Morningstar's analysts now expect earnings at commodity tech hardware companies to peak in 2028, and Dave Sekera is telling investors to sell two of them into the rally rather than wait for it. Most of the market is treating the memory and optical-component boom as the safest way to own AI; Sekera's argument is that every one of those companies is building capacity right now, and that supply always arrives.
"I think earnings drop like a rock, and then on top of that you'll have multiple contraction at the same point in time."
Sekera is Morningstar's chief US market strategist, which means the fair value estimates, star ratings and moat ratings he quotes all come from his own firm's analysts rather than from a screen. He records the show every Monday before the US market opens.
I listened to the full episode so you can skip it. 41 minutes of audio, 21 minutes of reading.
Here are the 13 takeaways that matter.
👤 Guest: Dave Sekera, chief US market strategist at Morningstar, who publishes the firm's stock picks weekly and grades them against his own past calls on air
🎙️ Host: Susan Dziubinski, of Morningstar, who has presented The Morning Filter every Monday before the market opens
📰 Published: 31 August 2026 on YouTube (Morningstar)
🔴 YouTube | ⏱️ 41 min | ✅ Time saved: 20 min
Key Takeaways
A September hike is now more likely than not, and he says the month does not matter
The odds moved from about 40% to over 60% in a week on Kevin Warsh's Jackson Hole remarks
Nvidia's call is the only place to see the whole AI economy at once
It sells to every lab, every hyperscaler and every data-center builder, so the guidance is an industry read
The software selloff was based on a premise the market is abandoning
Nobody replaces a major enterprise platform by writing their own with AI, he says
Cybersecurity spending accelerates into 2027, and the stocks have already run
Morningstar raised fair values on four of them and still rates three as roughly fairly valued
He calls the payrolls report low-quality information and watches only the trend
Commodity tech hardware earnings peak in 2028 on his team's forecast
Everyone is adding capacity, and the operating leverage that worked upward works downward
He got Lululemon wrong and said so on air
The stock has fallen a further 25% since he recommended it
September is historically the weakest month, so he is rotating into defensives
Consumer staples, utilities and healthcare for the buys; technology and consumer discretionary for the sells
A three-star rating is not a hold instruction
On The Trade Desk he says selling part of a position and harvesting the loss can beat waiting
1. Warsh Turned Hawkish
Dziubinski opened on last week's market activity, and Sekera started with the new Fed chair.
The market repriced a hike inside a week. He said Kevin Warsh's statement at the end of the previous week was more hawkish than expected, and that futures now put a September hike at over 60% against about 40% a week earlier
The odds of a hike by year-end are close to certain on his reading. He put it at a 90% probability the Fed hikes at least once by the December meeting
He does not think the timing is the story. "But in my mind, I don't think it really matters." Whether it lands in September, October or December, he said, it will not make a difference in the marketplace in the grand scheme of things
2. Why Nvidia's Call Matters
Sekera's answer to what drove the week was one company, said three times.
"Nvidia and did I mention Nvidia?"
The reason is not the market capitalization. He argued Nvidia matters because it has the most comprehensive view of anyone in AI: relationships with Anthropic, OpenAI and xAI, with all the hyperscalers, and with everybody building data centers
"You know, essentially when you think about AI, Nvidia touches the entire AI economic chain."
That is why he treats the earnings call as an industry report rather than a company report — most companies only discuss the one market they are in
The read-across from the quarter was split. Morningstar remains cautious on commodity-oriented tech hardware stocks even after they fell over the previous month, but thinks the companies at the leading edge of AI, Nvidia among them, have room to run for now
3. Software Isn't Dying
Dziubinski asked about the move in software stocks.
"I think the market is just finally starting to get comfortable with the thought that the death of software you know has been greatly exaggerated"
Morningstar's thesis held through 12 to 18 months of falling prices. Software companies are putting AI into their own products and adding economic value for their clients, he said
He concedes the risk is real at the small end. Smaller, very niche software providers may be at risk
The large platforms are a different case, and his reason is business risk rather than capability. "Clients are not going to replace those software providers by trying to vibe code their own platforms."
Doing so would risk disrupting the client's own business, and he said there are still economies of scale in using an outside vendor rather than building in-house
4. Cyber Runs Into 2027
The same AI thesis shows up on the spending side, and Dziubinski noted cybersecurity stocks had a good week.
AI is expanding cybersecurity budgets separately from the build-out. He said IT budgets for cybersecurity have expanded across the board because of AI, and that it is showing in results and in forward guidance over the past couple of quarters
"In fact, our analyst team is looking for an acceleration of cyber software spending all the way into 2027."
