CNBC Halftime Report
2026-05-05 · Hosted by Scott Wapner · CNBC
Executive Summary
Featured a marquee interview with Altimeter’s Brad Gerstner at Milken on his 55th birthday — Altimeter just had its best month in firm history as the AI infrastructure trade powered through Mideast volatility. The Investment Committee debated mega-cap tech leadership: only 5% of S&P companies have missed estimates (smallest share in 25 years ex-COVID), Q1 earnings up 16% ex-private markups (28% as reported), and tech earnings drove most of the strength. Joe Terranova added Apple, Amazon, Meta back into the JOET ETF on quality+momentum. Gerstner sees Nvidia as the first $10 trillion company, expects 2027 rollout of mandatory Trump Accounts ($1,000 per newborn in S&P 500 index). The Trump admin floated AI model pre-release vetting.
Key Stories & Changes
1. Brad Gerstner — Best Month in Firm History
Altimeter owns compute, logic, memory: Nvidia, SK Hynix, CoreWeave, ARM, Cerebras (board, IPO next week)
Mag Five capex this year: $800 billion total
Mega-cap revenues: 8 quarters ago $150B → now $350B combined
Amazon revenue growth +28% at scale; Google Cloud +63%; Azure +39%
Gerstner: “I think Nvidia will be the first $10 trillion company” — currently 13-14x fully-taxed GAAP earnings
Memory stocks at 5x earnings; “Samsung will do more in profits this year than Google”
Sold Microsoft (rotated to memory/SK Hynix) — owns 6 of 7 MAG7 (no Tesla)
2. Anthropic vs OpenAI — “Team America Winning”
Both on Altimeter cap table
Anthropic took lead Jan-April adding “historic levels of new revenue”
Claude Code + Opus 4.6 launched early December → enterprise adoption surging
Codex (OpenAI) accelerating in last 4 weeks — “incredible” buzz on X/Twitter
ChatGPT at 900M+ weekly actives (not yet 1B)
“It is the verb still in AI”
Both already raised $110-120B private rounds — “they’ve already IPO’d, just privately”
SpaceX and xAI going public end of June — “Elon is an N of 1”
3. Trump Accounts Launching July 4 (60 days out)
Every newborn in America gets $1,000 in S&P 500 at birth, automatic, stapled to SSN
Math: $1,000 + $50/month = $50K at age 18, $200K at age 30, $1M at age 55
Already 6M kids signed up (Treasury said 5M two weeks ago); target 10M by July 4
Eligible kids under age 2 get $1,000 seed; older kids get $250
Michael Dell + Susan donated $6.25 billion
Employers (Uber, Nvidia, Salesforce, AMD) committed to add to employee kids’ accounts
4. JOET ETF Rebalance — Mag 7 Reasserted
Joe Terranova added Apple, Amazon, Meta back to JOET
ETF now holds 6 of 7 Mag 7 (no Tesla)
Reduced financial sector exposure (had hurt performance)
Meta “barely got in” — 12-month momentum just over 1%
Rationale: monetization of AI capex now visible in Alphabet/Amazon earnings (52-week highs)
5. Earnings Backdrop
5% miss rate = smallest share in 25+ years (ex-COVID)
Q1 EPS growth 28% headline, ~16% ex-private markups (Anthropic, SpaceX, Databricks)
Mega-cap premium to S&P now only 13% (per Goldman) — narrowest in years
Tech NASDAQ +14% in last month, best since October 2002
Earnings revisions rising — but Amy Raskin warns these are lagging indicators
6. Microsoft / Software Debate
Bren Talkies: doubled down on Microsoft despite skeptical view in past
Citing Azure +40%, Copilot 15M → 20M seats, plus pivot to seat+usage pricing
“If they can execute, this stock will not be at $414 a year from now”
Joe Terranova: Apple will be next Mag 7 to hit 52-week high
7. AI Regulation & Pre-Release Vetting
WH considering executive order on AI model vetting before release
Working group of tech execs + government officials
Mythos model triggering attention
Anthropic + FIS building bank financial-crime AI agent
8. Memory Chips at 5x Earnings
Micron at S&P 52-week highs; Gerstner doing podcast with Micron CEO in coming weeks
Industry restructuring + hyperscaler collaboration could remove “boom-bust” stigma
Gerstner views memory as long-duration AI infrastructure play
