CNBC Halftime Report — The Return of the "Magnificent Seven"

Nvidia makes the chips that AI systems run on. Its results a week earlier were strong, but the rest of technology did not move on them straight away. Lebenthal's argument is that the delay was just a lag, and that the earnings functioned as a "clearing event" — a result definitive enough to settle an open question for everyone in the sector, not only for the company reporting.

The question it settled was whether the enormous spending on AI data centres is going to continue and whether it earns its keep. His evidence is other people's numbers, not Nvidia's: Amazon's chief executive saying a new data centre pays for itself in under three years, Alphabet's cloud business growing 80%, Microsoft's Azure 40%. If the customers are making money on the equipment, the orders keep coming.

Simpson makes it his final trade on the back of the Hugging Face acquisition. Worth noticing what nobody claims: neither of them argues the stock is cheap. The argument is that the spending is durable.

Microsoft is up 27% from the day it reported, having gone essentially nowhere for the previous year. Sechan's case has two parts, and the second is the more useful one.

First, size: it is one of the heaviest weights in the index, so when it moves the whole market moves with it — which is part of why the Mag 7 chart turned. Second, and this is the part that matters for anyone holding it, the price rose while the valuation fell. A flat year of earnings growth with a flat stock means the multiple — what you pay for each dollar of profit — compressed. So the rally started from a cheaper base than the chart alone suggests.

That is the same test Simpson applies to the whole group: a rise that comes with rising earnings is a re-rating you can hold; a rise that comes only from people paying more for the same profits is a rally you have to time.

Meta rose almost 6% on the week after releasing a new AI model, with Bank of America reiterating a buy at $810 — roughly 200 dollars above where it trades.

Sechan's specific point is the valuation gap: at about 18 times earnings it is the cheapest of the big cloud-and-AI companies, which is unusual for a company the market has just decided is winning. His rule for acting on that is worth separating from the enthusiasm — he is overweight already and would add "into the weakness," not into this week's strength.

Weiss adds the strategic turn that changed the story: the market expected Meta to push into renting out cloud capacity, and instead it said it would sell access to its AI models. Simpson's addition is the one a competitor cannot copy quickly — Meta already has billions of users to distribute anything it builds to, so a new product does not need a new sales channel.

Apple is about to launch its most expensive iPhone ever — a reported $1,319 for the top model against $1,099 last year — because the memory chips inside it have become dramatically more expensive. The company is passing that cost on rather than absorbing it.

The question the desk is really debating is elasticity: how much demand you lose for each dollar you add to the price. Lebenthal's answer is empirical — Apple has raised prices repeatedly and people have kept buying — with one structural support, the phone carriers who subsidise handsets to win customers. Sechan's is strategic: Apple owns the customer relationship, and in past technology waves the companies that owned the customer, rather than the ones that built the plumbing, captured the profits. That is why the market lets Apple trade at a premium.

Weiss is the sceptic, and his two objections are both checkable rather than atmospheric. App Store revenue growth is decelerating, per a Bank of America note. And the phone's headline feature — a genuinely improved Siri — has been promised and not yet delivered. If a much more expensive phone launches without the AI that was supposed to justify it, the pricing test gets much harder.

Oracle rents out computing power to companies building AI systems. Lebenthal's final trade is four words — "Oracle's got good momentum" — and the change from two sessions earlier is worth noting, because it is a change in the kind of reason he gives.

Then, he held the stock through a painful stretch on a fundamental argument: the world is short of computing capacity, so Oracle will benefit whatever the share price does. Now he is naming price behaviour instead. Momentum arguments and value arguments have different exit rules — a value thesis survives a falling price, a momentum thesis does not — so it is worth knowing which one you are actually holding.

KLA makes the inspection equipment that finds defects on semiconductor wafers during manufacture — a picks-and-shovels business inside the chip industry rather than a chip designer.

Sechan's whole pitch is that it is still 30% below its June high. That is a deliberate contrast with everything else discussed in the hour: the Mag 7 names have recovered and he says he would rather buy weakness than strength. This is him doing exactly that — reaching for the part of the semiconductor complex that has not participated.

FTAI's original business is aircraft engines — leasing them and refurbishing them — and it has been pushing into generating power for data centres. That second part is why Weiss thinks it is being mispriced.

His claim is narrow and testable: the political backlash against data centres is real and is hurting the companies that build and power them, but it is being applied to this stock indiscriminately, when the company's actual exposure does not warrant it. "Data center debate is misplaced with this stock. It's very cheap." Earlier in the hour he describes the same holding as having had good news and still trading down to 184 on nothing but the debate.

One caveat travels with the earlier reference: the automatic transcript drops the company's name from that sentence entirely, so tying the "184" line to FTAI is an inference from his own final trade in the same episode and his known position in the name. The final trade itself is verbatim; the earlier attribution is probable, not certain.

