SUCCESS

There are only a handful of industries in America where the infrastructure advantage is so deeply embedded that new competitors, regardless of their capital or tech, face a decade-long climb to approach parity. Telecom is one of them.But SpaceX’s Starlink is not a typical competitor. And AT&T (T) CEO John Stankey knows it.Stankey appeared on CNBC’s Squawk Box on Wednesday, July 22, to discuss AT&T’s strong second-quarter results and field questions about the satellite threat directly. And his answers were measured, confident, and revealing. AT&T closed the week on Friday, July 24, at $24.13, up 5.10% on the session following earnings, according to Yahoo Finance. The company reported Q2 revenues of $31.6 billion, up 2.3% year over year, and announced an accelerated $10 billion share buyback program for 2026.Stankey’s core message on Starlink was that competitors are welcome. Yes. But they are arriving very late to a party that AT&T has been hosting for decades.Also Read: History of AT&T: Timeline and FactsWhat AT&T CEO actually said about Starlink — the infrastructure argumentStankey did not dismiss satellite competition. In fact, he contextualized it in a way that tells you how AT&T thinks about the threat internally.There are going to be new competitors, and they’re going to be folks that come in. But the reality is that they’re coming to the game very late after this industry has been established.Stankey continued on Squawk Box. “They have to catch up with substantial amounts of infrastructure investment that’s been going on for decades inside hospitals, on university campuses, in stadiums, in tall buildings.”That last sentence matters the most, and here is the reason why. Indoor coverage is the problem satellite cannot solve with the same economics as terrestrial networks. More AT&T:AT&T stock price target cut puts dividend investors on alertAT&T may be left out of the Starlink deal everyone wantsAT&T leaves rivals flat-footed as bankrupt carrier foldsEvery high-rise apartment building, every hospital basement, every stadium concourse represents infrastructure that Starlink cannot serve from 340 miles above Earth. AT&T handles more than 98% of traffic generated by its converged customers on terrestrial networks today, according to the Interview. Satellite addresses the remaining fraction of time a customer walks off-grid. In fact, Stankey says AT&T will address that by next year through the partnerships they’ve already established.Related: AT&T may be left out of the Starlink deal everyone wantsThe scale comparison supports Stankey’s confidence. AT&T generated $31.6 billion in revenue in Q2 alone. According to a Reuters report, Goldman Sachs projected full-year 2026 Starlink revenue is approximately $15.6 billion. AT&T notes that it has more than 100 million U.S. consumers across mobile and broadband. As noted by Idem Est Research & Advisory, Starlink has approximately 12 million globally, as of June 2026.The wholesale strategy and why AT&T is not signing with Starlink as a main partnerStankey drew a specific line on wholesale network agreements that has direct implications for how AT&T approaches Starlink and the broader satellite ecosystem.We do wholesale agreements when we think there’s a part of the market that we can’t address with our distribution, our brand and our product.In U.S. suburban and metropolitan markets, AT&T can address those customers itself. The U.S. is also more disciplined than European markets, Stankey noted, precisely because American carriers have “very robust distribution, very well recognized brands, very pervasive infrastructure.”Rather than a bilateral deal with any single satellite operator, AT&T prefers a consortium approach.Related: Oppenheimer downgrades AT&T stock on SpaceX threat”We want to partner with everybody in the satellite ecosystem,” Stankey said, explaining that aggregating volume across multiple low-Earth orbit constellations, including AST SpaceMobile, Amazon Kuiper, and SpaceX, gives AT&T coverage for the small percentage of off-grid traffic at economical pricing without creating dependency on any single provider.This matters for investors tracking the AST SpaceMobile story, which I covered when Cramer called it a buy for the two-year horizon. AT&T’s consortium framing suggests the relationship is complementary but not exclusive, which limits both the upside and the risk for either party.