He separates the industry from the prices. Asked whether he is still enthusiastic, he said the business dynamics make cybersecurity one of the most attractive spaces to invest in right now, but the stocks have been on a tear and it is hard to tell which are overextended
The fair value increases were large and the ratings still came down to roughly fair. Morningstar's fair value estimate is its own analysts' view of what a share is worth, and the star rating compares that with the market price
Fortinet raised to $143 from $108, now three stars; the stock has more than doubled since he picked it on 22 September 2025
Okta raised to $200 from $124, now three stars, also more than doubled since picks in 2023 and 2024
Cloudflare raised to $266 from $235, now three stars
CrowdStrike raised to $152 from $133 and left at two stars, which he said is still a little overextended
An ETF he flagged in March under the ticker BUG is up 74% since
"I think the risk here is still more to the upside than to the downside." He said this about the sector's valuations overall, with Palo Alto Networks reporting the next day and Zscaler on 3 September
On what a three-star rating means in practice, he said it is technically a stock trading close to fairly valued on a risk-adjusted basis, and that he would not argue against staying long given the momentum
5. Payrolls Are Low Quality
Dziubinski put it to him that he is known for not caring about the non-farm payrolls report, and asked whether he would watch it anyway.
His objection is to the data, not to the labor market. He said the payroll numbers are generally very low-quality information, especially for thinking about the economy overall
The reason is revisions. They are revised multiple times after release, often by large amounts, so apparent strength or weakness in any single month can be revised away months or years later
"I'm really watching more what the trend is necessarily than any one individual month"
6. What Broadcom Must Say
Dziubinski noted Morningstar's $650 fair value estimate on Broadcom and shares trading well below it, and asked what he wants to hear on the call.
He is not listening for the quarter's numbers. The listening test is whether Broadcom is both keeping clients and adding them
On keeping them, he wants one specific thing confirmed. He wants confirmation that the deal between Marvell and Google is Google multi-sourcing its semiconductor needs rather than a step toward replacing Broadcom as its supplier over time
On adding them, he is going on what the analysts are hearing. Morningstar's team is hearing Broadcom has been bringing more new customers onto its platform, and he wants details
He read his analysts' thesis out verbatim. "Broadcom's growth potential over the next two years is being undervalued with 20 gigawatts of capacity for OpenAI and Anthropic alone creating the potential for 400 billion in cumulative revenue."
This is Morningstar's own analyst note, quoted by Sekera, not his own forecast
The catalyst he is waiting for is a guidance raise, not a beat. Current guidance is for more than $100 billion of AI chip revenue in 2027, and he thinks conservative guidance has held the stock back; an increase, plus comfort that Google and other clients are staying, could be a big catalyst
7. Lululemon: Wrong Timing
Dziubinski raised Lululemon, a pick of his from the 23 March episode, with the stock well below Morningstar's $280 fair value estimate.
He led with the error rather than the thesis. "Yeah, unfortunately, I think I read the situation wrong as far as when I made my first recommendation, you know, into the stock."
The setup looked like a five-star entry and was not. The stock peaked at the end of 2023 and had fallen 68% by the time he first recommended it, at which point it was trading well into the five-star range, with analysts pointing to several catalysts that should support a turnaround
"So obviously, we were very wrong about the timing here." The stock has fallen another 25% since, and over the same period the analysts cut the fair value estimate to $280 from $295
He still thinks the model is undemanding. A 4.6% five-year compound annual growth rate for revenue, which he described as inflation plus a little new product growth; a gradual operating margin recovery with 2026 as the low; and 12% earnings growth on a five-year compound annual basis
The valuation is what keeps him in it. The stock trades at 10.8 times Morningstar's 2026 earnings estimate and 9.5 times the 2027 estimate, so any sign of acceleration in North America would be welcomed and the low starting point gives it room to run on good news
8. Nvidia Beat and Guided Up
Dziubinski noted the stock rose more than 8% after earnings and Morningstar raised its fair value estimate by $30 to $310.
