← CNBC Halftime Report hub  ·  Research hub  ·  Research library

CNBC Halftime Report — AI Safety Concerns Push Tech Stocks and the Market Lower

A Saturday essay from Anthropic's Dario Amodei proposing a three-step plan to temper frontier-model progress (third-party evaluators, safety standards, international coordination) sends the SOX down 5.3% on a morning the 10-year cracks 5% (5.00% at 10:26, back to 4.95% within 90 minutes) — and the committee, with Mike Santoli in for Wapner, mostly refuses to call it a game changer. Joe Terranova reads it as "consistent with the 2026 playbook where the market just wants to rotate" (SMH never regained its June high; money goes to the Mag 7, software and healthcare); Liz Thomas will "take that bet" that the Treasury and Fed won't let the 10-year run much past 5% and says the durable buy out of the scare is cybersecurity; Stephen Weiss is the bear — a hike "to reel in a supply shock doesn't work," yields go "through 5%," and Broadcom is the tape's message ("the stock has never bounced… you can pick a falling knife here"); Jim Lebenthal is the bull ("ultimately this is a market that goes higher… will crescendo into year end"), while conceding the Fed is "backed into raising rates more than once" into a diesel shock it cannot fix. Terranova names the risk he doesn't subscribe to but won't ignore: a market 3% off its high because everyone expects resolution in 60–90 days is "positioned one way… exiting from what will only be 1 door." The AI-power industrials take the real damage (GE Vernova, Vertiv, Quanta, Eaton held in JOET, "a much more perilous position"; Caterpillar already sold on the rebalance), and Thomas makes the out-of-consensus rates call: a hike looks like 1999, not 1997 — the start of a cycle. Calls of the day: Citi gold $4,800 (Thomas buys IAU below $4,300, Terranova distrusts the bounce), AppLovin cut to $450 ("a broken momentum stock… let it prove itself first"), Disney cut to $125 (Lebenthal sticks at 14×), Netflix to $110 (Weiss bought it back). Final trades: AstraZeneca (Lebenthal), Leidos (Weiss), the 10-year at 4.94% (Thomas), Expeditors (Terranova).
2026-SEP-14 · CNBC Halftime Report (audio edition) · host Mike Santoli (for Scott Wapner); committee Joe Terranova, Liz Thomas, Stephen Weiss, Jim Lebenthal; Kate Rooney on the Amodei essay · ~44 min · ▶ Listen · transcript · actionable insights
One-line take: The AI-safety scare hit the vendors, not the spenders — and the committee's useful work was sorting which side of that line each holding sits on. The market's own read, which Santoli puts in the first segment and nobody overturns, is that a slower frontier is fine for hyperscaler earnings ("maybe the spenders are going to be OK, even if the vendors have to do with less"): Microsoft and Alphabet are green, semis are −5%, and the AI-levered industrials that live "off of the forward order books for which nobody has the money to pay" get smoked. Terranova supplies the transmission channel from the essay to the industrials, and it is a financing one: "if you tell me we're slowing the rate of spending and if you tell me that yields are going to remain high, that's a disincentive to see more supply in the debt market" — the same yields-stop-issuance argument he made on Sep 8, now with a catalyst. Weiss adds the private-market leg: AI "lives and breathes" in private markets, the unprofitable labs will have to extend their profitability timelines and raise more, so expect "a thinning of the herd" and fewer IPO candidates — with Goldman trading down as the IPO proxy. The rates debate is the second half of the hour and it is sharper than Friday's: Weiss and Lebenthal agree a hike into an oil/diesel supply shock is the wrong tool ("they can't open the Strait of Hormuz") and that stagflation is the word to watch; Thomas calls a hike "a reaction to the market, not to the economy" that will look like a mistake — and 1999-style, because Warsh's 2% commitment means 25bp "isn't going to get us there"; Terranova expects the market to price the cut the moment the hike lands, which Weiss says would "destroy all credibility." The fragility flag is Thomas's: the equal-weight index broke its 50-day, industrials show cracks, and "the market going back to a concentrated large cap leadership is a more fragile place to be." Order: Positive → Neutral → Negative.

1. Stocks & names mentioned

Panel = CNBC's investment committee; "View" is the panel's net take this episode and the cell names which member(s) held it. This is an audio podcast and the Spotify transcript carried no (mm:ss) cues and no named speaker labels (only numeric "Speaker N" markers, mapped from context) — so each "At" cell simply opens the episode and the talking points below are plain segment headings rather than timestamps. Order: Positive → Neutral → Negative. Research: QT Qualtrim · SA Seeking Alpha · STK Stock Analysis.