Freeport is one of the largest copper miners. Simpson bought more of it, and the reason he gives for preferring copper to gold is the useful part.

Gold is bought mostly as a store of value — it does not get consumed, so its price depends on what people want to hold. Copper gets used up: it goes into electrical wiring, into the grid, into data centres, into electric vehicles. So a copper position is a bet on physical demand you can point at, whereas a gold position is a bet on sentiment and interest rates. If you believe the data-centre build-out continues — which the same panel spent the first half of the hour arguing — copper is the more direct way to own it.

His discipline here is also worth copying: he adds on a pullback in an existing position rather than initiating something new because it is going up.

Agnico Eagle is a large gold miner. A miner is a leveraged way to own the metal: its costs are broadly fixed, so a move in the gold price shows up amplified in profits, in both directions.

What makes this entry unusually honest is that Simpson names, on air, the single thing that would prove him wrong — and it is a call his own colleague made twenty minutes earlier. Gold pays no interest, so it competes with cash and bonds; when interest rates rise, holding gold costs you more in forgone yield and the price typically suffers. "If Jim's right about rate hikes… then forget the gold trade. I'm completely off base with this."

That is the shape a position should have: an argument, and a specific, observable event that invalidates it. Citi's note supports him for now — bullish gold and silver into year end, and more resilient than copper or aluminium to rising energy costs.

CF Industries makes nitrogen fertilizer, which is manufactured largely from natural gas. Simpson bought it the previous week as the opening position in a broader "hard commodities" theme; the Freeport and Agnico adds this week extend it.

No separate case was made for the company itself here. Its role in the episode is to show that the metals purchases are a programme with a thesis behind it — rotating money toward physical goods — rather than two opportunistic dip buys.

Lebenthal's argument is not about the oil price at all — it is about business structure. Exxon is integrated: it finds and produces oil, transports it, refines it into fuels, makes chemicals from it, and sells it at retail. Those pieces do well at different points in the cycle, so when crude falls the refining and chemicals arms cushion the fall.

That is why his rule is "if you're starting to build your energy portfolio, you've got to start with ExxonMobil" — it is the position that requires the fewest correct forecasts. The supporting fact he cites is a six-year record of roughly doubling the S&P 500's return, in a sector most investors treat as a trade.

Chevron is the other US integrated major, and Simpson prefers it to Exxon on two operating numbers rather than a story: production up 20% in the latest quarter, and about $1.5 billion of cost savings already captured from its Hess acquisition.

"Synergies" is the word companies use when promising that a takeover will produce savings; most of the time it is a forecast. The reason this one carries weight is the verb — captured, not targeted. It has already shown up.

Baruch sold a high-end sporting-goods company and bought Walmart with the proceeds. The logic is a "trade down": when consumers get squeezed, they shift spending toward cheaper retailers, so the same economic weakness that hurts the first position helps the second. It also fills a hole — his portfolio was underweight consumer staples entirely.

His entry is a level, not a feeling. The stock is 25% off its high and sitting at $100, a price where an enormous amount of stock changed hands in 2025 — areas like that often act as support, because the people who bought there tend to defend the position. He also names his disaster plan in advance: the 200-week average sits at $80, he does not expect it, and he would buy heavily if it happened.

The operating case is three specifics: 94% of the US population reachable within three hours; 30% of same-day orders paying a fee that expands margins; and two competitive claims — that Walmart is quietly taking share from Amazon's core business, and that Sam's Club matches Costco on same-day delivery at a lower valuation.

Robinhood is the retail brokerage app. Its share price has historically tracked bitcoin closely, because a lot of its trading revenue comes from crypto.

Simpson does two things here that are worth separating. He likes the company and he tells you not to buy it today — after a sharp move and a broker target raise, "I'd be careful chasing it here." Liking a business and liking its price on a given morning are different judgements, and most commentary blurs them.

His longer thesis is that the bitcoin correlation is a legacy, not a destiny. Tokenization — issuing ordinary assets as blockchain tokens so they can be traded in smaller pieces, around the clock — is a genuine brokerage business rather than a crypto bet. His line for it: ten years out, "what, Bitcoin doesn't exist? Robinhood will."

Renick's segment is about the options market rather than the company, and the insight is a mismatch. Options are contracts giving the right to buy or sell later at a fixed price, and their cost rises with how much movement traders expect. Enormous volumes are trading in SpaceX, but because the stock has sat in a ten-point range for a month, the options have been priced for a sleepy stock. He thinks that is wrong: a share this contested does not stay quiet, and when it moves it will move hard.

The positioning detail tells you which way the pain runs. There are slightly more puts (bets on a fall) than calls outstanding, but nearly all of them are struck below the current price, and almost nobody has bought protection below $142. So the downside beneath that level is unhedged — if it breaks, there is nothing there to slow it. Sellers, in Renick's phrase, need to move faster than buyers do.