The quarter beat the company's own guidance by a wide margin. Second-quarter revenue of $96 billion, up 106% year over year, against guidance of $91 billion
The next quarter is guided to $108 billion, an 89% year-over-year increase, against consensus coming into the quarter of $105 billion
The number that mattered was the year after. New fiscal 2028 revenue guidance implies a 70% increase, which he said puts revenue at about $700 billion on Morningstar's numbers, against their prior estimate of $570 billion
The one negative was gross margin, and he dismissed it. Guidance takes it from 75% to 74% this quarter, as low as 71.5% in the January quarter, and 72.5% for fiscal 2028 — which he attributed to the sharp rise in memory prices and said does not concern him from a valuation perspective
The number he pulled out for the industry was the customers' budget. Nvidia's own forecast is for its top five US hyperscaler customers to spend $1.3 trillion on AI capital expenditure next year, a large step up from this year
On the purchase and lease commitments the market has been worrying about, he said he spoke to Brian Colello, Morningstar's analyst on the company, who is still comfortable with Nvidia's disclosure and rationale; across the firm's AI coverage the team thinks the industry is growing into those obligations
The stock trades at a 30% discount to fair value, which he said puts it well into four-star territory and still a buy
9. Marvell Fell on Good News
Marvell fell 10% after what Dziubinski described as good earnings.
"Yeah, honestly, I'm not sure why the market puked all over Marvell stock, you know, like they did."
"I think this is just one of those cases where the good results just weren't necessarily good enough for the marketplace."
The results themselves were strong on his account. Revenue up 37% year over year, fiscal 2028 guidance raised by 10%, and management pointing to stronger growth beyond fiscal 2028 than it had described before
Morningstar's thesis is about breadth of silicon. A differentiated portfolio across data-center needs, with broad-based demand in custom chips, interconnect and switching products, plus a recently announced agreement with Google
They moved the fair value the opposite way to the stock. Raised to $300 a share from $270, on a higher medium-term growth forecast taken from this earnings report, leaving the stock at almost a 30% discount and well into four-star territory
The next catalyst he named is a date. Marvell's investor day on 6 October, where he is looking for financial targets through calendar 2030 and more detail on the ramp of the Google chip business
10. Salesforce Vindicated
Salesforce rose more than 20% after earnings.
He put it down to the market catching up with a view Morningstar already held. Another quarter of solid top-line growth, revenue up 11%, toward the upper end of past guidance
The signal underneath was demand for the AI products. He said there are more and more signals on demand and adoption, and that the firm is looking for revenue acceleration in the second half
He quoted the analyst note on why the Anthropic deal matters. "Salesforce's newly-announced Anthropic partnership reinforces Salesforce's strategy of serving as the governed data and workflow layer underneath multiple AI models and interfaces. Customers will be able to use Claude on top of Salesforce without replacing Salesforce as the system of record."
He refused the credit for the call. Dziubinski noted the stock is up more than 50% since he picked it on 29 June. "Well, first of all, I'm not looking good on this one, Susan. I would say our analyst and our analyst team is looking pretty good on this one."
His advice on it now is to let a winner run. At a 7% discount to the $280 fair value estimate and a 0.7% dividend yield it is a three-star stock, but he said it may not be time to sell: still below fair value, still with strong upward momentum, and many investors probably underown software
His own speculation, flagged as such, is that fair value increases may come over the next couple of quarters
The check on that is the multiple — 18.5 times Morningstar's 2026 earnings estimate against a 13% five-year earnings growth rate — which he said is starting to get pretty full
11. The Trade Desk Question
The viewer question came from JC, who asked whether to give up on The Trade Desk after a 64% fall this year and a fair value estimate cut from about $60 at the start of the year to about $16.
Morningstar had been negative on it for years. He said it has been a sell recommendation a number of times, and was one of the most overvalued stocks against Morningstar's valuation across the firm's entire coverage as recently as December 2024
The reasons were specific and mostly not about AI. High-profile fee disputes with major ad agencies — Publicis, WPP and Omnicom — rising competition and lost market share, plus a concern about how AI might disrupt the business
The fall has been most of the way down. From about $120 a share in December 2024 to $13 and change now
His current model gives the company credit rather than writing it off. Revenue growth of 3.7% on a five-year compound annual basis, operating margin expanding to 18.7% by the end of the forecast period from 13.2% this year — which he noted is at the high end of a five-year range that has run from 7.2% to 20.3%
The valuation is still not cheap on those numbers. Earnings are expected to decline this year, which he thinks is the low, then grow at 12.5%, while the stock trades at 17.5 times this year's earnings against a $16 fair value estimate
He used the question to correct a common misreading of the rating. A three-star stock, for a long-term investor, should be expected to return the company's cost of equity if it performs in line with the forecast — here 10%, on a very high uncertainty rating
His answer is to sell into it rather than hold for the rating. "So in this case I think this could be a pretty good opportunity to sell, or at least a portion of it, even though it is a three-star rated stock"
The reason is that the whole approach is buying at a significant margin of safety below fair value, which cushions a wrong thesis and makes the position fall less than the market in a general selloff
The capital losses can offset gains elsewhere, which he said answers one of the most common questions he gets — what to do with positions carrying hundreds or thousands of percent of gains and a tax bill attached
If upward momentum takes it into two-star territory, he said that would be a better exit
12. September's Buys
Dziubinski turned to the picks, and Sekera explained the screen before the names.