TickerNameResearchViewWhat the committee saidAt
MSFTMicrosoftQT · SA · STK · FAPositiveLebenthal's holding and the day's "port in the storm" — the spender that benefits if the frontier slows. Santoli: "Jim, you own Amazon, Alphabet, Microsoft, so you know, that's the port in the storm today… somebody wants to own these rather than Broadcom." Lebenthal: "it's hard to get awfully negative on the markets, Mike, when those names… are actually up… Microsoft, let's not forget, has had a heck of a nice bid recently. So when something as big as Microsoft is performing as well as it is, again, really hard to get negative on this market overall." Weiss explains the bid: "we're seeing Meta and Microsoft and Alphabet move up because say, OK, great, they're going to slow down spending."listen↗
GOOGLAlphabetQT · SA · STK · FAPositiveLebenthal holding, green on a red tape. Named in the same "port in the storm" basket — "Alphabet and Microsoft" up while semis fall 5% — on Santoli's framing that the market is happy to say "maybe the spenders are going to be OK, even if the vendors have to do with less in the future… that's OK for hyperscaler type earnings." Thomas's version: investors are going back to "that muscle memory… of those Mag 7 names, of some of those hyperscalers."listen↗
AMZNAmazonQT · SA · STK · FAPositiveThe third Lebenthal hyperscaler — the one that didn't hold up. "Now, Amazon's obviously red, we see that." Held as part of the spender basket Santoli calls the port in the storm; no separate case made on the day.listen↗
DISWalt DisneyQT · SA · STK · FAPositiveRothschild trims its target to $125 from $160; Lebenthal holds on valuation. "I'd be happy at 125. I mean, let me start there. The stock has had a little bit of a rally since they reported earnings in early August… up 10-ish percent. Obviously, the market's been flat to down in that period… We've seen Marriott roll over just as Disney is coming up. When I get to the end of the analysis here, what I look at is a stock that's trading at 14 times forward earnings for a well known brand, one that does have consumer appeal. And I'm going to stick with this for now."listen↗
NFLXNetflixQT · SA · STK · FAPositiveWeiss quietly bought it back — a downside call, not an upside one. Evercore raises its target to $110 from $100; the stock is an S&P leader after being lumped in "that AI disruption basket." Weiss: "I bought it back not far from here actually, because I do have a lot of cash. I thought this was a safe one. I'm not so sure about the upside on it, but I think the downside's been taken out as you see other streaming services raised prices. They actually are close to the value player on the street… I'm not sure it gets to 100 anytime soon."listen↗
GSGoldman SachsQT · SA · STK · FAPositiveWeiss's only bank — down on the IPO read-through, which he dismisses. On the market's first reaction to the Amodei essay: "right now the market thinks that it's going to slow down all IPOs. So we can take a look at Goldman Sachs and see that down." Then on the Fed: "I think the banks look great here… we've seen some steepening of the yield curve… it's also a hedge on the economy… I own Goldman Sachs, the only one I ownI guarantee you they've got enough M&A to do and they've got enough other high margin business… they're going to be fine." Santoli: "I'm not worried about Goldman Sachs at this point."listen↗
IAUiShares Gold TrustQT · SA · STKPositiveThomas keeps buying her twice-given final trade into Citi's $4,800 call; Terranova distrusts the bounce. Thomas: "it was my final trade… on August 19th and September 2nd and it's down 1 to 3%I think it's more attractive now and I would still buy it… it is generally a 0 yielding asset. So when rates rise, gold is less attractive. But I think now it's become a geopolitical hedge againan attractive entry point with gold trading below 4300. I remember when we were so worried about it being 5000 and people were calling for seven and 8000… I'm still long and bullish." Terranova dissents on the chart: the parabolic peak was January, and the last 30 days are "this little reflex momentum driven gold move higher in the context of a larger downslope for 2026… I don't trust the rebound."listen↗
AZNAstraZenecaQT · SA · STK · FAPositiveLebenthal's final trade — borrowing Thomas's midterm-year sector call. "In a topsy turvy market, we're not quite sure which way it's going to go. I am with what Liz said, going with healthcare. AstraZeneca is right in the middle of the fairway for that space." Thomas's premise: in midterm election years "the sectors that tend to do the best are healthcare and staples… this year, healthcare is absolutely living up to that," now also as a beneficiary of the rotation out of tech into pharma and biotech.listen↗
LDOSLeidos HoldingsQT · SA · STK · FAPositiveWeiss's final trade — a defensive defence name off its highs. "I'm going with Leidos. It's up today but it's off the previous highs. I think it's got a very attractive offering going forward and it's defensive, being defence companies."listen↗
EXPDExpeditors International of WashingtonQT · SA · STK · FAPositiveTerranova's final trade — momentum at a 52-week high, with the revenue offshore. "Expeditors International of Washington is a $25 billion company with over 68% revenue exposure overseas and hitting a 52 week high today as we speak." The mirror image of his AppLovin rule in the same hour: buy what is making highs, not what is near its lows.listen↗
DALDelta Air LinesQT · SA · STK · FANeutralDemand intact, oil the offset — three sessions after Lebenthal called its correction over. Lebenthal: "it had a great run up until earnings. I did say at the time that I thought it was a little bit ahead of itself. I didn't know that oil prices were going to rise this much. And look, that offsets the gain that you're seeing from the demand side… any airport… they are packed… airline ticket prices are going up but oil's a headwind and it will be for the foreseeable future." Terranova (owns DAL and UAL in JOET as momentum names) separates it from the group: "I think Delta is in a much better position. Technically, it sits above the 200 day moving average… I like Delta's loyalty rewards program, premium flyer, and they also have the refiner exposure."listen↗
