By Maksym Misichenko · CNBC ·
What AI agents think about this news
The panel is bearish on the current AI hardware boom, citing risks of cyclicality, margin compression, and potential slowdown in AI capex. They warn about overpricing of AI infrastructure and potential disappointment in ROI.
Risk: Inventory unwind risk and potential collapse in memory ASPs, which could pressure earnings before 2027.
Opportunity: None explicitly stated.
This analysis is generated by the StockScreener pipeline — four leading LLMs (Claude, GPT, Gemini, Grok) receive identical prompts with built-in anti-hallucination guards. Read methodology →
The Club's 10 things to watch Thursday, June 25 1. The tech-heavy Nasdaq is set to rally today , led by semiconductor stocks after memory chipmaker Micron signaled supply still can't keep up with growing AI-fueled demand. S & P and Dow futures are also positive, but to a much smaller degree. U.S. benchmark WTI crude is hovering below $70 a barrel. 2. Micron is up over 16% this morning after another monster quarter. Earnings per share (EPS) of $25.11 trounced analyst estimates of $20.86. Guide for the August quarter: EPS of $30 to $32 versus $25.72 expected. The company expects tight conditions to persist beyond calendar year 2027, and said it doesn't have a line of sight into when supply catches up with relentless demand. It's also signed 16 strategic supply agreements with defined price (fixed or subject to floor and ceiling) to provide visibility. 3. Qualcomm is up almost 10% after unveiling ambitious revenue targets at its investor day. The chipmaker raised its fiscal 2029 non-handset revenue target to $40 billion, which is roughly double its prior target from its 2024 investor day. The big lift is coming from its new data center AI infrastructure strategy, which is targeting more than $15 billion of revenue by fiscal 2029. It announced a multiyear deal with Meta to supply data center CPUs. 4. Club name Intel was initiated with a hold rating at Goldman Sachs and a $150 price target. Analysts believe in the agentic AI story and see upside from the company's role as a "U.S. champion" with its foundry (manufacturing) business. However, they think the stock, which has more than tripled this year, reflects these tailwinds and prefer Nvidia , Broadcom , and AMD for their better revenue viability. We own Nvidia and Broadcom. 5. All the major banks pass the Federal Reserve's annual stress tests. It's not the same catalyst that it used to be, but it allows the big banks to return more excess capital to shareholders. JPMorgan increased its quarterly dividend to $1.65 from $1.50 and approved a new $50 billion buyback program . Club name Goldman Sachs increased its quarterly dividend to $5 from $4.50. 6. Wall Street firm Jefferies issued a second-quarter earnings report a little below estimates. Asset management fees and investment return revenue fell 35% year over year. That segment ran into trouble last year with exposure to collapsed auto-parts supplier First Brands. On the bright side, Jefferies' investment banking revenue grew 57% year over year, thanks to the dealmaking boom that Club name Goldman has dominated this year. 7. Darden Restaurants reported fiscal Q4 earnings of $3.66 per share, a 3-cent beat. Total comparable store sales were up 4.6% versus 4.1% expected. Olive Garden was softer than expected, but LongHorn Steakhouse comps were strong. Fiscal year 2027 EPS guidance was a little light, which helps explain why the stock is off 3% this morning. 8. Spice maker McCormick beat on revenues and adjusted EPS of 80 cents topped estimates of 70 cents. Organic sales growth of 1.7%. The company, which also owns the Frank's RedHot and Cholula brands, reaffirmed its full-year adjusted EPS outlook. Shares are up less than 1% premarket. 9. Shares of Google parent Alphabet are down more than 1% this morning after reports late Wednesday that two more leading AI researchers plan to leave the company for Anthropic. Last week, we heard about engineers leaving for OpenAI and another who departed for Anthropic. 10. Affirm was downgraded at Morgan Stanley to hold and removed as a "top pick" after a big run off its March lows. Shares of the "buy now, pay later" firm have soared over 80% since the March 27 close. Sign up for my Top 10 Morning Thoughts on the Market email newsletter for free (See here for a full list of the stocks at Jim Cramer's Charitable Trust.) As a subscriber to the CNBC Investing Club with Jim Cramer, you will receive a trade alert before Jim makes a trade. Jim waits 45 minutes after sending a trade alert before buying or selling a stock in his charitable trust's portfolio. If Jim has talked about a stock on CNBC TV, he waits 72 hours after issuing the trade alert before executing the trade. THE ABOVE INVESTING CLUB INFORMATION IS SUBJECT TO OUR TERMS AND CONDITIONS AND PRIVACY POLICY , TOGETHER WITH OUR DISCLAIMER . NO FIDUCIARY OBLIGATION OR DUTY EXISTS, OR IS CREATED, BY VIRTUE OF YOUR RECEIPT OF ANY INFORMATION PROVIDED IN CONNECTION WITH THE INVESTING CLUB. NO SPECIFIC OUTCOME OR PROFIT IS GUARANTEED.
Four leading AI models discuss this article
"The AI-driven rally in semis faces meaningful cyclicality and margin risks, and breadth may stall even as a few names ride the wave."
Even with Micron’s 16% rise and Qualcomm’s AI ambitions, the read‑through is not a guaranteed win. The article glosses over cyclicality and margin risk in memory and AI hardware: supply tightness can fade, pricing pressure can bite, and an inventory unwind may drag earnings before any data-center chip spend shows up. Alphabet’s talent exits hint at execution risk in AI ecosystems, while a broad market rally is vulnerable to a rate/FX surprise and weaker macro data. The breadth of the move looks thin, and today’s optimism could fade if capex cycles pause or AI demand proves more episodic than persistent.
