Market Commentary - Week of April 18, 2026

StoicFunds Signal Check

• Macro Regime: Shifting. S&P crossed 7,000 for the first time — up 4.5% for the week. Oil plunged 11.45% on Friday after Iran declared Hormuz "completely open." If oil holds lower and equities hold above prior highs, the regime could flip from Inflationary Bust toward Deflationary Boom on the next signal update.
• Market Stress: Likely improving. Three consecutive winning weeks. Nasdaq posted 12 straight positive sessions — longest since 2009. But one rally doesn't flip the signal. The system waits for confirmation.
• Momentum: Broad risk-on shift underway. Energy gave back gains (oil -13.2% for the week). Bank earnings strong. PE/private credit sector rallied hard. The signal rotation from energy-only BULL to broader participation is a data point to watch.
• The models will recalibrate. The war-driven regime has dominated for 10 weeks. If Hormuz stays open and a ceasefire holds, we could see a significant regime transition. The system reads price, not headlines.

Iran — Hormuz Opens, But It's Not Over

• Iran declared the Strait of Hormuz "completely open" on Friday. Oil plunged 11.45% to $83.85 (WTI), down 13.2% for the week. Second-largest single-day drop since the war began
• Israel-Lebanon ceasefire announced Thursday — 10-day truce. But Hezbollah did not agree to terms. Trump said the naval blockade on Iran specifically remains in effect
• Iran's framing: Strait has returned to its "previous state" under "strict management and control" by Iran's military. Not a US victory — Iran asserting sovereignty over the waterway
• Markets pricing in optimism: S&P hit all-time highs. But lost production during the conflict will keep energy markets tight relative to pre-war levels when oil traded below $65
• Key uncertainty: whether ceasefire becomes lasting peace or just another pause in escalation (the pattern since February)

Bank Earnings — The Credit Cycle Is Not Here (Yet)

• Big 4 reported: JPM, Wells, Citi, Bank of America. Investment banks crushed it; lending-focused banks were mixed
• Goldman Sachs: EPS $17.55 (beat), +24% YoY. Advisory revenue +89%. 21% ROTCE. But fixed income trading -10% and no operating leverage. Stock down on the print
• Morgan Stanley: EPS $3.43 (beat), +32% YoY growth. Best-in-class 27% ROTCE. Stock trying to break out
• JP Morgan: EPS $5.94 (beat). 23% ROTCE. Fixed income trading +21%. Positive operating leverage. Jamie Dimon cautionary on credit (as always), but credit data benign
• Wells Fargo: Missed on NII. 13bp sequential NIM decline — large for a lending bank. Lower tax rate masked the miss. Stock down sharply. Disclosed $36B private credit exposure — software is 17% ($6B), most with ~40% cushion
• Citi: EPS $3.06 (beat). 13% ROTCE — best in years under Jane Fraser. Nice operating leverage
• Bank of America: EPS beat. 16% ROTCE (up from stuck-at-14%). Benign credit data
• The non-accrual data matters most for the credit cycle thesis. JPM total non-accruing loans: $9.6B — up 11% YoY but down 3% sequentially. BofA: $5.8B — down 4% YoY, flat sequentially. There is no deterioration pattern forming. As the banks go, so goes the economy — and the banks say no recession is on the horizon while credit quality holds
• Private credit bank exposure is now disclosed: JPM $50B, Wells $36B, Citi $22B. Large in absolute terms but manageable — most are senior loans with subordination protection. If losses materialize, this is an income statement problem, not a systemic capital problem
• The PE/private credit sector rallied hard on the benign bank data. The credit cycle may still be coming, but bank earnings season says: not yet

Software & AI Threat — Too Early To Call

• The investment thesis on AI vs. software is still forming. Investors are worried about everything — AI replacing platforms, companies building internally, pricing power evaporating permanently. The threat landscape is broad and the fear is real
• The honest assessment: it's too early to know. Clarity on how AI reshapes software economics is likely 12+ months away. Bottom-fishing in a sector with this much narrative negativity remains dangerous
• Subscription fatigue emerging (eIBD): The average consumer now carries 14 active subscriptions. AI services cracked the top 10 subscription categories for the first time — competing fiercely for finite wallet share
• Regulatory pressure on subscription models growing: Adobe settled $150M with the DOJ over concealed subscription terms. The FTC filed a complaint against Uber over deceptive Uber One practices. The subscription economy's growth-at-all-costs era is encountering friction