On the desk, both owners decline the trade. Simpson holds it in a growth portfolio and says the timeframe is years, "not weeks or hours." Sechan has owned it privately for a long time and now publicly too, selling into strength when liquidity windows open, and describes the shareholder base as "evangelical" — devoted enough that good news produces outsized moves. Lebenthal's rejoinder is the one to keep: devotion is not a floor. "That evangelical support at times withers — and look at Tesla over the last week or so."

Adobe makes creative and document software. Lebenthal had been holding it for a specific event — the naming of a new chief executive — on the theory that an outsider, particularly one open to selling the company, would force the market to notice how cheap the shares had become. The company named an insider, and the stock fell 6%.

His response is the honest version of a failed catalyst trade: he says plainly that the market wanted an outsider, that he did too, and that the re-rating he was waiting for "is clearly not going to happen with the selection of an insider." He is not selling, but the reason for owning has changed from an event to the ordinary case — around 10 times forward earnings, growing free cash flow, a share count down roughly 10% in eighteen months, with next week's results as the test.

The exchange it triggers is the more transferable content. Wapner calls the buyback story "financial engineering." Lebenthal's distinction: buying back shares with borrowed money, or by starving research, is engineering; doing it out of cash flow that is also funding research is just capital allocation. Weiss's version is sharper than either — buybacks are not automatically bad, but shrinking the share count by 10% flatters earnings-per-share growth by 10%, so ask what the underlying growth was without it. And he raises an uncomfortable possibility about the CEO search: a wide net that lands on an insider may mean the outsiders did not want the job.

Suncor is a Canadian integrated oil producer, and Sechan makes an unusually specific claim for why it belongs in a portfolio right now: it has been a hedge against rising interest rates. Most equities fall when yields rise; energy has been going the other way, so the position offsets something else he owns rather than just adding to it.

Then he says he is inclined to take chips off the table — not because the reasoning changed, but because the position has run. That is a rebalancing decision rather than a change of view, and it is the reason the net stance here is neutral despite the compliments.

GE Vernova makes turbines and grid equipment — the hardware that powers data centres. Weiss's read has softened from two sessions earlier: it is "well off its highs, making a comeback," but "much more volatile" than the mega-cap technology names.

That volatility is the point of the mention. Demand for the equipment is not in doubt; what moves the price is politics. Local objections to data centres, and the electricity bills residents blame them for, create headline risk that arrives without warning and has nothing to do with the order book. Wapner's request for a chart of "data center backlash versus AI power names" is only half a joke — the two lines have been mirror images.

So the same investor can hold both this and the Mag 7 and treat them differently: one is exposed to earnings, the other to permitting and politics.

Caterpillar makes the heavy equipment that builds infrastructure, including data centres, and it is the second half of Weiss's pair of names that are "well off their highs" while mega-cap technology recovers.

It is cited to explain the rotation rather than to make a case either way. Money went back into the largest technology companies because they offer known businesses, cheap debt and defensible valuations; the build-out names offer the same underlying demand with an extra layer of political risk attached, and the market is currently paying up to avoid that layer.

An analyst trimmed the price target on Costco by $50, to $1,144. Simpson owns it and dismisses the cut on arithmetic — $50 on an $1,100 stock, with the target still $100 above the price, is not information.

What he does concede is the relative call. Asked to choose between Costco and Walmart he picks Walmart, and says he already holds more of it. Both are expensive membership-and-scale retailers, and he is not arguing the multiple is wrong; he is saying that if you are paying a high multiple for scale, the cheaper of the two is the better expression. Baruch reaches the same conclusion from the other side, arguing Sam's Club now matches Costco's same-day delivery at a lower valuation.

Amazon appears twice without anybody taking a position on the shares, and the two mentions point opposite ways.

As evidence, its chief executive's claim that a new data centre pays back in under three years is one of Lebenthal's three proofs that AI spending is profitable rather than speculative — a payback period is the plainest possible test of that.

As a risk, Baruch's Walmart case rests partly on the argument that Walmart is "eating into Amazon's core business largely unnoticed." Both can be true: the cloud business compounding while the retail business faces a sharper competitor.

Alphabet is named once, as a number rather than a position: its cloud business growing 80%, alongside Microsoft's Azure at 40%, in Lebenthal's list of reasons the AI build-out is earning a return.

The relevance is to the sector argument, not to the stock. If the companies renting out AI computing power are growing at those rates, the spending that created the capacity is being absorbed by real demand — which is the difference between a build-out and a bubble.

Tesla held an event for its Cyber Cab — a purpose-built self-driving taxi — and the stock fell 6.5%. The safety regulator is separately reviewing the rollout.

