Micron earnings, EUV bottleneck, and NASDAQ valuation is fair
Micron earnings, EUV bottleneck, and NASDAQ valuation is fair
7 hours ago•Mark Meldrum
Podcast1 hr 29 min
Listen to Episode
Note: AI-generated summary based on third-party content. Not financial advice. Read more.
Quick Insights
  • Micron (MU) is the clearest direct play on tight memory supply and AI demand; consider it only if you can tolerate memory-cycle volatility, since supply may remain constrained beyond 2027 but prices and margins could reverse.
  • Treat the cited $1,000–$1,240 MU range as an illustrative valuation—not a price target—and monitor capacity growth, customer agreements, and AI spending for signs the shortage is easing.
  • Nasdaq 100 (NDX)/QQQ has a cautiously bullish outlook, with possible new highs before year-end; use July technology earnings and guidance as a key test rather than assuming the AI investment boom will continue unchecked.
  • ASML (ASML) and Amkor (AMKR) may benefit from equipment and advanced-packaging bottlenecks, but the discussion provides no valuations or explicit buy signals, so treat them as watchlist names.
Detailed Analysis

Micron Technology (MU)

  • Micron’s latest quarter showed exceptionally strong results: revenue was $41.46 billion, up 74% quarter over quarter and 346% year over year. The host said most of the increase came from higher memory prices, while volume growth was only in the low single digits.
  • The host emphasized that Micron appears to be operating near capacity. Inventory was mostly work in progress, which he interpreted as evidence of a full production pipeline and limited ability to increase output quickly.
  • Micron expects supply tightness to continue beyond 2027, with supply and demand potentially moving toward balance in 2028. New facilities and additional capacity take years to come online, and equipment availability is a constraint.
  • Micron’s customer agreements could make some revenue more predictable: 16 strategic agreements cover portions of DRAM and NAND volume, with pricing floors and, for the largest agreements, ceilings. The host said the agreements could eventually cover about 40% of revenue, leaving the rest exposed to spot pricing.
  • The host sees potential structural support for memory demand from AI infrastructure, greater memory use per GPU, and expanding use in inference servers, on-premises systems, edge devices, and vehicles.
  • He also stressed that memory remains cyclical. High prices could reduce demand, encourage customers to bring memory production in-house, or attract new capacity and competition. Slower AI capital spending or more efficient use of memory could also ease the shortage.
  • As an illustrative valuation exercise—not a price target—the host applied an 8–10× multiple to annualized guided EPS of about $31, assuming no growth. That produced a rough range of $1,000–$1,240 per share, compared with the $1,135 price he cited.

Takeaways

  • The discussion presents Micron as a potential beneficiary of tight memory supply and structurally rising demand, but with significant exposure to the memory cycle.
  • Watch for evidence that capacity is increasing, customer agreements are limiting pricing upside, or AI spending and memory demand are moderating.
  • The host’s valuation example depends heavily on assuming zero growth and should not be treated as a forecast or recommendation.

ASML (ASML)

  • The host described extreme ultraviolet (EUV) lithography equipment as a key constraint on memory production. He said ASML planned to ship 60 EUV units in 2026 and 80 in 2027, with those units shared between logic and memory manufacturers.
  • According to the discussion, ASML is fully booked, so new semiconductor capacity depends not only on building facilities but also on obtaining the necessary equipment. More advanced memory nodes may require more EUV layers, increasing the number of machines needed to achieve a given production level.

Takeaways

  • ASML was discussed as a critical supplier in the semiconductor capacity chain, with equipment availability potentially limiting how quickly memory supply can expand.
  • The transcript does not provide an ASML valuation, price target, or explicit stock recommendation.

Amkor Technology (AMKR)

  • The host identified advanced packaging as another bottleneck, alongside front-end lithography. He mentioned Amkor in connection with outsourced 3D-packaging work and described advanced packaging capacity as a constraint on semiconductor production.

Takeaways

  • Amkor may be relevant to investors following the advanced-packaging supply chain, but the transcript provides no company-specific financial analysis or stock recommendation.
  • The broader point is that packaging capacity, not just chip fabrication, may limit how quickly AI hardware supply can grow.

SK Hynix (000660.KS)

  • The host cited SK Hynix’s reported signal that it would be careful about expanding high-bandwidth memory production, including a shift toward serving the consumer market.
  • He interpreted such caution as a possible sign of capital discipline among memory suppliers. In a market with only a few major producers, companies may be wary of repeating past cycles of overbuilding capacity.

Takeaways

  • Capital discipline among major memory producers could help sustain margins, but it is not assured; the industry remains cyclical and capacity decisions can change.
  • The host offered no specific valuation or recommendation for SK Hynix.

