deal that shipped ahead of schedule, plus a China handset recovery signal. The strategy thesis — own the infrastructure layer, not just the AI application layer — was validated in the most concrete way possible this week. Honorable mention: core_satellite — five straight winning weeks, ATH close, zero drama. That's the point.

Next Week Preview: May 4 – May 8

Economic Calendar

Political / Central Bank

Geopolitical Watchlist

• Iran/Hormuz: Did the Pakistan-mediated peace proposal produce a response? If the Strait reopens, oil retraces $15-20 fast — and the entire warflation_hedge thesis needs reassessment.
• Trump-Xi summit: May dates still unconfirmed. Any announcement would move CNY, EM equities, and tariff-exposed names immediately.
• UAE OPEC exit: First full week since UAE's departure (effective May 1). OPEC market share now below 30% for the first time.
• META/AMZN capex digestion: After AH drops of 7-9% and 3%, do these names recover as the capex-as-moat narrative takes hold, or do they trade on free-cash-flow compression?

Monday Setup

Scenario A: Iran de-escalates (~35% probability)

Pakistan-mediated proposal produces a framework. Strait partially reopens. Oil retraces to $90-95. Inflation fears fade, real rates ease, META/AMZN recover from their AH drops, tech broadly leads. Action: add to ai_mega_ecosystem on the dip; trim warflation_hedge energy sleeve partially; watch gold — a re-bid to $4,700+ would confirm the move is real.

Scenario B: Capex anxiety dominates (~30%)

No Iran news either way. Markets digest the $515B capex week — META/AMZN/MSFT continue to underperform; semis and AI infrastructure layer outperform. Yields hold elevated (4.35-4.45%). NFP anticipation keeps risk appetite cautious. Action: rotate within tech toward ai_infrastructure_layer (QCOM, data center cooling, power); maintain core_satellite core.

Scenario C: Hormuz re-escalates (~35%)

Iran rejects the Pakistan proposal; blockade hardens. Brent retests $115-120. The PCE price index at 4.5% + core PCE at 3.2% + oil re-spike = stagflation narrative takes hold. Action: warflation_hedge extends; uranium_renaissance re-bid; defensive rotation toward quality_factor and low_vol_quality.

Position Sizing

• Core (locked strategies): keep at full weight, no changes
• Tactical adds: max 2% per name, only post-signal, only post-NFP
• Cash: hold 15-17% through NFP on May 8
• No directional adds before May 8 — the stagflation/recovery read depends on that print

---

The $515 Billion Week: What Happens When the World's Three Largest Spenders All Pull the Same Trigger at Once

A meditation on the mathematics of the AI infrastructure arms race, why the hyperscalers are being punished for winning, and where the actual returns are going to accrue.

On Wednesday evening, April 29, 2026, three technology executives sat in front of earnings call moderators and committed their companies to spending, in aggregate, somewhere between $515 billion and $535 billion on capital infrastructure in a single fiscal year. Meta raised its 2026 capex guidance to $125-145 billion. Microsoft is tracking toward $190 billion for its fiscal year (roughly $47.5B per quarter). Amazon guided full-year 2026 capex to approximately $200 billion, up from $132 billion in 2025 — a 52% increase.

Let me put $515 billion in context. It exceeds the annual GDP of Norway. It is roughly half the combined annual revenue of every airline in the world in 2024 (which totaled approximately $996 billion). It is more than seven times what the entire US semiconductor industry invested in fab capacity in a single year. It is, to borrow a Keynes phrase, a genuinely staggering number — and it was committed across three earnings calls over roughly six hours.

What they're buying. The vast majority of this capital is going into data center construction, GPU servers (primarily Nvidia H100s and B200s, plus custom silicon from each company's own chip teams), networking, cooling infrastructure, and power. Each dollar of data center investment requires roughly $0.30-0.40 in associated power infrastructure — transformers, substations, backup generation, and increasingly, direct agreements with nuclear and natural gas plants. The rough math on $515B of data center capex implies somewhere in the range of 50-80 gigawatts of new power demand over the three-to-five-year infrastructure buildout. For context: the entire US grid capacity is approximately 1,250–1,300 gigawatts. These three companies are planning to absorb, indirectly, 4-6% of US total generating capacity for AI workloads.