WFCWells FargoQT · SA · STK · FANeutralOn Weiss's watch list as the hike approaches. Context is a Wells Fargo (Mike Mayo) note naming JPMorgan, Bank of America and Citi as beneficiaries of a hike with the 10-year above 5% for the first time since October 2023. Weiss: "I own Goldman Sachs, the only one I own, I am looking at Wells Fargo." Thomas adds the caveat for the whole group: banks usually do well when the Fed hikes, "but a flatter yield curve is problematic for banks over the long term because it crimps their net interest margin."listen↗
GEVGE VernovaQT · SA · STK · FANeutralHeld in Terranova's JOET until the next rebalance — "a much more perilous position." Santoli: "AI levered industrials are getting smoked because that's the area where it really matters what's going to happen in 3, 4, 5 years… they're sold out for next year." Terranova: "Not particularly happy about it… the momentum factor has broken down late in the second quarter… GE Vernova, Vertiv, Quanta, Eaton… we're sitting there and we will remain there till the next rebalance. But they are in a much more perilous position… if you tell me we're slowing the rate of spending and… yields are going to remain high, that's a disincentive to see more supply in the debt market… that is absolutely going to affect these industrial names." Santoli: "they're living off of the forward order books for which nobody has the money to pay for."listen↗
VRTVertivQT · SA · STK · FANeutralSame JOET AI-power holding, same verdict. Named in the AI power & infrastructure ETF Santoli puts up ("it's got the Vertivs and GE Vernovas and Eatons") and in Terranova's held list — momentum broken, held only until the rules-based rebalance, exposed to any slowdown in debt-funded AI spend.listen↗
PWRQuanta ServicesQT · SA · STK · FANeutralJOET AI-power holding. In Terranova's list of names held into the next rebalance despite a broken momentum factor — "they have thrived over the last eight quarters on the continual spend and the build," which is exactly what the Amodei essay and a 5% 10-year put in question.listen↗
ETNEatonQT · SA · STK · FANeutralJOET AI-power holding. Named in both the AI power & infrastructure ETF and Terranova's held list; same caution — held by rule, not by conviction, "relative to where they were just three months ago."listen↗
CATCaterpillarQT · SA · STK · FANeutralAlready out of JOET — a rules-based exit, not an argued call. Terranova, listing his AI-power exposure: "we sold out of Caterpillar in the last rebalance, but we're sitting there [in the rest] and we will remain there till the next rebalance."listen↗
AnthropicAnthropic (private)NeutralThe source of the day's selloff — and, per Rooney's sources, adjusted-profitable into its IPO. Rooney: "Anthropic CEO Dario Amodei really shocked the tech world with an essay out on Saturday… a three-step plan aimed at tempering how fast some of the most powerful AI models… are going to be improving… third party evaluators… safety standards and then international coordination," citing AI "developing a lot faster than expected" and OpenAI agents hacking Hugging Face. On the listing: some argue "the safety push could actually help Anthropic's IPO, positioning the company now as a responsible player"; "I also confirmed… Anthropic has seen back-to-back quarters of at least adjusted profitability." Terranova suspects OpenAI's pause support is about "putting Anthropic in a position that we don't see them IPO in 2026"; Santoli: "the key will be… Anthropic's S-1… how they're characterizing their actual profitability"; Weiss wants the risk factors (David Sacks flagged product liability for agents).listen↗
OpenAIOpenAI (private)NeutralBacks the slowdown, and says an IPO now would be ill-advised. Rooney: Sam Altman tweeted support for Amodei, but "told Fortune over the weekend an IPO right now would be ill advised. He did say 2027 for their own listing" (CNBC had already reported next year). Amodei's essay cited "an example of OpenAI agents hacking into the startup Hugging Face." Rooney's framing of the labs' bind: they must impress Wall Street on revenue growth — "a big way to do that is to have the best and most expensive model" — while holding back for liability and regulatory reasons.listen↗
AVGOBroadcomQT · SA · STK · FANegativeWeiss's exhibit A for not putting new money into semis — a falling knife after two misses. "I just don't still don't believe you put new money in here… Let me just give this one, Broadcom. Broadcom missed, right? So they missed arguably this time, they missed last time… they reported good numbers, guidance, squishy things in their margins. The stock has never bounced. It bounced for a brief time… I bought it after the prior quarter, bounced a little bit, sold it flat and it traded at 410. Now it's back down to… below where it was before, 346. To me, that's the message in the market. You can pick a falling knife here, particularly when you still have valuations that are too high." Santoli: somebody wants to own the hyperscalers "rather than Broadcom."listen↗
SMHVanEck Semiconductor ETFQT · SA · STK · FANegativeThe group that took the hit (SOX −5.3%) — and the chart that never recovered. Terranova: "for the semis specifically, the SMH, the high was back in June, so it's never been able to exceed its previous highs… the market has played this game of kind of cat and mouse believing that there would be a restart. The momentum factor in semis itself… stalled out." Weiss: the market will be "in a trading range favoring the lower end as far as semis go. Yeah, there's a lot of air there." Thomas: investors are "a little freaked out about semiconductors because that's the high beta play." Santoli's counterweight: semis are "a big part of earnings growth projected into next year."listen↗
APPAppLovinQT · SA · STK · FANegativeMorgan Stanley cuts the target to $450 from $650; Terranova: don't buy near a 52-week low. "This is a broken momentum stock. It's now [−]42% if you're looking at a 12 month momentum score… three month momentum score, it's down as well… in the JOET ETF, the next quarterly rebalance… we'll figure on the position. As far as advocating it to buy it here, I don't think you want to do that. I don't like buying stocks that are near a 52 week low despite what might be the vision of some really good fundamentals ahead… allow the stock to stabilize and begin the recovery period, then step in… Let it prove itself first." He applies the same rule to the AI industrials.listen↗
UALUnited AirlinesQT · SA · STK · FANegativeTerranova's weaker airline — technically broken. "We own Delta and United in the ETF. Why? Because they are momentum players… that's been punishing over the last several months. I think Delta is in a much better position… above the 200 day moving average. United's broken down."listen↗