AT&T is raising prices by $5 per month for fiber and copper internet plans, starting August 16, targeting customers established between June 2024 and July 2025.Kevin Carter/Getty Images

AT&T’s Q2 results and the buyback signalThe financial performance underneath Stankey’s July 22 Squawk Box commentary was genuinely solid, according to AT&T’s earnings release.Revenue of $31.6 billion, representing growth of 2.3% year over yearAdjusted EBITDA reached $12.3 billion, up 5.2% year over year. Free cash flow was $4.7 billion, up from $4.4 billion. Advanced Connectivity service revenue of $23.5 billion grew 5.1%, with Advanced Connectivity operating income up 20.3%. Added more than 646,000 total internet net adds, including 367,000 fiber and 279,000 fixed wireless. Postpaid phone churn was 0.86%.The $10 billion buyback acceleration is the number that sent the stock up 5.10% on the last session of the week. AT&T returned $4.1 billion to shareholders in Q2 alone, including $2.2 billion in common share repurchases. The company is also on track to reach 40 million total fiber locations by the end of 2026 and 60 million by 2030, according to the same statement.On the consumer side, my colleague reported that AT&T is raising prices by $5 per month for fiber and copper internet plans, starting August 16, targeting customers established between June 2024 and July 2025. Price increases alongside volume growth are the combination that drives the margin expansion embedded in the Q2 results.Also Read: AT&T Inc.Latest News and StoriesAT&T is up 0.64% year-to-date but has returned 89.07% over three years, according to Yahoo Finance. The stock had been range-bound for much of 2026 as investors debated the satellite threat. Stankey’s direct engagement with that question, paired with a buyback acceleration and strong free cash flow, is why the Friday, July 24 session looked the way it did.Related: AT&T quietly raises prices again as customer behavior shifts

International Business Machines (IBM) CEO Arvind Krishna has spent the past 10 days apologizing.On July 14, he told shareholders his company had faltered. Eight days later, he opened the second-quarter earnings call by admitting that IBM fell short on execution.Krishna then told CNBC on Thursday, July 23, that only 2% of IBM’s software could be replaced by applications built by artificial intelligence models. Everything else, he argued, helps clients get ready for AI rather than compete against it.That is a bold number from a chief executive whose stock recently recorded the worst single trading day in its 115-year history.IBM shares rose 4.01% to $214.94 by midday Friday, July 24, up $8.30 from Thursday’s close of $206.65, but the stock is still down more than 18% over the past month and sits near its 52-week low of $199.19.What Krishna’s 2% claim means for IBM software revenueSoftware is IBM’s profit engine. It brought in $7.76 billion in the second quarter, up 5%, and makes up roughly 45% of the company’s revenue, CNBC reported.Krishna’s argument rests on a distinction most investors skip. More AI Coverage:The AI honeymoon appears over amid stock sell-offGoldman Sachs backs surprising non-AI stocksMark Cuban sees a problem with the AI spending spreeMost of that revenue is infrastructure software, the layer that manages data, security, and hybrid cloud plumbing. AI tools need that layer to run.Applications are the exposed category. IBM’s Tririga lease management product, acquired in 2011, earns about $2 million a year from Starbucks (SBUX), which is phasing it out before a 2027 support cutoff.Put a dollar figure on it. Two percent of a $7.76 billion quarterly software business is about $155 million a quarter, or roughly$620 million a year.Krishna made the same case to analysts on the earnings call, telling them IBM’s AI strategy is the right one and the shortfall came down to execution, Yahoo Finance reported.