AI Race — Anthropic Overtakes OpenAI

• Anthropic hit $30B ARR as of early April — surpassing OpenAI (~$25B). The growth differential is stark: ~10x/year versus ~3-4x/year. Secondary markets now price Anthropic higher than OpenAI for the first time
• OpenAI is navigating an identity crisis. Internal tensions between consumer and enterprise strategy are becoming public. The $850B valuation is finding no buyers on secondary markets. Questions about revenue accounting — with ~$8B attributed to channel partner revenue-sharing — are adding pressure
• The enterprise revenue flywheel is decisive. Enterprise customers pay for code tokens on a metered basis — usage scales, and so does revenue. Consumer subscribers have lower willingness to pay and only 3-4% convert to premium. The company capturing enterprise coding at scale wins this race
• The growth rate gap is the entire story. At the same revenue level, the company growing 3x faster captures the network effects — more tokens, more data, more reinforcement learning, more compute funded by actual revenue rather than investment capital. Efficiency outstrips subsidy every time
• The Mythos question: Anthropic's decision to withhold its most powerful model may have been partially driven by compute constraints — a model potentially 10-20x the token cost of Opus simply couldn't be served at scale. Strategic altruism and infrastructure reality aren't mutually exclusive

Compute Crisis — The Five-Alarm Fire

• Hyperscalers control 60% of all compute. Frontier labs have grown large enough that dependency on AWS, Azure, and GCP is now a strategic liability — not just a cost issue
• For Anthropic and OpenAI, securing independent compute infrastructure is existential. Without owned land, power, and shell capacity, revenue growth could hit a wall — not from product quality or adoption, but from the inability to serve demand. The historical parallel: Friendster didn't fail because users left. It failed because the site couldn't keep up
• Data center backlash is intensifying. Maine passed the nation's first statewide ban on large data centers. A town approved a $6B data center build, then half the board was voted out to undo the decision. An estimated 40% of contested data centers are getting cancelled — more than double last year's rate. The total economic value at risk: ~$162B
• Public sentiment on AI is shifting negative. The average consumer doesn't see the value yet — most of AI's impact is showing up in enterprise, not in daily life. Data centers have become the physical manifestation of that disconnect: the visible target for populist resentment
• The buildout continues regardless. Colossus is expanding to 555,000 GPUs across three buildings ($18B investment), with excess capacity being leased to companies like Cursor. Meta's Prometheus cluster targets 150,000 GPUs for 2026. The ratepayer protection pledge ensures new data centers bring their own power — grid-neutral by design. But policy solutions don't solve political opposition

Allbirds → NewBird AI — The Signal Under the Meme

• Allbirds sold its brand assets for $39M (10% of its IPO raise), rebranded as "NewBird AI" with a $50M convertible note for data center services. The stock surged ~580% before pulling back sharply
• The late-90s parallels are obvious. Change your name to [anything].AI and the stock pops — the same pattern as adding .com in 1999
• But there's a serious signal underneath. The market is screaming that we are massively compute constrained. Bloom Energy is going vertical because it solves on-site power. Jane Street committed $1B plus a $6B compute deal with a neoscaler. When sophisticated capital is chasing the same thesis as meme stocks, pay attention to the thesis

Market Performance

• S&P 500 broke 7,000 for the first time — closed at 7,041.28. Best weekly performance since May 2025 (+4.5%)
• Nasdaq: 12 consecutive positive sessions, longest winning streak since 2009
• Oil: WTI crashed to $83.85 (-11.45% Friday, -13.2% for week) on Hormuz opening
• Philly Fed manufacturing unexpectedly surged to highest since January 2025. New orders also best since early 2025. Prices paid hit 8-month high
• ASML: Beat (EPS +24%, sales +17% to $10.3B) but Q2 guidance fell short. TSMC: EPS +63%, sales +39%, raised full-year outlook. Both stocks fell after results then pared losses
• Netflix: EPS +86% (boosted by $2.8B Warner Bros. termination fee). Revenue +16%. But Q2 guidance lower — stock dived Friday
• Revolution Medicines: +41% in a day on "unprecedented" pancreatic cancer trial results — patients survived 13.2 months vs. 6.7 months on chemo
• Amazon buying Globalstar ($11.6B) — satellite play competing with SpaceX Starlink
• Credo Technology soared on DustPhotonics acquisition — silicon photonics play for data center connectivity
• United-American Airlines merger pitched to Trump by United CEO Scott Kirby. Would face massive regulatory hurdles
• eVTOL pilot program (eIPP): FAA tapped 8 companies for battery-powered air taxi trials spanning half the country. 3-year timeline. Joby, Archer, others in the mix

Valuation Framework — Banks

• Banks with large IB/trading businesses (Goldman, Morgan Stanley, JPM) command 20%+ ROTCE and ~3x tangible book
• Lending-centric banks (Wells, BofA) stuck at 14-16% ROTCE, 1.8-1.9x tangible book
• Why the gap: Lending is hyper-competitive. NIMs in long-term decline. Banks compete with each other AND a multitude of non-banks
• Goldman rerating: From 1.3x tangible book (value stock) to nearly 3x. Driven by capital markets recovery
• Citi turnaround: Consistently below 10% ROTCE since GFC. Now 13% under Jane Fraser — stock rerating from below tangible to 1.3x. Still a steep discount but moving