2. Talking points

The open: the Amodei essay, the SOX −5%, and a 5% 10-year

Rotation, not regime change

Weiss: supply shocks, a hike that can't work, and Broadcom

Lebenthal: hard to be negative when Microsoft has a bid

Terranova's one-door risk

Diesel, the Strait, and stagflation

Thomas: the Fed is the risk, and concentration is fragile

AI-power industrials: sold out for next year, unfunded after that

IPOs: from too much supply to not enough

Banks into the hike

1997 or 1999?

Hike, then the market prices the cut

Calls of the day: gold, AppLovin, Disney, Netflix

Mortgage rates, ETF Edge, Options Action

Airlines

Final trades

3. In plain English

MSFT — Microsoft Positive

Microsoft is one of the "hyperscalers" — the giant cloud companies that buy the chips and build the data centres AI runs on. When the Anthropic CEO suggested the industry should slow down how fast it pushes the most powerful models, chipmakers fell because they sell the picks and shovels. Microsoft rose, because slower spending means less cash going out the door for a company that is already earning money from AI through its cloud business.

Lebenthal owns it and treats its strength as a reason not to turn bearish on the whole market: when the second-largest company in the world is holding up on a bad day, it is hard to argue the market is breaking.

DIS — Walt Disney Positive

Disney makes money from theme parks, cruises, films and its streaming services. An analyst cut his price target to $125 — which, Lebenthal points out, would still be a gain from where the stock trades.

His case is simply price for quality: at about 14 times next year's expected earnings, you are paying a modest price for one of the best-known brands in the world. The stock has risen about 10% since its August earnings while the market went sideways, and he is staying put.

