CNBC Halftime Report

2026-05-21 · Hosted by Scott Wapner · CNBC

Executive Summary

The Halftime Report spent the hour debating what Nvidia’s earnings would mean for the AI trade, anchored by a wide-ranging Jeff Bezos interview on Squawk Box, SpaceX’s S1 filing, and the OpenAI IPO announcement. Ahead of the 4pm print, the investment committee — Josh Brown, Joe Terranova, Steve Weiss, and Jim Lavinthal — dissected Nvidia’s position as the world’s largest company, the AI “wave vs. bubble” debate, and the cybersecurity sector’s re-emergence. Key data points offered pre-earnings: Nvidia has driven 20% of S&P 500 year-to-date gain and semiconductors account for 78% of S&P 500 price appreciation so far in 2026. Post-earnings detail was not yet available at broadcast time. SpaceX’s Goldman Sachs-led IPO was confirmed as expected to be “at least double” Alibaba’s $25 billion 2014 record, with fees potentially exceeding half a billion dollars.

Key Stories & Changes

1. The Case for Nvidia: Pre-Earnings Setup

  • 80% year-over-year revenue growth from an all-time high in revenue — Josh Brown: “no analog to this having happened ever in history”

  • 119% year-over-year EPS growth; gross margins 75%

  • Data center revenue of $73.2 billion — three years ago the business was $4 billion

  • Nvidia trading at 25× forward earnings — below growth rate; trading at less than market multiple per Paul Meeks (on other shows)

  • Nvidia has driven 20% of S&P 500 gains year-to-date

  • Semiconductors responsible for 78% of S&P 500 price appreciation YTD

  • Joe Terranova: Nvidia’s market cap is ~$5.5 trillion — bigger than the entire healthcare sector and financial sector

  • 93% of semiconductor companies that reported this earnings season beat estimates; Intel, Texas Instruments, AMD all saw double-digit post-earnings moves

  • Nvidia’s stock has dropped 5 of the last 7 earnings prints despite strong results — runs up into the print, often sells on the news

2. Jeff Bezos on AI, Bubbles, and Space

  • Bezos on AI bubble: “We’re in a phase where every experiment is getting funded… good ideas will pay for all the losers… even if it does turn out to be a bubble, it’s driving investment that is very healthy”

  • Josh Brown contrasts: this AI buildout is funded by the world’s most profitable companies from free cash flow, not equity IPO proceeds feeding into server purchases (unlike 1999)

  • Bezos cautioned on data centers in space: Elon’s “two or three years” probably “a little ambitious”

  • Steve Weiss: private market AI companies (Figure AI: $100M invested at $2.5B valuation, now at $30–36B valuation with no real revenues) — that’s where the bubble damage will manifest, not in public semis

  • Josh Brown’s “wave vs. bubble” argument: stocks can correct while technology investment remains healthy and societally productive

3. SpaceX IPO: Goldman Sachs Wins Lead-Left

  • Goldman Sachs selected as lead-left underwriter; Leslie Picker reported based on three sources

  • Offering size: expected at least double Alibaba’s $25 billion — likely >$50 billion

  • Bank fees: >1% of offering size = upwards of $500 million for Goldman and co-underwriters

  • Morgan Stanley: second in line; BofA, Citi, JPMorgan in alphabetical order

  • S1 expected to flip the same day (confirmed later in the episode)

  • Steve Weiss: IPO market wants to be “first one out” in their category; ego and strategic importance in sequencing matters

4. OpenAI IPO Filing: Racing Anthropic

  • Wall Street Journal reported OpenAI preparing IPO filing as soon as Friday

  • Potential listing as soon as September 2026

  • Banks: Goldman Sachs and Morgan Stanley (per Kate Rooney / CNBC reporting)

  • OpenAI racing to beat Anthropic to market — rivalry intensified after Anthropic’s Dario Amodei disclosed $30 billion annual run rate (up from $9B as of end of December)

  • Josh Brown: OpenAI “doing jujitsu” — pretending not to be in a rush, then acting urgently

  • OpenAI is strongest in consumer LLM; Anthropic stronger in enterprise code; market battleground shifting

  • Mackenzie Sigalos: IPO could come as soon as Friday for confidential filing; listing as soon as September; clears Musk legal overhang

  • Steve Weiss noted OpenAI has “too many stakeholders to fail” at this point; likely to be added to indices very quickly after listing

5. Cybersecurity’s AI-Driven Renaissance: Josh Brown’s Best Stocks

  • Josh Brown highlighted three cyber stocks in “Best Stocks in the Market” update:

  • Fortinet (FTNT): beat on top + bottom; revenue +20% year-over-year; EPS growth +40%; cybersecurity networking segment (66% of billings) up 32% year-over-year, accelerating; RSI 80–90 (parabolic, not a buy here)