Samsung Electronics (005930.KS)

  • Samsung was discussed as one of the major memory suppliers and as a company using a different naming convention for memory process layers than Micron.
  • The host’s broader argument was that the small number of major memory producers may create incentives to avoid aggressive capacity expansion, though he did not say that such discipline was guaranteed.

Takeaways

  • Samsung’s capacity decisions are relevant to the memory supply outlook, but the transcript does not provide a company-specific investment view or valuation.

Taiwan Semiconductor Manufacturing Company (TSM)

  • The host raised TSMC as a potential competitive threat to memory makers. If unusually high memory margins persist, he said TSMC could consider entering or expanding in memory, despite having stayed away from the business historically.

Takeaways

  • Potential entry by a large semiconductor manufacturer is a risk to the long-term pricing and margins of existing memory producers.
  • This was presented as a hypothetical threat, not as a stated TSMC plan.

Cerebras (Private)

  • The host mentioned Cerebras’ claim that its high-bandwidth-compute accelerator could reduce the amount of high-bandwidth memory needed. He said he had not investigated the claim.

Takeaways

  • If accelerator designs materially reduce memory requirements, that could weaken demand for high-bandwidth memory.
  • The transcript does not establish whether Cerebras’ claim is technically or commercially proven.

Alphabet (GOOGL, GOOG)

  • The host said Google had discussed using better algorithms to reduce memory intensity. He presented algorithmic efficiency as one possible way the memory shortage could eventually ease.

Takeaways

  • Improvements in software efficiency could reduce memory demand per AI workload, creating a risk for memory suppliers if those gains are substantial.
  • The transcript does not provide a separate investment view or valuation for Alphabet.

Apple (AAPL)

  • The host said that high memory prices could motivate consumer-facing companies such as Apple to secure their own supply or bring more of the process in-house.
  • He described this broader customer response as a risk to memory suppliers, alongside potential demand destruction in consumer markets.

Takeaways

  • Apple’s potential response illustrates how sustained high input prices can encourage customers to seek supply security or alternative arrangements.
  • No specific Apple investment recommendation was made.

Nasdaq 100 (NDX) and Invesco QQQ Trust (QQQ)

  • The host described the Nasdaq’s valuation as broadly fair relative to recent history. He cited a forward P/E of 23.14 and a trailing P/E of 32.46; the trailing figure was near the one-year average he cited and above the five-year average.
  • The Nasdaq was about 5.3% below its cited June 2 all-time high. The host said the market’s pullbacks felt more like jitters than the start of a bubble bursting.
  • He expected the market to reach new highs at some point before year-end, with Q2 earnings and technology-company commentary in July as possible catalysts. He also warned that the infrastructure build-out will eventually end and that earnings could challenge the optimistic view.
  • The host disclosed significant exposure to the Nasdaq 100, describing 18 puts and four long contracts. He said he did not expect to increase the position.
  • He also discussed illustrative options pricing: a 20-day, one-delta QQQ put near $655 with a quoted premium of $5.25, and a one-delta call near $749 with a quoted premium of $3.76. These were market observations, not explicit trade recommendations.

Takeaways

  • The host’s stance was cautiously bullish: he expected possible new highs by year-end but characterized Nasdaq valuation as fair rather than clearly cheap.
  • July technology earnings and guidance were identified as important tests of the AI infrastructure thesis.
  • The options figures were examples from the discussion, not recommendations; options can involve substantial risk.

S&P 500 (SPX)

  • The host said the S&P 500 was about 4% below its cited June 2 all-time high.
  • He calculated a forward multiple of 20.82 using operating earnings of $353.22 and an index level of 7,354. He noted that this differed from another source’s cited forward multiple of 22.06, and that the methodologies were unclear.
  • In his historical comparisons, the S&P 500 appeared broadly fairly valued.

Takeaways

  • The host viewed the S&P 500 as roughly fairly valued, with earnings and guidance likely to determine whether the market can advance.
  • The transcript provides no specific S&P price target.

Russell 2000 (RUT) and iShares Russell 2000 ETF (IWM)

  • The host noted that IWM reached an all-time high even as the Nasdaq and S&P 500 pulled back.
  • Despite the rally, he described the Russell 2000 as expensive relative to its five-, ten-, and twenty-year valuation comparisons. He also cited the index’s P/E as 15.5% above its 200-day average.

Takeaways

  • The discussion highlights a contrast between strong small-cap price momentum and relatively expensive valuations.
  • Investors may want to distinguish recent market strength from the valuation evidence the host cited.

Nike (NKE)

  • Ahead of Nike’s earnings, the host said he did not expect a quick turnaround. He raised the question of whether the company’s brand and share price made it a value opportunity or a “value trap.”
  • He said he did not think Nike was going away, but suggested that a recovery could require patient capital rather than expecting improvement within a quarter or two.