IBM chief executive Arvind Krishna told CNBC that only 2% of the company’s software faces real replacement risk from AI-built applications.Marvin Samuel Tolentino Pineda / Getty Images

The 42% mainframe collapse behind IBM’s AI defenseIBM’s Z mainframe revenue fell 42% in the second quarter, and the transaction processing software tied to those machines dropped 9%, according to CNBC.One quarter earlier, Z revenue had grown 48%.The reversal came from IBM’s own customers. Krishna wrote in IBM’s July 14 letter to investors that clients spent late June redirecting capital toward servers, storage, and memory to lock in supply ahead of price increases.Memory prices are climbing because AI chip production absorbed manufacturing capacity. So IBM’s clients bought hardware first and pushed software purchases into the next quarter.That sequence hurts twice, because IBM collects about $3 of software revenue for every dollar of mainframe hardware it sells.Inside IBM’s second-quarter 2026 resultsRevenue: $17.2 billion, up 1% year over year, according to IBM’s earnings releaseSoftware: $7.76 billion, up 5%, with Red Hat up 11% and data offerings up 19%Consulting: $5.33 billion, flatInfrastructure: $3.84 billion, down 7%, including the 42% drop in IBM ZOperating earnings: $2.93 a share, up 5%Free cash flow: $4.8 billion for the first six months, flat year over yearWhy Wall Street cut IBM stock price targets anywayKrishna handed analysts a second data point. About one-third of the deals that slipped out of the second quarter have already closed in July.  IBM normally recaptures two-thirds to three-quarters of slipped deals within six months, Benzinga noted. Krishna called that a deferral rather than destroyed demand.Related: Oppenheimer sends warning on IBM after shares crashWall Street trimmed targets regardless. Morgan Stanley analyst Erik Woodring cut his target to $190 from $293 on July 23 while holding an equal weight rating, according to GuruFocus. Stifel analyst David Grossman moved to $235 from $290 and kept a buy rating, Investing.com reported.Grossman told clients IBM’s stock will likely stay stuck in a narrow range, with more room to fall than to rise.Guidance came down with them. IBM now expects full-year constant currency revenue growth of 4% to 5%, trimmed from more than 5%. Oppenheimer had already downgraded the stock after the July 14 warning.How IBM stock compares with the S&P 500 and software peersThe scoreboard is unforgiving.IBM: Down about 30% in 2026 through Wednesday’s close, according to CNBCS&P 500: Up about 10% across the same stretchiShares Expanded Tech-Software Sector ETF: Down 17%IBM is trailing its own sector by 13 percentage points, which points to execution problems specific to Armonk on top of the industry-wide anxiety about AI.IBM now trades at 19.08 times earnings against a 52-week high of $332.46, and yields 3.15% on a quarterly payout of $1.69 a share. Its free cash flow held at $4.8 billion for the first half.IBM has paid consecutive quarterly dividends every year since 1916. That streak is the biggest reason income investors held on through a 25% one-day crash.What has to happen before IBM stock earns back its valuationKrishna and CFO Jim Kavanaugh are asking investors to accept a two-part promise: The delayed deals come back, and mainframe demand recovers.Kavanaugh told Yahoo Finance that IBM sees no evidence of clients walking away from the mainframe, and that installed capacity points to a record year against prior programs.Four things have to land for that promise to hold:The remaining two-thirds of slipped deals close by the fourth quarterSoftware growth reaches the 6% to 8% full-year range management now guides toFree cash flow rises by about $1 billion for the year, as reaffirmed on July 22Z mainframe revenue stops declining by early 2027IBM is also committing more than $10 billion to quantum computing over five years, which will not offset a 42% mainframe decline inside 2026.What IBM’s next two quarters mean for ordinary investorsKrishna’s 2% claim is testable, which is what makes it useful.Red Hat grew 11% in the quarter, while transaction processing software fell 9%. Transaction processing is tied directly to the mainframe cycle, so that split supports his timing argument.Here’s the practical read. At $214.94, IBM’s stock price already assumes 4% to 5% revenue growth and stable cash flow. It assumes nothing more. If software growth jumps back into double digits, that upside isn’t priced in yet.Buying now means trusting that the delayed deals land on schedule. Waiting costs you the first leg of any rebound and buys you one more quarter of evidence.The number to watch on the October earnings call is Z mainframe revenue. Krishna has said software should catch back up within a year, and the mainframe line is where that claim will show up first.Related: OpenAI just disclosed something genuinely alarming