Why the stocks went down. Meta fell 7-9% afterhours. Amazon fell 3%. Microsoft fell ~3%. All three beat on earnings. All three reported accelerating revenue growth. All three have pristine balance sheets. The reason for the selloff is simple: free cash flow. When you raise capex by $30-68 billion without a corresponding guarantee that revenue will absorb it in the same time frame, operating margins compress. Microsoft guided its Q4 operating margin to 44% — down from 46.3% in the quarter just reported. Investors who bought these stocks on 45-50% operating margin assumptions suddenly hold shares in companies with 44% margins trending further down. The selloff is not about disbelieving the AI thesis. It is about the timing mismatch between when the cash goes out and when the revenue comes back.

GOOGL was the exception. Alphabet rose 6.5% afterhours on the same night. The reason is instructive: Google Cloud came in at $20.03 billion versus an $18.05 billion estimate — an 11% beat — with revenue growing 63% year-over-year. Google's Cloud backlog nearly doubled to $460 billion. But crucially, Google's capex guidance did not shock the market to the same degree as Meta or Amazon. The buyside rewarded the beat while accepting the spend. The implicit message: "We are past the point where we are deploying speculative capex — we have the revenue to justify it." Meta and Amazon have not yet demonstrated that at the same scale. The divergence in afterhours reactions — GOOGL +6.5%, META -7-9%, AMZN -3% — is a real-time verdict from institutional investors on which company has the revenue visibility to justify its spending.

The picks-and-shovels asymmetry. This is where the strategy implication gets interesting. When three of the five largest companies in the world simultaneously commit to record capital expenditure programs, the obvious question is: who actually gets the money? Not the hyperscalers themselves — their stocks go down. The money flows to the supply chain: chip designers, power equipment makers, fiber companies, cooling specialists, and the networking infrastructure layer. Qualcomm's +9-16% this week is a small example of the dynamic. QCOM announced it was shipping data center chips to a "large hyperscaler" ahead of schedule. It did not name the customer. It did not need to. Investors know that a $200B annual capex program at Amazon, a $190B program at Microsoft, and a $145B program at Meta need chips, switches, racks, power convertors, and software-defined networking at a scale that the market has barely begun to price.

The relevant historical parallel is the 1999-2000 telecom capex boom. WorldCom, AT&T, and Sprint collectively spent roughly $120 billion laying fiber optic cable across North America and transatlantic routes. The telecom companies themselves went bankrupt. But the fiber they laid powered every meaningful internet application from 2003 to 2015. The engineers who built Google, YouTube, Facebook, and Netflix inherited an infrastructure surplus they did not pay for. The question is whether the current AI infrastructure buildout is analogous — and if so, whether the hyperscalers are playing WorldCom, or whether they are playing the internet applications that came after.

The critical difference from 2000: these companies are not building infrastructure they cannot monetize. Azure's revenue grew 40% this quarter. Google Cloud grew 63%. AWS grew 28% — its fastest rate in 15 quarters. The infrastructure they are building is already generating revenue faster than most analysts expected. The $515B bet is not speculative in the way that telecom fiber was speculative. It is an acceleration of a supply-demand imbalance that already exists. Cloud demand is growing faster than supply can be provisioned.

The Iran variable no one saw coming. Meta's daily active people came in at 3.56 billion versus a 3.62 billion estimate. Part of the miss? Internet disruptions in Iran — where the ongoing US blockade has degraded connectivity — and WhatsApp restrictions in Russia. The Iran war is showing up not just in oil prices and defense budgets, but in the user-growth metrics of a social media company. This is a reminder that the Hormuz crisis is not a siloed geopolitical event; it is embedded in the operating results of companies across sectors we do not normally associate with the Middle East. Meta lost active users to a naval blockade. Chevron missed revenue because Hormuz production is disrupted. Boeing is dealing with delivery complications from restricted airspace. War reaches further into quarterly earnings than the earnings call agendas acknowledge.

The compounding bet. The real question for the next 18 months is whether the hyperscalers' capex proves prescient or excessive. In 2023, a consensus view held that cloud growth was slowing, AI was a feature not a product, and hyperscaler capex would normalize down. That view was wrong — spectacularly so. Azure +40%, Google Cloud +63%, AWS +28% says the infrastructure investment was correct. The 2024-2025 camp argued that the current buildout was a bubble, that LLM monetization was unclear, and that the capex would depress returns for years. This week's results argue that camp is also wrong: monetization is happening faster than the skeptics modeled.

