MERIDIAN RESEARCH

Two cross-currents dominate the tape this morning. First, energy supply structure is shifting: the UAE is reportedly exiting OPEC effective May 1 (CNBC), removing one of the cartel's most disciplined producers and adding ambiguity to the OPEC+ quota framework heading into summer driving season. Concurrently, Politico is reporting energy-desk consensus around another pump-price spike, and Barron's notes the Nasdaq is set to open lower on "Iran peace fears" — a paradoxical phrase that I read as the market pricing risk-off on a Middle East flare-up scenario, not an actual de-escalation. Net read: oil-linked tail risk is being repriced higher. This matters for CPI persistence (currently running on the FRED CPI index series; YoY trajectory remains elevated) and therefore for the Fed reaction function. The fed funds rate sits at 3.64% (FRED) with the 2Y at 3.78% (FRED) — a slightly inverted front-end signaling the market still expects modest cuts but is not pricing aggressive easing.

Second, AI capex digestion is becoming a real narrative, not a passing one. CNBC reports OpenAI missed internal revenue targets, with Oracle and chip-suppliers selling off pre-market. This is the first concrete data point challenging the "demand is uncapped" framing for AI infrastructure spend. It does not invalidate the secular thesis, but it does increase the dispersion of outcomes for hyperscaler-dependent names (NVDA, MSFT exposure indirect, CEG via data center power demand). I'm treating this as a signal to monitor, not yet a thesis-breaker — one anchor tenant missing internal targets is not the same as cloud capex rolling over.

Third, consumer health is deteriorating at the margin: Axios reports >50% of Americans say their finances are getting worse. Combined with the pump-price risk, this is a yellow flag for discretionary consumer names (CAVA, CELH, COST premium tier resilient but watch traffic).

HIGH CONVICTION (RECOMMEND):

• NVDA (8/10) | +29% 1M, +5.3% 1W: The OpenAI revenue miss (CNBC) is the first material data point that could pressure the H2 demand curve. NVDA is up sharply over the last month, which raises near-term setup risk. Thesis intact, but I'm flagging for re-underwrite if (a) hyperscaler capex guidance gets cut on next earnings cycle, or (b) Oracle/CoreWeave-type tier-2 customers signal order push-outs. No conviction change today. Cramer commentary noted but ignored per Rule 4.
• MSFT (8/10) | +19.6% 1M: Q3 earnings preview cycle underway. Stellantis partnership and the Meta/MSFT job reduction headlines (CNBC, Yahoo) are consistent with the AI-driven productivity reallocation thesis — flat-to-down headcount with rising revenue per employee is the bull case in motion. Conviction unchanged at 8/10; awaiting Q3 print.
• MELI (7/10) | +10.7% 1M, -3.7% 1W: Cantor and Jefferies positive (analyst input, not a conclusion). Brazil/LatAm consumer remains structurally underpenetrated in e-commerce and fintech. No change.

MONITORING — NOTABLE MOVES:

• MEDP (5/10) | -19.4% 1W, -11% 1M: Multiple securities class action filings (Rosen, Bragar Eagel, Levi & Korsinsky) alleging concealment of backlog deterioration. This is exactly the bear case I flagged at initiation. I am moving MEDP to active downgrade review. If the next 10-Q confirms backlog erosion, conviction goes to 3/10 and the name comes off the watchlist. Do not add on weakness.
• FOUR (6/10) | -9.7% 1W: No company-specific news driving the move; news feed returned irrelevant matches. Likely sympathy selling with the broader payments/fintech complex. Thesis unchanged — short-term volatility is not a sell signal (Hard Rule 7).
• CSGP (6/10) | -12% 1M: "Rising AI Fears Hurt CoStar Group in Q1" (Yahoo). This is the structural risk I called out at initiation — proprietary commercial real estate data being commoditized by LLMs. Watching closely; if next earnings shows subscriber attrition or net revenue retention compression below 100%, conviction goes to 4/10.
• AAON (6/10) | -9% 1W, +12.6% 1M: Volatile but no fundamental news. HVAC/data-center cooling thesis intact. Hold at 6/10.
• KTOS (5/10) | -10% 1W: No catalyst visible. Defense/drone thesis intact but conviction was already low. Monitoring.
• BWXT (6/10): Supportive sector news on SMR/nuclear fuel supply chain (Yahoo). Reinforces the "sole U.S. naval reactor manufacturer + SMR optionality" thesis. Conviction unchanged at 6/10; would consider upgrade on a clean Q1 print.
• CEG (6/10) | +6.5% 1W: Evercore ISI bullish resumption. Data center power demand thesis remains the dominant driver. If OpenAI revenue miss spreads to a broader hyperscaler capex pullback, CEG's PPA economics get pressured. Conviction unchanged but tied to AI capex narrative — same risk vector as NVDA.
• LLY (7/10) | -5.2% 1W: No company-specific negative news; weakness consistent with broader healthcare rotation. GLP-2 and oncology pipeline thesis intact. Holding 7/10.
• FICO (5/10): Mizuho initiated Outperform calling overreaction (Yahoo). Cross-check: my prior thesis flagged Fannie/Freddie regulatory risk as the swing factor. Mizuho's view is an input, not a conclusion. No change.

All other monitoring names: no thesis-altering data today.

No new initiations today. The pipeline already added five names in the past week (NTRA, MDB, FDS, BRK-B, DEEP). I'd rather deepen work on the existing 30+ monitored names — particularly the ones now at decision points (MEDP downgrade, CSGP risk reassessment) — than dilute attention with new names.

What I'm screening this week:

• Energy E&Ps — given the UAE/OPEC dislocation and pump-price spike risk, looking at low-cost domestic producers with strong balance sheets. Not yet at thesis-formation stage.
• Defensive consumer staples — Coca-Cola's beat (CNBC) is a reminder that pricing power in staples is still functioning; KO peers worth a look as recession hedge.

• MEDP — securities class action cluster. Three law firms now soliciting plaintiffs alleging backlog deterioration concealment. Pattern is consistent with material misrepresentation cases. Trigger for downgrade to 3/10: confirmation of backlog decline in next filing, or auditor-related disclosure.
• AI capex digestion (NVDA, CEG, MSFT indirectly). OpenAI internal target miss (CNBC) is the first crack. Trigger for action: any hyperscaler (MSFT, AMZN, GOOGL, META) cutting FY26 capex guidance by >10% on next earnings.
• Oil supply shock (broad portfolio). UAE-OPEC exit + Iran tail risk = material upside risk to oil prices, which feeds CPI, which constrains the Fed, which compresses long-duration multiples (DDOG, NET, MDB, CRWD, TTD all sensitive).
• Consumer rollover (CAVA, CELH, COST). Axios data on >50% of Americans reporting worsening finances. Trigger: comparable traffic decline >LSD% in next CAVA print.
• CSGP — AI disintermediation risk is now showing up in price (-12% 1M). Need to validate Q1 net revenue retention.