For most of the past two years, investors bought nearly every company tied to semiconductors, from chip designers to equipment makers, and those stocks rose across the board.Enterprise software, on the other hand, got treated as collateral damage, priced as though large language models would eventually make the whole category redundant.That assumption is now getting tested, and not by the software companies themselves.Piper Sandler told clients on Wednesday that five infrastructure software names are positioned to solve the problem chief information officers complain about most: Running AI agents at scale costs far more than anyone budgeted.Piper Sandler’s argument is that the customer data these companies already store can cut the number of tokens an AI agent needs to process, which lowers the cost of running it.Why Piper Sandler says these 5 software stocks cut AI token costsThe note, led by analyst Rob Owens, named Elastic (ESTC), GitLab (GTLB), MongoDB (MDB), Snowflake (SNOW), and Atlassian (TEAM) as the primary beneficiaries, Investing.com reported.A token is a chunk of text that is often smaller than a word. AI models charge by the token, counting both what you send in and what you get back.Related: AI is quietly changing how portfolios are managedOwens wrote that the proprietary data already sitting inside these platforms can make models “significantly more accurate and efficient while dramatically reducing token usage costs.” That will let companies expand AI adoption without costs rising too much.Early deployments showed token usage falling by 50% to 75% when clean organizational context was fed directly to the agent.The mechanism is simple enough. AI uses fewer tokens and answers faster when given clean, organized data instead of messy data.The token math that changed enterprise AI budgets in 2026Here is the part that confused a lot of investors this year: Token prices fell, yet AI bills went up anyway.Owens noted that output tokens on newer frontier models run about 50% cheaper than the prior generation, yet improved reasoning capabilities caused consumption to increase.Reasoning models think in tokens, so a single query that once cost a few hundred tokens can now cost tens of thousands.More AI Stocks:AMD just landed its biggest AI deal yetThe AI honeymoon appears over amid stock sell-offMorgan Stanley sends strong verdict on memory stocksSnowflake’s pricing documentation shows how detailed this has become. The company splits AI usage onto a separate consumption meter so customers can track token spend against regular processing costs.That shift changed corporate behavior. Companies moved away from what Owens calls “Tokenmaxxing,” or throwing unlimited model capacity at every problem. Instead, the companies shifted toward model routing, which sends easy queries to cheap models and hard ones to expensive models.What the consumption pricing model means for revenueVendors price context layers on consumption rather than per seat. That matters because the per-seat model is exactly what the market fears AI will destroy as headcounts shrink.Piper Sandler called this an attractive incremental growth opportunity that also strengthens long-term competitive advantages.Put plainly, if a customer’s AI agents run more queries next quarter, the vendor gets paid more without signing a single new user.Three things have to hold for that thesis to work:Enterprises must keep expanding agent deployments rather than pausing them.Context layers must stay difficult enough to replicate that model vendors do not absorb the function.Consumption revenue must grow faster than any decline in traditional seat licenses.Owens said conversations with management teams and channel partners confirmed that organizations are turning to software to make AI more efficient.How these 5 software stocks have actually tradedThe stocks Owens named have not moved as a group.MongoDB has been the standout, with a market capitalization near $27.7 billion in mid-July, up more than 62% from last year, according to StockAnalysis data. The stock traded around $307 on July 21.Elastic went the other direction. Shares sat near $50 in recent trading, and Jefferies cut its target to $75 from $95 while keeping a Buy rating.GitLab has been the weakest of the five. Analysts carry an average Hold rating with a 12-month target of $34.50, roughly 4% above where shares trade.Snowflake sits in between, with 33 analysts rating it Strong Buy at an average target of $302.26.Atlassian rounds out the group with shares sitting near $86 as of the time of writing, well below the average analyst target of $139.70 reported on Yahoo Finance. KeyBanc set the most recent target at $115 on July 8 while keeping an Overweight rating, which points to about 33% above where the stock trades.