Takeaways

  • The host’s view was cautious: Nike may have long-term value potential, but he did not expect a near-term turnaround.
  • The transcript gives no price target or explicit buy recommendation.

General Mills (GIS)

  • The host said General Mills had been under pressure for a long time and that the market expected it to miss earnings.
  • He described having owned 1,000 shares and selling calls for several months before exiting, saying the premiums were not attractive and he did not see a convincing turnaround.

Takeaways

  • The host expressed a bearish-to-cautious view, citing ongoing pressure and limited near-term recovery prospects.
  • No price target or specific recommendation was provided.

AI Infrastructure and Semiconductor Supply Chain

  • The host’s broader thesis was that AI infrastructure investment is not necessarily finished. He argued that more models and broader adoption could increase demand for computing infrastructure, even if open-source models are cheaper.
  • He identified bottlenecks in semiconductor equipment, memory, advanced packaging, and energy availability. Announced capital spending may not translate immediately into completed projects if companies cannot secure energy or equipment.
  • He also said data-sovereignty requirements could lead countries to duplicate infrastructure rather than rely on a single shared system.
  • Risks include slower AI capital-spending growth, more efficient algorithms or hardware, excess capacity, and the eventual end of the infrastructure build-out.

Takeaways

  • The discussion favors looking across the AI supply chain—not only at model developers—for potential investment opportunities, while recognizing that bottlenecks can delay revenue and capacity growth.
  • The host’s central uncertainty is how much current demand and high margins are sustainable versus cyclical.
Ask about this postAnswers are grounded in this post's content.
Episode Description
0:00 - 7:40 July Live sessions7:41 - 18:59 Economic data 19:00 - 43:51 Micron Earnings43:52 - 51:11 WIP Inventory and EUV bottleneck51:12 - 1:07:46 Memory market sustainability 1:07:47 - 1:11:49 NASDAQ 100 valuation1:11:50 - 1:15:09 World market valuations1:15:10 - 1:22:59 SPY and earningsPDF repository/July Live Sessions:https://7584408165302.gumroad.com/Link to CFA subscription: https://www.markmeldrum.comLink to Applied Level Subscription: https://www.markmeldrum.com/subscriptions/Every 2 weeks, I highlight a company in a new or emerging market segment, or a stock that is newsworthy. I develop an AI 40-prompt infographic for each highlighted company. If you want this e-mailed to you the day before the Market Outlook is released, click here: https://drmarkmeldrum.substack.com/ Below are links to data sources I use or will use in the Market OutlookMajor Foreign holders of UST:https://ticdata.treasury.gov/resource-center/data-chart-center/tic/Documents/slt_table5.htmlRealtor.com:https://www.realtor.com/research/november-2025-data/Economic calendar:https://tradingeconomics.com/calendarThis Week in Earnings:https://lipperalpha.refinitiv.com/wp-content/uploads/2026/06/TRPR_82221_857.pdfMoney Market flows:https://www.ici.org/research/stats/mmfFed Balance Sheet:https://www.federalreserve.gov/monetarypolicy/bst_recenttrends.htmSIFMA:https://www.sifma.org/resources/research/us-corporate-bonds-statistics/Freddie Mortgage Rates:https://www.freddiemac.com/pmmsTreasury:https://home.treasury.gov/resource-center/data-chart-center/interest-rates/TextView?type=daily_treasury_yield_curve&field_tdr_date_value=2023SOMAhttps://www.newyorkfed.org/markets/soma-holdingsEFFRhttps://www.newyorkfed.org/markets/reference-rates/effrReverse Repohttps://www.newyorkfed.org/markets/desk-operations/reverse-repoFedWatch Tool:https://www.cmegroup.com/markets/interest-rates/cme-fedwatch-tool.html?redirect=/trading/interest-rates/countdown-to-fomc.htmlSTIR analyticshttps://www.cmegroup.com/tools-information/quikstrike/short-term-interest-rate-analytics.htmlForward Curves:https://www.chathamfinancial.com/technology/us-forward-curves Learn more about your ad choices. Visit megaphone.fm/adchoices
About Mark Meldrum
Mark Meldrum

Mark Meldrum

Mark Meldrum is a podcast of long-form talks and lectures from Mark Meldrum, covering ideas that reward a full listen. New episodes continue the same deep-dive format listeners come for.Links:Substack: https://drmarkmeldrum.substack.com/aboutGumroad: https://7584408165302.gumroad.com/Instagram: https://www.instagram.com/applied_level/Twitter: https://twitter.com/AppliedLevel/ Made by a listener who resonates deeply with Mark Meldrum's work, so it can travel with you on walks, drives and quiet nights. An unofficial tribute, not affiliated with Mark Meldrum — all rights remain theirs. If you're the creator, we'd love to hear from you.