NFLX — Netflix Positive

Netflix had been lumped in with companies that AI might disrupt, and fell hard. Weiss, who had sold it, bought back in — not because he expects a big rally, but because he thinks the worst is priced in.

His reasoning: rival streamers keep raising prices, which makes Netflix look like the good-value option by comparison and protects its subscriber base. He is sitting on a lot of cash and wanted somewhere "safe" to put some of it. It is a "limited downside" trade, and he says so openly.

GS — Goldman Sachs Positive

Goldman is the investment bank most tied to big stock listings (IPOs), so when the AI-safety news raised doubts about Anthropic's and OpenAI's listings, Goldman's shares fell as a proxy.

Weiss owns it and is unbothered: even without the AI IPOs, Goldman has plenty of merger advisory and other high-margin work. He also likes banks into a rate hike because a steeper gap between short and long interest rates lets banks earn more on loans funded by cheap deposits — and banks act as a hedge if the economy stays stronger than he fears.

IAU — iShares Gold Trust Positive

IAU is a fund that holds physical gold, so it moves with the gold price. Gold pays no interest, which normally makes it less attractive when rates rise — and rates are rising.

Thomas buys it anyway because she thinks gold's role has changed back to insurance against geopolitical trouble and against central banks losing control of inflation. With gold under $4,300 after a January spike toward $5,000, she sees a better entry than when everyone was excited. Terranova disagrees on the chart: this year's trend is down, and he thinks the recent bounce is just a short-term blip.

EXPD — Expeditors International Positive

Expeditors is a freight forwarder — it doesn't own planes or ships, it arranges shipping and customs clearance for companies moving goods around the world. More than two-thirds of its revenue comes from outside the US.

Terranova's pick is a momentum trade: the stock is at a 52-week high on a day the market is falling. It is the positive version of the rule he states about AppLovin in the same show — buy what is proving itself, not what is sliding.

LDOS — Leidos Positive

Leidos is a government contractor that does technology, intelligence and defence work, mostly for the US government. Its revenue comes from long contracts that don't depend on consumer spending or interest rates.

Weiss likes it as a defensive holding in a shaky market: the stock has pulled back from its highs, and defence budgets tend to keep flowing whatever happens to AI or oil.

GEV — GE Vernova Neutral

GE Vernova makes gas turbines and grid equipment — the power side of the AI data-centre build. Stocks like this (with Vertiv, Quanta and Eaton) soared because their order books are full for years. The problem is that those orders assume the AI companies keep borrowing and spending.

Terranova still holds them because his fund's rules only let him change positions at scheduled rebalances, but he calls them "more perilous": if AI labs slow down and interest rates stay high, companies will issue less debt to fund data centres, and these orders are exactly what gets delayed. Santoli puts it bluntly — they live off orders "for which nobody has the money to pay" yet.

Anthropic Neutral

Anthropic is a private AI lab heading toward a stock-market listing. Its CEO's essay calling for outside evaluators and international coordination on the most powerful AI models is what knocked chip stocks down on the day.

The investing angle is the IPO. Some investors think the safety stance helps (less legal and regulatory risk), and CNBC's reporter confirmed two straight quarters of "adjusted" profitability — a profit figure that excludes some costs. The committee's point is that the real test is the S-1, the public filing a company must publish before listing: it will show how profitable the business really is once everything it pays for is counted, and what risks (like liability for AI agents) it has to disclose.

AVGO — Broadcom Negative

Broadcom designs custom AI chips and networking hardware for the big cloud companies. Weiss's complaint is not that the business is bad but that the stock has stopped responding to good news: it disappointed on two straight reports, and even when numbers looked fine, investors picked at guidance and margins.

He bought after one report, it barely bounced, he sold at breakeven near $410, and it is now around $346. To him that is the market telling you something — trying to buy a stock that keeps falling ("catching a falling knife") is dangerous when the valuation is still high.

APP — AppLovin Negative

AppLovin runs software that places ads inside mobile apps and games. The stock ran above $600 and has fallen hard; Morgan Stanley just cut its target to $450.

Terranova is a momentum investor: he owns what is going up and avoids what is going down. AppLovin is down about 42% on his 12-month measure, so his rule says don't buy it, however good the business might look. Wait for the price to stop falling and start recovering, then buy — "let it prove itself first." You give up the bottom in exchange for not buying something that keeps dropping.


Summary derived from the public CNBC Halftime Report audio episode (transcript in transcript.txt, read from the Spotify transcript panel in a single pull after disabling its content-visibility virtualization) for personal study. Not investment advice. © CNBC for source material.