Piper Sandler says enterprise data platforms are becoming the cost control layer for corporate AI deployments.SOPA Images / Getty Images

Where this fits against the broader software selloffPiper Sandler is not alone in making this argument.Morgan Stanley told clients this week that sentiment on software has become too negative, naming eight Overweight companies positioned for the AI era, Yahoo Finance reported.The firm raised a similar question: What happens to software growth once AI companies stop selling tokens below cost?The backdrop explains why these calls keep coming. The S&P 500 software industry index has fallen more than 25% from its October highs. The iShares Expanded Tech-Software Sector ETF (IGV) tells a similar story. It’s down 13% this year. Meanwhile, the S&P 500 has gained close to 10% over the same stretch.Risks investors should weigh before buying the thesisThe counterargument to Owens’ call is that AI model providers build retrieval and memory features directly into their own platforms.Nothing stops a frontier lab from building its own retrieval and memory tools, which would make a third-party context layer less necessary. Several labs have already started doing this.There is also a timing problem. Piper Sandler describes a critical window opening, which is analyst language for a call that has not yet shown up in reported revenue.None of these five companies breaks out context-layer revenue as its own line item in filings. That means investors are betting on an analyst estimate, not a disclosed number.Two further limits that also matter:The 50% to 75% savings figure comes from early use cases, not audited results across a customer baseConsumption pricing cuts both ways, since AI budget cuts would hit revenue faster than annual seat contracts wouldWhat to watch next on these AI software stocksThe next earnings cycle should settle a lot of things.Snowflake, MongoDB, and Elastic all report consumption metrics that investors can check to verify Piper Sandler’s call.Each company’s management comments on AI-driven usage will tell you whether context layers are actually producing revenue.Watch net revenue retention specifically. If existing customers are spending more as agent deployments expand, that means the consumption approach is working.Also watch whether GitLab and Atlassian, the two most seat-dependent names on the list, can show credit or consumption revenue growing while seat counts stay flat.For readers deciding what to do with this, the practical read is that the five names carry very different risk profiles despite sharing a common call. MongoDB has already priced this call in. GitLab has not. Related: Cathie Wood buys $8.7 million of beaten-down AI stock

If your portfolio has been riding the same cluster of mega-cap technology stocks since 2023, Morgan Stanley has a timely warning. The S&P 500climbed about 20% from its April low to a record high near 7,620 on June 2, fueled by optimism over the U.S.–Iran ceasefire and persistent enthusiasm for artificial intelligence.Since then, the benchmark has stalled, closing near 7,457 on July 17 and struggling to gain traction despite strong corporate earnings. Giant stocks are pulling in opposite directions, and those offsetting moves are keeping the broader market locked in place.Morgan Stanley Wealth Management chief investment officer Lisa Shalett laid out those dynamics in her July Global Investment Committee presentation. Her conclusion carries a pointed implication for anyone who remains heavily concentrated in the same handful of Big Tech names.Semiconductor stocks have ballooned to a historic share of the S&P 500The core of Shalett’s case is about how lopsided the S&P 500 index has become, and exactly where that imbalance is concentrated.The 10 largest stocks now represent about 40% of the S&P 500’s total market value, according to Morgan Stanley Wealth Management’s July 2026 Global Investment Committee note. A modest pullback in just a few of those names can erase gains from hundreds of other companies in the AI trade.JPMorgan strategist Mislav Matejka advises buying dips triggered by geopolitical tensions despite lingering risks.The risks of renewed flareups remain, but we believe one should keep using any dips on the back of adverse geopolitical headlines in order to add.Chipmakers have surged to extreme valuations, with semiconductor market capitalization growing to about 18% of the S&P 500, Shalett noted. That figure stood at roughly 3% for most of the index’s modern history, illustrating how concentrated the rally has become.Investors have simultaneously punished the “Magnificent Seven” hyperscaler stocks over concerns about the cost of their massive AI infrastructure buildouts.That creates an index-level stalemate where one group of trillion-dollar names gets bid higher while another gets sold, and a concentrated portfolio goes nowhere.Morgan Stanley says the AI trade is entering a cost-conscious phaseThe real substance of Shalett’s note goes beyond diagnosing the stall and into where the AI investment cycle is heading next.Enterprises are moving from an early adoption phase focused on maximizing AI usage to a disciplined approach that prioritizes cost control, the committee observed. More Tech:Microsoft cuts thousands as Xbox faces rude awakeningSpectrum makes significant decision as customer losses mountGiant troubled satellite TV company files Chapter 11 bankruptcyThat transition is pushing what the firm calls “hybrid engineering” across the AI technology stack, Shalett explained in the presentation. In practical terms, companies are becoming more willing to blend expensive frontier AI models with lower-cost open-source alternatives and to diversify their hardware choices. That trend could pressure chipmakers whose valuations assume limitless demand while rewarding cloud providers that adapt to leaner enterprise budgets, the firm noted.