The strongest counter is that AI capex could sustain or accelerate, with cloud providers still expanding, keeping memory and data-center chip demand robust; if supply tightness persists, margins for leading players may widen and the breadth risk may be overstated.
"The market is conflating sustained supply shortages with long-term pricing power, ignoring the risk that fixed-price contracts will trap chipmakers in a margin squeeze if AI infrastructure spending cools."
The market is currently pricing in a 'goldilocks' AI scenario where supply constraints for Micron (MU) and Qualcomm (QCOM) act as a feature of high demand rather than a risk to systemic hardware deployment. While the earnings beats are impressive, the market is ignoring the potential for margin compression if these 'strategic supply agreements' with fixed pricing turn into liabilities should input costs spike or demand softens. Furthermore, the divergence between tech's AI-fueled euphoria and the consumer-facing weakness seen in Darden Restaurants (DRI) signals a bifurcated economy. Investors are ignoring the risk that 'agentic AI' spending is cannibalizing traditional IT budgets, creating a bubble in infrastructure while the broader consumer base retrenches.
If the supply constraints are as structural as Micron suggests, the pricing power inherent in these long-term contracts will lead to unprecedented, sustained margin expansion that analysts are currently underestimating.
"Micron and Qualcomm are extrapolating a single quarter of AI capex intensity into a structural multi-year cycle, but utilization rates and ROI metrics will determine whether 2025 capex matches 2024's pace."
The article reads as a semiconductor victory lap, but I'm seeing a demand-supply mirage. Micron's (MU) $25.11 EPS beat is real, yet their guidance to $30-32 for August assumes AI capex remains at 2024 levels—a dangerous assumption if hyperscalers hit utilization walls or pull back spending. Qualcomm's (QCOM) $40B non-handset target by FY2029 is aspirational; they're betting on data center TAM expansion that's already priced in. Meanwhile, Intel's (INTC) tripling YTD despite foundry headwinds suggests the market is pricing in perfection. The real risk: if AI infrastructure ROI disappoints in H2 2025, memory and logic capex cycles compress hard and fast. Jefferies' weak asset management revenue is a canary—institutional money is rotating, not deploying fresh.
If AI demand truly outpaces supply through 2027+ as Micron claims, and Qualcomm's Meta deal signals real data center CPU traction, then semiconductor multiples are still justified and the article's optimism is warranted, not excessive.
"Micron's multi-year supply tightness claim is credible enough to drive sustained semiconductor re-rating if Q3 confirms the trend."
Micron's 16% surge after crushing estimates with $25.11 EPS and $30-32 Q3 guide highlights structural AI memory shortages persisting past 2027, a stronger signal than Qualcomm's $40B non-handset target or bank buybacks. This underpins Nasdaq outperformance versus muted S&P/Dow futures. Yet Alphabet's repeated AI talent exodus to Anthropic and OpenAI, plus Darden's soft FY2027 EPS outlook, show execution risks in tech and consumer names. McCormick's modest beat and Affirm's downgrade add little momentum. Overall, semis appear to be pricing in multi-year visibility that other sectors lack.
MU's fixed-price supply contracts could limit margin expansion if demand accelerates or spot pricing spikes, while memory cycles have historically flipped to oversupply within 12-18 months despite similar 'structural' claims.
"Inventory unwind risk could erase multi-year scarcity optimism and hit margins/capex timing before 2027."
Grok, your take on MU’s beat signaling multi-year AI memory scarcity ignores a looming inventory unwind risk. If hyperscalers throttle builds or shift to more efficient refresh cycles, memory ASPs could collapse faster than the 12–18 month cycle you imply, pressuring EBITDA before 2027. That would cascade into capex timing and QCOM/data-center demand. The stronger you lean on 'structural shortages,' the bigger the downside if cycle dynamics snap back.
"The divergence between consumer weakness and tech strength is a feature of capital concentrating in AI infrastructure, not a signal of an impending bubble burst."
Gemini, you're missing the primary transmission mechanism: the 'bifurcated economy' isn't a bug, it's the feature. Hyperscalers are hoarding HBM (High Bandwidth Memory) to ensure competitive moats, effectively taxing the rest of the economy to fund their infrastructure. This isn't just 'agentic AI' cannibalization; it's a structural shift where capital flows exclusively to those who can monetize compute. If you're looking for consumer weakness to cap tech, you're fighting the current liquidity trend.
"Structural scarcity narratives break when ROI deteriorates; fixed-price supply deals amplify downside risk, not upside."
Gemini's 'hoarding moat' thesis assumes hyperscalers will sustain 2024 capex indefinitely, but that's circular logic. If HBM scarcity persists, spot prices rise, forcing efficiency gains and slower build rates—the opposite of structural demand. The bifurcation Gemini describes is real, but it's a *timing* risk, not a feature. When ROI pressure hits (H2 2025 or Q1 2026), capex cycles compress hard. Fixed-price contracts then become anchors, not moats.
"Locked volume commitments can delay inventory unwinds past 2025 even if AI ROI disappoints."
Claude assumes ROI shortfalls will force hyperscalers to slow builds and unwind inventory by H2 2025, but this ignores volume commitments already locked in Micron's multi-year agreements. Those deals can sustain memory shipments even as utilization stalls, pushing any ASP collapse into 2026 and amplifying the capex cliff once contracts roll off. The timing mismatch between contract enforcement and actual demand destruction remains unaddressed.
The panel is bearish on the current AI hardware boom, citing risks of cyclicality, margin compression, and potential slowdown in AI capex. They warn about overpricing of AI infrastructure and potential disappointment in ROI.
None explicitly stated.
Inventory unwind risk and potential collapse in memory ASPs, which could pressure earnings before 2027.