9. Nvidia Price Action — “Terribly Under-owned”
Nvidia at $195, lower than 6 months ago
Concerns over Trainium / TPUs / Cerebras share gains
Gerstner: “token per watt per dollar” still favors Nvidia; ChatGPT 5.5 trained on new Blackwell
Vera Rubin LPX (Grok acquisition) rolling out later this year — “another huge step forward”
Trillion-dollar future order pipeline cited
10. GameStop / eBay — Investment Committee Debate
Jim Lavinthal: “I think the deal is credible” — TD’s $20B financing letter + GameStop’s $9B cash + retail love
$125 offer “way too cheap” if deal happens — Lavinthal long eBay since mid-90s
Joe Terranova: “I almost wish this news didn’t come out today” — concerns about quality of new shareholders chasing on M&A speculation rather than fundamentals
Trends Identified
1. AI Model Race Is “Team America” — Not Single Winner
Gerstner repeatedly framed Anthropic vs OpenAI as multi-horse race in same way Mag 7 split cloud profits. “I would buy both today in the public market” — total addressable market still very early. Counters narrative that one model wins all.
2. Privatized IPOs and Long Compounding Reality
Gerstner’s blunt point: companies like OpenAI ($110B raise) have effectively IPO’d at trillion-dollar valuations privately. Retail investors entering at IPO shouldn’t expect “doubling/tripling” — should target 20-30% annual compounding. The classic “value capture in private markets” debate framed plainly.
3. Trump Accounts as “Largest Social Contract Shift Since Social Security”
This isn’t a small detail — Gerstner argues mandatory $1,000-at-birth in S&P 500 (automatic from 2027) plus employer contributions creates “401K-like benefit” that aligns 40M+ kids with capital markets. Major narrative shift for retail demand on US equities long-term.
4. Earnings Revisions Hide Underlying AI Spending Distortion
Amy Raskin pushed back on revision-based bullishness: AI revenue gets booked immediately for chip suppliers, but capitalized over years for hyperscalers — creating math distortion. Half of Q1 earnings strength came from private-company markups (Anthropic, SpaceX, Databricks).
5. Memory Re-Rating from Cyclical to AI Infrastructure
Gerstner’s case for memory stocks at 5x earnings rests on durability of demand through 2028-2029 — not cyclical boom-bust. Industry restructuring + hyperscaler collaboration could justify multiple expansion.
6. Fed at Crossroads with Warsh Coming In
Steve Liesman on John Williams’ speech: Fed sees “elevated level of concern” on Mideast → 3% inflation forecast for 2026, GDP 2-2.25%, supply chain emergency risks rising. Williams “well positioned” line stands but reflects unease as Warsh nominee transitions in. —-
Sentiment Analysis
Overall Market Sentiment: Bullish on AI Infrastructure, Cautious on Macro
The desk and Gerstner agree on AI tailwinds but acknowledge geopolitical/oil pressure on the margin. Underlying tone is “dips will be bought” given earnings strength.
Risk Factors Highlighted
Mideast escalation: Oil at $120+ would “take its toll” on market over 3-4 months
AI capex/revenue mismatch: Eventually catches up via depreciation
Earnings revisions are lagging: Stocks lead, not follow
Mag 7 narrowness: Only 20% of S&P stocks beat index in last month
Fed regime change: Warsh transition + Fed independence concerns
Inflation breakeven 5-year highs: 5/5 swap pricing rising
Microsoft / Meta capex without monetization clarity: Margins under scrutiny
GameStop dilution risk: Even if deal happens, equity issuance hurts shareholders
Berkshire underperformance: $400B cash drag, no buybacks, -40% relative
Retail IPO dynamics: Buying $1T+ company is “not a get-rich scheme”
Mythos / AI cyber capabilities: Government vetting may slow innovation
Energy supply chain crisis: Williams flags emergency supply chain risk
This episode was covered in today’s The Market Signal — 2026-05-05, a cross-source synthesis of multiple podcast reports.