The premise is seasonal and he flagged it as such. "September historically is the weakest month of the year for the US market." "Now, of course, past performance not an indicator of future performance, yada yada yada, all those kind of disclaimers", he added, and said he is not forecasting a selloff — only positioning for one
The screen was his usual one. Four- or five-star stocks with an economic moat — a durable competitive advantage, ideally a wide one — preferring low or medium uncertainty ratings, though he will buy high-uncertainty names with enough margin of safety
The final cut was visual. He looked at the charts of what the screen returned and picked the ones that looked like a good setup for a general pullback
Mondelez, and Emerging Markets
The sector call needs separating from the sector's valuation. "From a sector perspective, we think the sector is overvalued because of Walmart and Costco." Those two skew consumer defensive too high, he said, while many individual names in it are undervalued
Historically the sector holds up. Investors rotate into defensives in a September selloff because they fall less
The stock is a four-star name at a 19% discount with a 3.3% dividend yield, a low uncertainty rating and a wide economic moat, first picked on 27 October 2025 and re-recommended in January
The fundamentals are better than the food group's. Last quarter's organic sales growth was 2.2%, with North America at 3.4% and improving sequentially; Europe is still weak and he would like to see it at least stabilize
"But when I think about Mondelez and why Mondelez is my top pick, its emerging market exposure is still the basis for why I think it's the best pick of the food names."
Emerging markets are 40% of total sales and grew 7.4% in the second quarter, and he expects that to be where the food sector's growth comes from
The model is a 3.3% five-year revenue growth rate with margin expansion back toward historical averages and 9.2% earnings growth from 2028 through 2030, on 20 times this year's earnings falling to 18 times next year's
Alliant, a Second Bite
Utilities are the classic defensive holding — stable earnings and high dividend yields — and he said the sector has generally outperformed the S&P 500 in September
A four-star stock at a 12% discount with a 3.2% dividend yield, a low uncertainty rating and a narrow economic moat, first recommended on the 12 January episode
"And I think the market is just giving you kind of that second bite of the apple on this one." The stock ran from $65 to $78 on rising interest rates and has fallen back to $68, which returns it to four-star territory from three
What it owns is regulated. It is the parent of two regulated utilities, Interstate Power and Light and Wisconsin Power and Light
He is forecasting the top of management's own range. Earnings growth at the high end of the 5% to 7% guided through 2027, accelerating after that on a four-year capital investment plan that is 24% larger than the prior one, with data-center construction adding to it
Baxter, and a 35% Margin
A four-star stock with a 35% margin of safety — the gap between Morningstar's fair value estimate and the price — and a narrow economic moat based on switching costs and intangible assets
The dividend is not the reason to own it. He noted the company cut it about a year ago, to something closer to a cent a share
The second quarter was the turn. Revenue and earnings both came in much better than expected and management raised guidance, giving a 2026 earnings range of $1.95 to $2.15 a share — which puts the stock at 12.5 times the midpoint
The analyst's read is about management execution. Morningstar's analyst sees early success relative to expectations from the new management team on operating efficiency
The model looks conservative to him. A 3.4% five-year revenue growth rate and operating margins expanding to 12.5% by 2030 from a 6.2% low this year, which only returns the company to historically normal levels
The chart supports the entry. He said the stock looks to have bottomed earlier this year and is making higher lows
13. September's Sells
The three sells come from the sectors that have historically underperformed in the month.
Technology sells off furthest and fastest in a general market decline, because high-growth stocks do, though he said Morningstar still likes a number of individual AI names
The second risk to technology is rates, and he gave a level. "If the 10-year were to start hitting like a five handle and go over 5%."
"Growth stocks have very long duration." In plain terms, most of the cash flow that justifies the valuation arrives many years out, so discounting it at a higher rate cuts what an investor will pay today by more than it does for a company earning now
Consumer discretionary is the other sector to avoid, both because it falls faster than the market and because it is more economically sensitive; a negative wealth effect from falling markets would hit those stocks hard
SanDisk and Ciena Premiums
Both are rated two stars and both trade above fair value. SanDisk at almost a 50% premium, with a very high uncertainty rating and no economic moat, already off its highs and rolling over; Ciena at a 40% premium, very high uncertainty, with a narrow moat
His objection is what the businesses are. He said both are generally commodity-oriented tech hardware, and that supply will eventually catch up with demand
The capacity is being added now. Companies are redesigning and redeploying assets toward the highest-margin, highest-demand products and building new facilities
"In fact, our analyst team is forecasting that 2028 will be the peak as far as earnings for all of these tech commodity type of companies."