Morgan Stanley says companies are shifting toward cost-efficient AI, favoring hybrid models and disciplined spending over unlimited infrastructure investment.Kasipat Phonlamai/Getty Images

How semiconductor concentration risk affects index fund investorsInvestors holding S&P 500 exchange-traded funds carry sector concentration that is not always visible.Cameron Dawson, chief investment officer at NewEdge Wealth, quantified the scale during a recent interview on the Thoughtful Money program. A decade ago, semiconductors accounted for about 2% of the S&P 500, and today that figure sits near 18%, she noted.That means a passive investor who believes they hold a diversified portfolio has nearly one in five dollars exposed to the chip trade. Chip stocks are projected to deliver about 133% year-over-year earnings growth in the second quarter of 2026, according to data compiled by the London Stock Exchange Group (LSEG) and cited by earnings research head Tajinder Dhillon.That single sector accounts for roughly 44% of the entire S&P index’s profit expansion, and the Philadelphia Semiconductor Index has fallen about 20% from its late-June record high, entering bear-market territory, according to Bloomberg.Morgan Stanley’s positioning adjustments for the second half of 2026Shalett’s team outlined several moves in the July note that reflect the firm’s view on where market leadership is migrating. Investors with large gains in semiconductor holdings may want to capture profits, especially where earnings expectations appear stretched, the committee recommended. The team also suggested selectively revisiting hyperscaler stocks that are retooling their businesses to serve cost-conscious AI demand across enterprise customers.On the fixed-income side, the firm pointed to intermediate-term bonds as a priority allocation, with longer-duration bonds worth adding if yields climb further, Shalett noted. The committee also emphasized global diversification, noting that non-U.S. equity markets have continued to outperform and offer a reason to broaden exposure.Gargi Pal Chaudhuri, BlackRock’s chief investment and portfolio strategist for the Americas, has argued that “continued dispersion” will define 2026, urging investors to broaden holdings beyond the AI trade, according to her February appearance on Yahoo Finance’s Market Domination Overtime.Chaudhuri told Yahoo Finance that ‘continued dispersion’ will be a defining 2026 theme as leadership broadens away from the AI infrastructure trade, with markets rewarding companies that convert AI spend into profitability rather than those simply announcing more capex. The shift from AI builders to AI adopters is gaining momentumShalett’s thesis builds on a pattern that Morgan Stanley has been developing since February, when the firm first flagged a rotation from mega-cap tech.In an earlier note, the committee argued that capital would eventually flow from AI “builders” selling infrastructure to AI “adopters” boosting margins with the technology. Health care, energy, software, and financial services were all identified as sectors with meaningful tailwinds from AI adoption in the report.That rotation appeared to stall during the spring rally as chipmakers recaptured attention, but the July semiconductor selloff suggests it is now resuming. Morgan Stanley’s Global Investment Committee framed the takeaway as a call to broaden, not exit, the AI theme, arguing that the next phase of returns is more likely to reward selective exposure across sectors than continued concentration in the same handful of Big Tech names.Related: Morgan Stanley sends strong verdict on memory stocks