Simpson owns it and does not defend the event, which makes his explanation more useful than a defence would be. A large part of what Tesla is valued on is not cars: it is robotics, autonomy and a future robotaxi fleet. Those are promises with no revenue attached yet, so the only evidence available is demonstrations. A disappointing demonstration therefore does real damage — "all this does is delay the reality of that coming to fruition, if it ever does."

His two specifics are worth holding on to. Tesla's approach to self-driving relies on cameras and software rather than the additional sensors Waymo uses, so the two are not interchangeable and progress in one does not validate the other. And the test that actually settles it is commercial, not technical: vehicles operating with paying customers.

Zscaler sells cloud-based network security. It beat expectations, guided up, and the stock fell — which is why Wapner uses it rather than a name that missed.

The pattern across the week is the lesson: Zscaler down on a beat, Palo Alto down 10% on a report that "wasn't bad," Fortinet down, the software index off 4%. When good results stop producing good reactions across an entire group, the variable being repriced is not the businesses. It is how many people already own them — and a crowded position has nobody left to buy it.

Palo Alto is the largest cybersecurity company. Two sessions ago it reported record results and fell 10.5% in a day; here the only update is that the weekly loss is still about 10% — it did not bounce.

That confirmation matters more than a fresh opinion would. On the day of the print, the reading offered on this desk was that positioning was full and the unwind "will get worked off over the course of time," not in one session. A stock that stays down all week after a record quarter is the evidence for that view rather than against it, and nobody stepped up with a bid.

Fortinet sells firewalls, the closest comparison to Palo Alto's core business. It gets two words — "Fortinet down" — and that is the whole content.

It belongs in the table anyway, because the read-through it did not get is the story: the sector leader posted a record quarter and its nearest peer fell with it rather than rising. That is what factor-driven selling looks like from the inside.

TJX runs TJ Maxx, Marshalls and HomeGoods — off-price retailers that buy surplus branded inventory and sell it cheaply. Simpson trimmed a long-held favourite, and how he got there is the transferable bit.

The headline numbers beat on both revenue and profit, and the stock still behaved badly. Rather than concluding the market was wrong, he looked underneath and found the reason: Marmaxx, the core US division, grew comparable sales 1% against 6% the previous quarter. A beat driven by everything except the main business is a weaker beat than it looks.

He then argues against himself out loud — this could be seasonal, and in a two-tier economy both the squeezed and the comfortable shop at off-price stores — and trims anyway, because the chart is breaking down. That order of operations is the discipline: form the view, check the specific line that would falsify it, and let price arbitrate when the two disagree.

Nobody on the desk would buy it, and Lebenthal's explanation is a claim about the category rather than the company.

Branded apparel used to be a cyclical business: fashion turned, a brand caught the trend, the stock went up, and you could trade around it. His argument is that this has become something else — "a secular decline in these businesses that they can't get out of the way of." Secular means it does not turn back with the cycle.

The suspected cause is the market that was supposed to be the growth story. China was where every Western brand was going to expand; instead it has produced competitors of its own that are taking the share. If that is right, no amount of a good season fixes it, which is what makes "no touch" a category verdict rather than a price call.

Nike is named twice, by two people, on the same side — grouped with Lululemon in Lebenthal's no-touch list, and used by Baruch as proof that the weakness spans the whole price range from premium to mass market.

That breadth is the argument. If the top end, the middle and the discounters are all struggling at once, the problem is unlikely to be any one company's execution. Nobody offered a level to buy or a catalyst to wait for.

On Holding is the Swiss running-shoe brand. Lebenthal confirms he is out — sold in April — and folds it into the same verdict as Lululemon and Nike.

Against this archive's own record that is a hardening. The last substantive discussion of the name, after a 19% single-day fall, still had a long-term owner defending the shift toward apparel and direct-to-consumer sales as the higher-margin part of the business. Here the individual defence has been dropped in favour of the category one: the whole group has stopped behaving like the stable business it used to be.

Amer Sports owns Wilson, Arc'teryx and Salomon. Baruch sold the entire position, and the striking thing is that he still likes every part of the business — this is a sell that contains no criticism of the company.

What broke was the reason he bought it. The position worked because Chinese consumer stimulus was driving demand for premium brands; tariffs then ate the benefit and squeezed margins on top. When the thesis that justified a position stops being true, the position is unjustified regardless of how good the brands are.

Wapner presses the obvious objection: these are luxury brands sold to the top of a two-tier economy, so why should a weak consumer matter? Baruch concedes the point and redirects it to margins — the tailwind that took the stock from $20 through $30 was a theme, and the theme is gone. Note the mechanics he actually acted on: he waited through the most recent results, saw a good report, watched the stock go down anyway over the following fortnight, and only then stepped aside on a $19 cost basis. "I don't want to hold on and see that continue to deteriorate."