The nvidia_supply_chain and picks_and_shovels_ai strategies are positioned for the asymmetry: the companies that supply the infrastructure rather than own it tend to see revenue immediately when capex budgets are committed, without waiting for the hyperscalers' FCF cycles to normalize. ai_infrastructure_layer is the same thesis expressed differently — own the stack beneath the application layer, where $515B in committed capex translates to purchase orders, not aspirational guidance.

The week's scorecard: three companies committed half a trillion dollars in AI infrastructure, and their stocks fell. The chip companies, the power equipment makers, the networking layer, the cooling companies — they will quietly receive those purchase orders over the next 24 months. History tends to reward the people who supply the gold rush more reliably than the people who run the mines. The AI gold rush just got a $515B deposit.

Now go enjoy your Saturday in Omaha if you're there — it's Greg Abel's first time at the center of a 40,000-person Q&A. Worth watching even if you don't own a single B-share.

---

Sources:
- S&P 500 Snapshot: Weekly Win Streak Continues — Advisor Perspectives (May 1, 2026)
- Stock market today: S&P 500, Nasdaq jump to fresh records — Yahoo Finance (May 1, 2026)
- Dow surges nearly 800 points, S&P 500 posts first close above 7,200 — CNBC (Apr 29, 2026)
- Stock market today May 1, 2026 — TheStreet
- KOSPI hits intraday record above 6,750, ends 1.38% lower — Korea Times (Apr 30, 2026)
- Oil briefly touches $126, its highest price in four years — CNN Business (Apr 30, 2026)
- Brent oil tops $118 after Trump says he will blockade Iran — CNBC (Apr 29, 2026)
- Oil prices rise despite Iran's proposal to reopen Strait of Hormuz — Al Jazeera (Apr 28, 2026)
- UAE quits OPEC: What that means — Al Jazeera (Apr 29, 2026)
- UAE's shock OPEC exit — CNBC (Apr 28, 2026)
- Federal Reserve FOMC Statement — Fed.gov (Apr 29, 2026)
- Fed interest rate decision April 2026 — CNBC
- GDP advance estimate Q1 2026 — BEA (Apr 30, 2026)
- PCE inflation rate March 2026 — CNBC (Apr 30, 2026)
- Personal income and outlays March 2026 — BEA
- Microsoft FY26 Q3 press release — Microsoft IR
- Microsoft Q3 2026 earnings — CNBC
- Meta Reports First Quarter 2026 Results — Meta IR
- Meta stock sinks on capex hike — Yahoo Finance
- Alphabet Q1 2026 earnings — CNBC
- Alphabet Q1 2026 results SEC filing
- Amazon Q1 2026 earnings — CNBC
- Apple Q1 FY2026 earnings record — Yahoo Finance
- ExxonMobil Q1 2026 results — Exxon IR
- Exxon and Chevron Q1 2026 earnings — CNBC (May 1, 2026)
- Qualcomm shares soar on China orders, hyperscaler customer — CNBC
- Ford beats Q1 earnings on $1.3B tariff relief — Electrek
- General Motors Q1 2026 earnings — CNBC (Apr 28, 2026)
- Mastercard Q1 2026 beat, April slowdown — Yahoo Finance
- Japan likely spent $34.5 billion in FX intervention — Bloomberg (May 1, 2026)
- 10-year Treasury yield jumps after Fed holds rates — CNBC (Apr 29, 2026)
- Berkshire shares struggle into annual meeting — CNBC (May 1, 2026)
- Hegseth defends Iran war in first Congressional testimony — CNBC (Apr 29, 2026)
- 2026 Strait of Hormuz crisis — Wikipedia
- T. Rowe Price Global Markets Weekly Update

---

Disclaimer

The information, strategies, research reports, pre-market analyses, trade recommendations, and all other content contained in this repository are provided for educational and research purposes only and do not constitute financial advice, investment advice, trading advice, or any other form of professional advice. Nothing herein should be construed as a recommendation, solicitation, or offer to buy or sell any securities or financial instruments.

All content in this repository — including but not limited to pre-market research reports, strategy analyses, trade recommendations, leaderboard rankings, and research findings — has been generated in whole or in part by large language models (LLMs) and automated systems. AI-generated content may contain errors, inaccuracies, hallucinated facts, incorrect dates, fabricated statistics, or misleading information. Users must independently verify all information, data points, dates, financial figures, earnings results, economic indicators, and factual claims before relying on them for any purpose.

Past performance, whether actual or backtested, is not indicative of future results. Trading and investing involve substantial risk of loss. Consult a qualified, licensed financial advisor before making any investment decisions.