  • Palo Alto Networks: 1,500 platform customers spending 19% more year-over-year; subscription business growing 33%; RSI near 90

  • CrowdStrike (CRWD): Josh owns personally; went from low $300s during “SaaS apocalypse” to knocking on $650; up 44.5% since April 30th rebalance; up 45% on 1-year basis

  • Brown: “More AI means more cyber threats, not less” — George Kurtz specifically said this

  • “The proposition of these companies being disrupted out of business by an employee vibe coding their own security patches” was “ludicrous”

6. Retail Sector: TJX, Ross, Walmart Setup

  • TJX: up 6% (best day since August 2024); raised full-year outlook; earnings up 29%; serving cost-conscious consumer

  • Joe Terranova: TJX heading to April 9th all-time high at $165; represents both cost-conscious consumer theme and scale advantages

  • Ross Stores: up 3% in sympathy; reports tomorrow (same narrative)

  • Walmart: reports tomorrow morning; Joe expects strong print; “Walmart is a tech company” — giving it valuation premium

  • Street expects Walmart to show consumer resilience, cross-tier appeal (value + wealthier consumers)

1. The AI “Wave vs. Bubble” Debate Sharpening

Bezos’s assertion that even a bubble is healthy for civilization — because investment drives technology — was debated intensely. Josh Brown’s “wave” framing is the most analytically precise: the infrastructure buildout is real and funded by companies with free cash flow, not borrowed equity, unlike 1999. But Steve Weiss identified exactly where bubble-like dynamics are concentrated: private market AI companies with no revenues at 10–20× the valuation growth of the previous round. The public/private distinction matters for investor risk.

2. The Sequencing of Mega-IPOs Matters Strategically

Multiple analysts on the show explained that OpenAI’s apparent pivot to faster IPO timing is partly competitive ego and partly strategic: being first out in a category matters for pricing, narrative, and subsequent comparables. Goldman Sachs winning SpaceX’s lead-left underwriting — and apparently working on OpenAI — concentrates enormous fee and allocation power in one institution. The availability of these deals simultaneously creates both capital competition and a “legitimizing” moment for the entire AI sector.

3. Semiconductors’ Dominance of Market Returns Is Unsustainable — and That’s Fine

Accounting for 78% of S&P 500 YTD price appreciation while Nvidia alone drives 20% creates index concentration risks. Joe Terranova cited Goldman Sachs data showing momentum factor is in the 100th percentile of positioning over the last five years — historically cyclical but not secular. The extreme leverage ETF inflow into semis amplifies this: moves that should be 2% become 5% in both directions. The good news: these companies aren’t going to zero (unlike 1999 dot-coms); valuation corrections are possible without permanent capital loss.

4. Cyber Is the AI Trade’s Durable Second Derivative

The cybersecurity sector’s renaissance — CrowdStrike up 45% in a year, Fortinet/Palo Alto parabolic — directly contradicts the narrative that AI would disrupt enterprise security. Instead, more AI infrastructure means more attack surface, meaning more cyber spending. This is a durable second-derivative play on AI that doesn’t require betting directly on GPU demand. The stocks are extended (RSI 80–90) but the thesis is intact. —-

Sentiment Analysis

Overall Market Sentiment: Constructively Bullish With Bubble Awareness

The investment committee was broadly bullish on AI infrastructure, dismissive of simple bubble comparisons, but aware of concentration risks and private market excess.

Risk Factors Highlighted

Nvidia Post-Earnings Sell Pattern: Stock has sold off 5 of last 7 quarters after earnings; run-up into print leaves limited post-announcement upside

Semiconductor Momentum at 100th Percentile: Goldman Sachs desk data shows momentum factor maxed out; cyclical correction expected; leveraged ETF amplification adds volatility

Private Market AI Bubble: Companies like Figure AI (valued at $30–36B with no real revenues) represent where bubble dynamics are concentrated; public investors buying IPOs may be on the wrong side

Mega-IPO Supply Competition: SpaceX, OpenAI, Anthropic all coming to market; capital competition is real; something must be sold to buy

Cybersecurity Stocks Technically Extended: RSI 80–90 across Fortinet, Palo Alto, CrowdStrike; not entry points; risk of sharp mean reversion on any sector rotation

OpenAI Governance Uncertainty: Stories of CFO/Sam Altman differences on IPO timing and target achievement; internal uncertainty at a company targeting $1T valuation

Space Data Center Timeline Risk: Bezos explicitly said Elon’s 2–3 year data center in space timeline is “probably a little ambitious” — could disappoint investors pricing in near-term infrastructure in space

Straits of Hormuz Reopening = Rotation Risk: If straits reopen, oil falls, AI/semi trade may face rotation as cyclicals become more attractive; Santoli noted this exact dynamic in the day’s trading

This episode was covered in today’s The Market Signal — 2026-05-21, a cross-source synthesis of multiple podcast reports.

Keep Reading