The operating leverage runs both ways. "I think earnings drop like a rock, and then on top of that you'll have multiple contraction at the same point in time." Once supply is sufficient, prices fall and margins contract
Brinker at a 66% Premium
The reason it appears now is the run. Dziubinski noted they have not discussed it before; Sekera said it is up to a 66% premium to fair value, well into one-star territory
The ratings make the premium harder to justify, not easier. A medium uncertainty rating means Morningstar thinks it can model the company's free cash flow further out than most, so he does not think the stock should trade that far above fair value
The narrow moat is already in the model. It rests on fixed-cost leverage from spreading costs across a large restaurant chain, and it means the model already assumes excess returns on invested capital for at least the next ten years before competition erodes them
The margin assumption is the fragile part. Morningstar models 4% five-year revenue growth, peak operating margins of 10.6% this year fading to 9.9% by 2030 — against a 6.5% historical average
"I mean, the restaurant business historically is just a low margin business, a very tough business." In any economic downturn he would not be surprised to see those margins evaporate quickly
"So market is pricing in significantly more growth, significantly more operating margin expansion than what we currently forecast." Earnings growth after this year is modeled at 7.3% and the stock trades at 21 times earnings
Bonus Insights
The show is off for a week. Dziubinski said there is no episode on 7 September for the Labor Day holiday, and that the next one is on 14 September, ahead of the Fed meeting
Sekera lost his voice partway through the software answer and stopped to apologize and reach for more coffee before finishing the point
Sekera's bottom line is that the AI trade has split in two: the companies at the leading edge and the software vendors that embed AI still have room, while the commodity hardware suppliers are being priced as though capacity will never arrive — and with September historically the weakest month, he would rather own staples, utilities and healthcare into it.
Products, Companies & Tools Mentioned
Nvidia (The quarter that set the tone: $96B of revenue, $700B implied for fiscal 2028, and a $1.3T hyperscaler capex forecast; still a buy at a 30% discount on Morningstar's numbers)
Broadcom (Undervalued on his analysts' view, with 20 gigawatts of OpenAI and Anthropic capacity behind a possible $400B of cumulative revenue; the catalyst is a guidance raise)
Marvell (Fell 10% on results he thought were good; Morningstar raised fair value to $300 from $270 and is watching the 6 October investor day)
Salesforce (Up more than 20% on 11% revenue growth and AI adoption signals; his analyst says the Anthropic deal makes it the governed data layer under other models)
Anthropic, OpenAI and Google (The counterparties running through the whole episode — Broadcom's capacity, Marvell's custom-chip deal and the Claude-on-Salesforce partnership)
Lululemon (The pick he says he got wrong on timing; down a further 25% since, at 10.8 times Morningstar's 2026 estimate)
Fortinet, Okta, Cloudflare and CrowdStrike (The four cybersecurity names whose fair values Morningstar raised this earnings season; the first three now three-star, CrowdStrike still two)
Palo Alto Networks and Zscaler (The two cybersecurity companies reporting in the week ahead)
BUG (The cybersecurity ETF he pointed to in March for investors who did not want single-stock risk; up 74% since)
Mondelez (His top pick among food names: four stars, 19% discount, wide moat, and 40% of sales in emerging markets growing 7.4%)
Alliant Energy (Four stars at a 12% discount after falling back from $78 to $68; parent of Interstate Power and Light and Wisconsin Power and Light)
Baxter (Four stars at a 35% margin of safety, on a second quarter that beat and a raised 2026 earnings range of $1.95 to $2.15)
SanDisk and Ciena (The two sells: 50% and 40% premiums to fair value on businesses he calls commodity tech hardware, with the earnings peak modeled in 2028)
Brinker International (A 66% premium and one star; the model already assumes margins well above the company's 6.5% historical average)
The Trade Desk (The viewer's question: down 64% this year, from about $120 in December 2024 to $13, against a $16 fair value estimate)
Walmart and Costco (The two stocks he says make the consumer defensive sector look overvalued while the names underneath it are cheap)
Publicis, WPP and Omnicom (The ad agencies whose fee disputes with The Trade Desk were part of Morningstar's original sell case)
Books & Resources Mentioned
Morningstar (The source of every fair value estimate, star rating, moat rating and uncertainty rating quoted in the episode; viewers are pointed there for detail on the names)
If this was worth your time, send it to someone who has to have a view on this.
Get the latest market chatter as it happens: