Micron's FQ4-26 guide of $50.0B and 86% gross margin: a single quarter larger than every full year through FY2024, with SK hynix's P&T7 build as the contestable mirror

Micron Technology Inc (MU $959.48 -1.17%) is in the middle of an inventory cycle inversion so extreme that its single-quarter FQ3-26 results now exceed every full year of revenue the company has reported through fiscal 2024, and management's FQ4-26 guide of $50.0B ± $1.0B would do it again. The setup cuts against the cheap cyclical reading the memory name carried for most of the last decade: DRAM and NAND pricing moved up mid-sixties percent and high-seventies percent respectively, on CFO Mark Murphy's framing on the 2026 Q3 earnings call for the prior quarter (FQ2), and the company's stock trades at a P/E of 21.7× against Nvidia Corp (NVDA $212.06 +2.30%)'s 32.5× and Broadcom Inc. (AVGO $396.81 +2.67%)'s 66.0×. The market-cap line tells the same story. Micron at $1.1T now sits between SK hynix and Taiwan Semiconductor Manufacturing Co Ltd (TSM $415.58 -1.34%) and above Applied Materials Inc (AMAT $562.80 +1.60%), while a YTD return of +236.3% on a stock that closed at $959.48, off its $1213.56 high by 20.9%, has left both bulls and bears with something to defend.

The thesis is that Micron is no longer a price-taking commodity memory name but an AI-cycle beneficiary that has converted a one-year DRAM/HBM supply squeeze into a multi-quarter earnings reset, and the contestable claim is whether FQ4-26's guided $50.0B revenue band and ~86% gross margin can hold once HBM4E volumes ramp and SK hynix's incremental packaging capacity comes on line.

Key takeaways

• Micron Technology Inc trades at a P/E of 21.7× against Nvidia Corp's 32.5× and Broadcom Inc.'s 66.0× on TTM revenue of $90.3B and TTM EBIT of $59.2B, a multiple gap that frames the read.
• The FQ4-26 guidance is revenue of $50.0B ± $1.0B at the midpoint, gross margin of 86%, and GAAP diluted EPS of $30.73 ± $1.00, a single-quarter print that exceeds every full fiscal year Micron reported through FY2024.
• Adjusted free cash flow reached $18.3B in Q3 2026 against $0.803B in Q4 2025, with operating cash flow at $25.39B against $0.024B in Q3 2023, the operating reset that funds the Tongluo, Gujarat, Hiroshima, Singapore, and New York build cycle guided to a fiscal 2026 capex above $25B.
• The balance sheet ended FQ3-26 with $25.0B in cash and equivalents against $582M in short-term debt and $100.7B in stockholders' equity; CFO Mark Murphy told investors on the 2026 Q3 earnings call that total debt had been reduced by over $5B in three quarters and that the company had earned two credit upgrades to a solid triple-B, the strongest net cash position ever.
• The bear mirror is SK hynix Inc. (SKHY $169.50 +2.56%)'s Won 7,093,100,000,000 (5.88% of equity as of 2025-12-31) Cheongju P&T7 advanced-packaging investment and Won 39,890,534,790,000 (US$26,507,100,000) ADR allotment completed July 14, 2026, the direct competitor build arriving on the same timeline as MU's HBM4E CY2027 ramp.

The setup

Price performance vs peers (Semiconductors and cross-industry comps)

Valuation anchors

• Indicated dividend yield: 0.1% (latest declared $0.15/share × 4 payments a year = $0.60 annualized, on the $959.48 last close).

Most recent payout: $0.15 (ex 2026-07-06) vs $0.115 a year earlier (ex 2025-07-07): +30.4% year over year.

The market's read on Micron Technology Inc is hard to call cheap or one-directional. The stock closed at $959.48, +236.3% YTD and +778.5% over 52 weeks, but it is already off its $1,213.56 high by 20.9%, a drawdown that sits between Nvidia Corp's −10.0% and Netlist Inc (NLST $2.48 -1.98%)'s −26.7% from peak. Implied dividend yield on the latest declared $0.15 quarterly payout lands at 0.1%: small in absolute terms, but the payout itself stepped up by 30.4% versus the year-earlier $0.115 (ex 2025-07-07) FQ3-26 release: the first real sign that capital return is becoming a thing the board will defend, not just a token. Short interest of 2.8% of shares is materially below Netlist Inc's 11.0% and roughly in line with Formfactor Inc (FORM $112.75 -0.56%)'s 2.6% and Applied Materials Inc /De's 2.5%; the bearish side of the tape is not heavily armed. Revenue growth of +49.5% YoY sits well below Nvidia Corp's +125.9% but above Broadcom Inc.'s +7.9%, and Micron Technology Inc's TTM revenue of $90.3B is now between Broadcom Inc.'s $75.5B and Nvidia Corp's $253.5B. What the tape is not pricing is the FQ4-26 guide of $50.0B ± $1.0B in revenue, ~86% gross margin, and $30.73 ± $1.00 GAAP diluted EPS, which would be a single quarter larger than every full fiscal year Micron has reported through FY2024. Sanjay Mehrotra framed the disconnect on the FQ3-26 release: "Micron's record fiscal Q3 financial results and even stronger outlook for Q4 reflect the strategic value of memory in the AI era."

The contrarian read against the consensus memory-cyclical frame is in the dollar trajectories of FY2025 segment revenue. Cloud Memory $13.5B (36% of total) and Mobile and Client $11.9B (32%) versus a Total of $37.4B, with Core Data Center revenue at $7.2B (19%), a mix that is structurally different from the NAND-tilted 2022–2024 cycle. DRAM itself made up $28.6B (76% of FY2025 product revenue) with NAND at $8.5B (23%). The setup the bull case rests on is that Micron Technology Inc's HBM4 generation is the gating supply for an AI capex cycle that posted visible post-print confirmation: Alphabet (GOOGL $317.69 -7.13%)'s Q2 FY26 capex of $44.92B came on top of $35.67B in Q1 and $27.85B in Q4'25 8-K filed 2026-07-22, Microsoft (MSFT $381.58 -2.24%)'s 9-month FY26 capex of $80.15B was +69% Y/Y 8-K filed 2026-04-29, and Meta (META $627.17 -2.58%) raised 2026 capex to $125–145B from $115–135B citing "higher component pricing this year" 8-K filed 2026-04-29. Taiwan Semiconductor Manufacturing Co Ltd's Q2 CY26 revenue of US$40.20B (+33.7% Y/Y, advanced ≤7nm at 77% of wafer mix) and Q3 guide of US$44.6–45.8B 6-K filed 2026-07-16 is the foundry-side confirmation that the HBM base-die substrate is also running tight. The bear's read is the supply-side mirror: SK hynix Inc.'s board approved a Won 7,093,100,000,000 (~5.88% of equity as of 2025-12-31) increase to the Cheongju P&T7 advanced-packaging facility investment, with the stated purpose of meeting "global demand for artificial intelligence ('AI') memory semiconductors" 6-K filed 2026-07-22: direct competitor capacity coming onto a bottleneck the bull case implicitly assumes Micron Technology Inc will own. The contestable claim is whether the HBM4E volume ramp FQ3-26 release and the multi-year Strategic Customer Agreements referenced on the 2026 Q3 earnings call can hold share and pricing once SK hynix's packaging supply is fully online through the 11/26/2025–12/31/2032 investment window.

What management teams are saying

Micron Technology Inc's management runs every recent discussion of demand, pricing, and capacity through the same sieve: supply cannot keep up. Sanjay Mehrotra told investors on the 2026 Q3 earnings call that AI demand is pushing DRAM and NAND data center bits TAM past half the industry for the first time in calendar 2026, and that traditional server demand is being pulled along by agentic AI and broad refresh. He added that the company is still only filling 50% to two-thirds of demand from its key customers. Mark Murphy, the CFO, framed the durability question by pointing to greenfield lead times, the rising HBM trade ratio, and what he called the physical constraint of bringing on new supply.

The pricing evidence lines up. Murphy said on the same call that DRAM prices increased in the mid-sixties percentage range and NAND prices in the high-seventies percentage range quarter-on-quarter, with both moves described as driven by tight industry conditions. He pushed back on the idea that this is a familiar memory cycle, arguing that any new capacity really needs to be greenfield, a physical constraint that takes a lot of time. The pricing framing matters because Micron sits at the supplier end of an AI capex pipeline whose customers are explicitly budgeting for higher memory component costs. Meta raised 2026 capex guidance to $125–145B from $115–135B and cited higher component pricing as one reason for the move.

The HBM4 thesis is the load-bearing claim on the demand side. Mehrotra told investors earlier on the 2026 Q1 earnings call that Micron had already locked in pricing and volume for its entire calendar 2026 HBM supply, including its HBM4, and projected HBM TAM to roughly triple from approximately $35 billion in 2025 to around $100 billion in 2028. He framed the $100 billion milestone as arriving two years earlier than previously expected. The 8-K released June 24, 2026 confirms HBM4 is in high-volume shipments for the lead customer and that HBM4E volume production is expected in calendar 2027.

Capacity expansion is the supply-side response. The 10-Q filed June 25, 2026 catalogs a multi-region buildout: a $1.8 billion acquisition of the Tongluo fab from Powerchip closed in March 2026 and is expected to support meaningful product shipments beginning in mid-calendar 2027; an assembly and test facility in Gujarat that has commenced commercial shipments and is ramping in 2026; a Singapore advanced packaging facility whose groundbreaking was in January 2025 with meaningful capacity in the first half of calendar 2027; a new NAND cleanroom in Singapore with initial wafer output by late calendar 2028; and a New York DRAM site where groundbreaking on the first fab occurred in January 2026, with supply slated for 2030 and beyond. Murphy told investors on the 2026 Q3 earnings call that fiscal 2026 capital spending is now projected to be above $25 billion, with management expecting to continue increasing fiscal 2027 OpEx as R&D ramps for an unprecedented set of long-term opportunities.

Strategic Customer Agreements are the contractual lock-in layer. Mehrotra said on the 2026 Q3 earnings call that Micron signed its first five-year SCA and is in discussions with multiple other customers. Murphy described the structure on the same call as a multi-year agreement with specific commitments, distinct from the typical one-year LTAs of prior cycles. The same 8-K released June 24, 2026 noted that the board declared a quarterly dividend of $0.15 per share, a 30% step-up from the $0.115 cadence of the prior eight quarters.

Peers frame the same bottleneck from the other side of the table, and most corroborate the supply-tight read.

Nvidia Corp's Colette Kress, the CFO, told investors on the 2027 Q1 earnings call that demand for GB300 and VL72 was particularly strong, with frontier model builders and hyperscalers each having cumulatively deployed hundreds and thousands of Blackwell GPUs in what she called the fastest product ramp in the company's history. Her data-center disclosure on the same call put Q1 FY27 data center revenue at a record level, with hyperscale and ACIE roughly splitting the segment. The Vera Rubin ramp, she said, is set to begin in Q3 and then scale. The 8-K filed June 18, 2026 added a $25 billion debt raise across seven tranches, evidence that Nvidia is preserving firepower ahead of its own customer commitments. The 8-K filed May 20, 2026 reports a Q2 FY27 outlook of $91.0 billion plus or minus 2% in revenue, with no Data Center compute revenue from China assumed.

Taiwan Semiconductor Manufacturing Co Ltd's C.C. Wei told investors on the 2026 Q1 earnings call that leading-edge capacity remains tight, with advanced packaging capacity described as very tight as well. The 6-K filed July 16, 2026 reports Q2 CY26 revenue of US$40.20 billion, up 33.7% year-on-year, gross margin of 67.7%, and an advanced-technology (7nm and below) share of wafer revenue at 77%. The same filing guided Q3 CY26 revenue to US$44.6 to $45.8 billion. The deck in the same filing lifts full-year 2026 USD revenue growth to slightly above 40%. Capex is the variable that has stepped up most. The FY2025 20-F (filed April 16, 2026) put FY2025 capex at US$40.895 billion and the 2026 budget at between US$52 billion and US$56 billion. On the 2026 Q1 earnings call, the CFO, Wendell Huang, framed the next three years' capex as significantly higher than the past three years combined.

Broadcom Inc.'s Hock Tan told investors on the 2025 Q4 earnings call that AI semiconductor bookings during the quarter exceeded $30 billion, that Q3 AI semiconductor revenue was guided to about $16.0 billion, and that FY2026 AI semiconductor revenue would land at about $56 billion with FY2027 in excess of $100 billion. The 8-K filed June 3, 2026 reports Q2 FY26 AI semiconductor revenue at $10.8 billion, up 143% year-on-year. The 8-K filed July 6, 2026 extends Broadcom's collaboration with Apple through 2031 for multiple generations of custom ASIC silicon. Tan said on the same call that visibility now runs all the way to 2028, and that the constraint he flagged was lead time, not component shortage.

SK hynix Inc.'s board told investors in a 6-K filed July 22, 2026 that it had approved a Won 7,093,100,000,000 increase in investment for the Cheongju P&T7 advanced packaging plant, equal to 5.88% of equity as of December 31, 2025, with the stated purpose of meeting global demand for AI memory semiconductors. The board resolution cited increased production needs and the acceleration of the cleanroom opening schedule. A separate 6-K filed July 15, 2026 reports an actual issuance amount of Won 39,890,534,790,000 from a third-party ADR allotment, equivalent to US$26,507,100,000 at the closing-date FX. The 6-K filed July 22, 2026 clarified that SK hynix has not pursued an acquisition of Intel's Ohio fabrication facility.

Applied Materials' agreement to acquire NEXX from ASMPT, disclosed on the 2026 Q2 earnings call, is a panel-level advanced-packaging deal aimed squarely at the same bottleneck. Lam Research guided June quarter revenue to $6.60 billion plus or minus $400 million on the 2025 Q3 earnings call, with the quarter-on-quarter step-up implying accelerating memory-fab tool demand. KLA's FY2026 10-K filed February 20, 2026 puts TTM capex meaningfully above the prior year, consistent with memory-fab capacity additions underway.

The two smaller names diverge sharply from the bull case. Formfactor Inc's Mike Slessor told investors on the 2025 Q4 earnings call that probe-card demand from a high-performance compute leader caused that customer to cross the 10% threshold for the first time, with probe cards for HBM applications and DDR sustaining demand. The 8-K filed April 29, 2026 reports Q1 2026 revenue of $226.1 million and guides Q2 to $240 million plus or minus $5 million. Slessor is explicit that pricing and ASP are not driving the gross margin improvement, that utilization is already very high, and that the Farmers Branch site will provide the next leg of capacity beginning late this year and ramping through 2027. Netlist Inc's 10-Q filed May 12, 2026 reports that the term of the SK hynix Supply Agreement supply provisions expired in April 2026 and that purchases continue on a purchase-order basis, and the same filing slips the timing band for incremental industry fabrication capacity from late 2026 or 2027 to late 2027 or 2028.

The peer tape points one direction, and the offset is the competitor capacity coming online. Mehrotra's framing on the 2026 Q3 earnings call, that Micron can fulfill only 50% to two-thirds of demand from key customers in the medium term, is the bull case in a single sentence. The mirror image is SK hynix's P&T7 expansion and the new Tongluo fab, which both add supply into the same bottleneck over the next two to three years.

The numbers turn a once-stuck operating engine into a record supplier

Valuation vs peers (Semiconductors and cross-industry comps)

Valuation anchors

• Indicated dividend yield: 0.1% (latest declared $0.15/share × 4 payments a year = $0.60 annualized, on the $959.48 last close).

Performance vs peers (Semiconductors and cross-industry comps)

Micron Technology Inc management guidance, from the 8-K filed 2026-06-24

The KPI block shows the operating scoreboard the rest of the section is read against. Two metrics are running at the extreme right of their trajectories: adjusted free cash flow reached $18.3B in Q3 2026, up from $0.803B in Q4 2025 and $0.857B in Q2 2025, having been negative in Q2 2023 (−$1.81B) and Q3 2023 (−$1.36B); operating cash flow reached $25.39B in Q3 2026 against $0.024B in Q3 2023, a roughly thousand-fold reset across the cycle. Cash, marketable investments, and restricted cash extended to $30.2B in Q3 2026, triple the $9.72B reported in Q2 2024. The sprint is in investments in capital expenditures, net, which accelerated through fiscal 2026 from $3.13B in Q1 2025 to $3.09B in Q2 2025, $2.66B in Q3 2025, $4.93B in Q4 2025, $4.5B in Q1 2026, and $7.1B in Q3 2026, a pace that, on the company's own framing, is above the $25B management told investors on the 2026 Q3 earnings call for fiscal 2026 capex in aggregate.

The shape of the swing matters more than the end points. Operating cash flow turned from $0.024B in Q3 2023 to $25.39B in Q3 2026, while capex accelerated in the same window: the company both expanded output and converted a higher share of revenue to cash. Adjusted free cash flow of $18.3B in Q3 2026 is the cleanest read on what the cycle is worth once the build is paid for. The cash balance rising from $9.22B in Q3 2024 to $30.2B in Q3 2026, across the same period that capex escalated, is the balance-sheet translation of the operating reset. Mark Murphy told investors on the 2026 Q3 earnings call that the company had reached record levels of cash and investments of $16.7B at the FQ2 quarter end and had reduced total debt by over $5B in the last three quarters, putting it at the strongest net cash position ever and earning two credit upgrades to a solid triple-B. The KPI block corroborates that timing: the run rate of cash generation ramped materially starting in Q1 2026.

The segment mix reads as a different company. Cloud Memory (CMBU) did not exist as a reported segment in FY2021 or FY2022; it was $1.9B in FY2023, $3.8B in FY2024, and $13.5B in FY2025, a 36% share of the year's $37.4B total. Mobile and Client (MCBU) has grown more slowly but more steadily, from $7.4B in FY2023 to $11.7B in FY2024 to $11.9B in FY2025, a 32% share in FY2025. Core Data Center (CDBU) moved from $2.1B in FY2023 to $5B in FY2024 to $7.2B in FY2025, a 19% share. The legacy CNBU and MBU lines were discontinued as segments.

The product mix is more concentrated. DRAM was $20B, $22.4B, $11B, $17.6B, and $28.6B across FY2021–FY2025, ending at a 76% share of FY2025 product revenue. NAND was $7B, $7.8B, $4.2B, $7.2B, and $8.5B, ending at a 23% share. Other Product Sales was $659M, $561M, $356M, $281M, and $297M, a 1% share. The split shows the cycle dependence: DRAM led the FY2025 fold, and NAND remains a meaningful but secondary line.

The peer comparison is harder to read than the KPI block because each company names its segments differently. Nvidia Corp's FY2026 split is Compute and Networking at $193.5B (90%) and Graphics at $22.5B (10%), a different cut entirely, anchored on GPU rather than memory. Broadcom Inc.'s FY2025 split is Semiconductor Solutions at $36.9B (58%) and Infrastructure Software at $27B (42%), a hybrid revenue base whose Software line is a VMware legacy. Netlist Inc's FY2026 split is dominated by Resales of Third Party Products at $180.1M (95%) with Modular Memory Subsystems at $8.6M (5%), a reseller model rather than a fab model. Formfactor Inc's FY2025 split is Probe Cards at $637.9M (81%) and Systems at $147.1M (19%), a different stage of the memory supply chain. Direct product-line competition with Micron is strongest against SK hynix, which uses its own segment structure and is not broken out in the same table.

The valuation table shows where the market is pricing the reset. Micron Technology Inc trades at $1.1T market cap, 21.7× P/E, 11.8× EV/Revenue, and 18.0× EV/EBIT, a clear discount to Nvidia Corp's 32.5× P/E, 20.1× EV/Revenue, and 31.4× EV/EBIT, and to Broadcom Inc.'s 66.0× P/E, 25.6× EV/Revenue, and 59.0× EV/EBIT. On the same table, Formfactor Inc trades at 129.6× P/E, 10.1× EV/Revenue, and 120.6× EV/EBIT, and Applied Materials Inc /De at 52.1× P/E, 15.1× EV/Revenue, and 52.8× EV/EBIT. TTM revenue of $90.3B at Micron Technology Inc sits between Broadcom Inc.'s $75.5B and Nvidia Corp's $253.5B; TTM EBIT of $59.2B is roughly 65% of revenue, a margin profile that the P/E does not fully capture but the EV/EBIT does. Taiwan Semiconductor Manufacturing Co Ltd and SK hynix Inc. have no P/E, EV/Revenue, or EV/EBIT on file in the table; the figures are not reported, and the comparisons stop there.

The performance table adds the trajectory. Micron Technology Inc's last close of $959.48 is +778.5% on 52 weeks and +236.3% YTD, leading the peer set on the year and on the trailing period. Formfactor Inc is +224.2% on 52 weeks and +103.3% YTD, and Netlist Inc is +308.1% on 52 weeks and +184.3% YTD. Nvidia Corp is +27.0% on 52 weeks and +13.8% YTD; Broadcom Inc. is +42.4% on 52 weeks and +15.1% YTD. Revenue growth on the same table: Micron Technology Inc +49.5% YoY against Nvidia Corp's +125.9%, Broadcom Inc.'s +7.9%, Netlist Inc's +28.2%, Formfactor Inc's −11.3%, and Applied Materials Inc /De's −12.6%. The short base is the guidance-implied read on the runway.

The guided numbers are the test. The 8-K filed June 24, 2026 sets FQ4-26 revenue at $49.0B–$51.0B ($50.0B midpoint), gross margin at 86%, GAAP operating expenses at $1.9B, non-GAAP at $1.7B, GAAP diluted EPS at $29.73–$31.73 ($30.73 midpoint), and non-GAAP EPS at $30.00–$32.00 ($31.00 midpoint). The 10-Q filed June 25, 2026 puts the quarter-end balance sheet at $25.0B in cash and equivalents, $26.9B in accounts receivable, $134.1B in total assets, $33.4B in total liabilities, and $100.7B in stockholders' equity. The 86% gross margin and $30.73 EPS midpoint are the loading for the next test: they are the numbers the FQ3-26 results beat, and they are what the guide says the next quarter will land on.

The read-through

Nvidia Corp's Q1 FY27 set the demand ceiling the bull case rests on, and the 8-K filed May 20, 2026 puts data center revenue at a record $75.2B (+21% Q/Q, +92% Y/Y), with the segment split between hyperscale ($37.9B, +115% Y/Y) and AI clouds/industrial/enterprise ($37.4B, +74% Y/Y). Colette Kress told investors on the 2027 Q1 earnings call that Blackwell was "the fastest product ramp in our company's history," and the 8-K filed May 20, 2026 shows Q2 FY27 revenue guided to $91.0B ± 2%. The transmission to Micron Technology Inc is straightforward: Nvidia is the most-likely identity of MU's unnamed 17% FY2025 customer and HBM4 lead customer (one customer = 17% of FY2025 revenue, primarily CMBU FY2025 results), so a +92% Y/Y data-center ramp at the GPU buyer implies that MU's HBM4 high-volume shipments and the HBM TAM Sanjay Mehrotra sized at roughly $35B (2025) to around $100B (2028) on the 2026 Q1 earnings call have a customer of matching scale absorbing them. The conditional: if the Blackwell and Vera Rubin ramps convert Nvidia's data-center dollars at the trajectory Kress described, MU's FY2026 DRAM and HBM bit-shipment forecasts sit on top of a customer base still in acceleration, not saturation.

The supplier side of the same chain moves in lockstep. Taiwan Semiconductor Manufacturing Co Ltd's 6-K filed July 16, 2026 reports Q2 CY26 revenue of US$40.20B (+33.7% Y/Y), with 2nm at 3%, 3nm at 30%, 5nm at 33%, and 7nm at 11% of wafer mix, and advanced (≤7nm) at 77% of wafer revenue. The deck in the same filing guides Q3 CY26 revenue to US$44.6B–$45.8B and lifts full-year 2026 USD revenue growth to "slightly above 40%". The 20-F filed April 16, 2026 sets the FY2025 capex actual at US$40.895B and the FY2026 budget at US$52B–$56B. The transmission is the HBM4/HBM4E base-die substrate: TSMC fabricates the customized logic base dies for HBM stacks, so the +36% Y/Y revenue and 77% advanced-mix print means the HBM4 ramp's substrate is also running tight, supporting MU's framing that HBM bit growth will be greenfield-gated for at least the next two years. The conditional: if TSMC's leading-edge wafer supply continues to track the 12% Q/Q Q2 step, MU's HBM4E CY2027 ramp has the base-die backing to clear that timeline; if 2nm yield slips, the HBM4E roadmap moves with it.

Hyperscaler capex is the second demand-side transmission, and the four most-recent datapoints all corroborate escalation. Alphabet's 8-K filed July 22, 2026 shows purchases of property and equipment stepping from $23.95B (Q3'25) to $27.85B (Q4'25) to $35.67B (Q1'26) to $44.92B (Q2'26), with the same filing reporting a $49.6B equity raise earmarked "for general corporate purposes, including capital expenditures to scale AI infrastructure and global compute." Microsoft's 8-K filed April 29, 2026 reports nine-month FY26 capex of $80.15B, up roughly 69% Y/Y from $47.47B in the prior nine months. Meta's 8-K filed April 29, 2026 raised 2026 capex guidance to $125B–$145B from $115B–$135B, citing "higher component pricing this year": direct disclosure that memory/AI-component pricing is a binding constraint on hyperscaler build. Amazon (AMZN $233.66 -4.57%)'s 8-K filed July 9, 2026 closed a $25B notes sale for general corporate purposes, a balance-sheet precursor to the Q2 capex line AWS will report later. The transmission to MU is that all four of MU's largest disclosed hyperscaler buyers (via the CMBU mix at 36% of FY2025) are explicitly budgeting for higher memory component pricing and the build-out to absorb it. The conditional: if hyperscaler capex holds the +69% to +88% trajectory already on file, MU's DRAM bit demand has a multi-quarter order book behind it; if any one of the four re-anchors lower, the AI memory TAM line on MU's CY2028 outlook takes the same step down.

The equipment vendor tape validates the fab build. Applied Materials Inc /De's 8-K filed May 14, 2026 reports a NEXX acquisition from ASMPT targeted at "panel-level advanced packaging technologies … designed to enable chipmakers and systems companies to build larger-body AI accelerators": the same advanced-packaging step that gates HBM4/HBM4E output. Lam Research (LRCX $319.78 +0.15%)'s 8-K filed April 22, 2026 reports Q3 FY26 revenue of $5.85B (+9.4% Q/Q) and guides the June quarter to $6.60B ± $400M (+12.8% Q/Q at midpoint). KLA (KLAC $218.73 +1.88%)'s 10-K filed February 20, 2026 puts TTM FY26 capex at $387M, +31% Y/Y from $296M. The transmission to MU is supply-chain conviction: each of these vendors is adding capacity into MU's own build cycle (the Tongluo, Gujarat, Hiroshima, Singapore, Idaho, and New York sites), and into the HBM base-die advanced-packaging capacity that backs the HBM4E ramp. The conditional: if AMAT's NEXX deal and Lam's +12.8% Q/Q guide hold, MU's HBM4E CY2027 timeline has the tool-side backing; if LRCX misses its June guide, MU's 1-gamma DRAM node ramp faces a tighter equipment window.

Broadcom Inc.'s 8-K filed June 3, 2026 sets the second customer-cluster reference: Q2 FY26 AI semiconductor revenue of $10.8B (+143% Y/Y), Q3 FY26 AI semiconductor guided to ~$16.0B (+>200% Y/Y), and FY2026 AI semiconductor at ~$56B with FY2027 "in excess of $100B". Hock Tan told investors on the 2025 Q4 earnings call that AI semiconductor bookings in the quarter exceeded $30B and that "visibility runs all the way to 2028." The 8-K filed July 6, 2026 extended the Apple custom-ASIC collaboration through 2031, adding another multi-year ASIC volume commitment. The transmission to MU is the custom-ASIC adjacency: Broadcom's TPUs (for Google) and Meta MTIA blocks consume HBM stacks on the same 12–18 month design cycle Sanjay Mehrotra flagged on the 2026 Q3 earnings call, so a Broadcom FY2027 AI line "in excess of $100B" implies incremental HBM demand that has not yet shown up in MU's FY2026 guide. The conditional: if Broadcom's $30B+ bookings quarter is repeatable, MU's HBM TAM CAGR sits closer to the high end of management's 2025–2028 path; if those bookings slip into 2028, MU's CY2027 HBM4E ramp builds more slowly.

Formfactor Inc's 8-K filed April 29, 2026 is the cleanest read on the equipment-and-consumables demand side of MU's HBM build: Q1 2026 revenue of $226.1M (a record, +32.0% Y/Y), with DRAM probe cards up on HBM applications. Mike Slessor told investors on the 2025 Q4 earnings call that a high-performance compute customer crossed the 10% threshold for the first time, driven by HBM probe cards. The 8-K filed April 29, 2026 guides Q2 2026 revenue to $240M ± $5M. The transmission to MU is the test-stage corroboration: HBM probe-card volume rising implies HBM die output rising at MU's customers. The conditional: if FormFactor's $240M Q2 guide and the +9.4% Q/Q LRCX guide hold, MU's HBM4E CY2027 ramp has multiple independent supply-chain datapoints backing it; if any one misses, MU's HBM bit trajectory loses a corroborating signal.

SK hynix Inc. is the bear mirror. The 6-K filed July 22, 2026 reports a Won 7,093,100,000,000 increase in investment for the Cheongju P&T7 advanced-packaging plant, equal to 5.88% of equity as of December 31, 2025, with the board's stated rationale being "increased production needs and in connection with the acceleration of the opening schedule of the cleanroom." The 6-K filed July 15, 2026 reports an actual issuance amount of Won 39,890,534,790,000 (US$26,507,100,000) from a third-party ADR allotment, dated July 14, 2026. The transmission to MU is direct competitor capacity arriving on the same bottleneck: SK hynix is adding HBM packaging capacity on roughly the same timeline as MU's HBM4E CY2027 ramp. The conditional: if SK hynix's P&T7 investment reaches the 2032 endpoint with the same speed-up of cleanroom opening the board cited, MU's HBM4E pricing window narrows in CY2027–CY2028 and the 86% FQ4-26 gross margin guide becomes harder to defend into FY2027. The second SK hynix 6-K filed July 22, 2026 clarified that the company has not pursued an Intel Ohio fab acquisition, so the U.S. advanced-packaging footprint stays unchanged: the bottleneck relief comes from the Korean P&T7 build, not from U.S. capacity.

Netlist Inc's 10-Q filed May 12, 2026 is the small-cap bear signal: management slipped the timing band for incremental industry fabrication capacity from "late 2026 or 2027" in the FY2025 10-K filed March 19, 2026 to "late 2027 or 2028" in the Q1 2026 10-Q filed May 12, 2026, and reports that the SK hynix Supply Agreement supply term expired in April 2026 with purchases now on a purchase-order basis. The transmission to MU is that the small-cap DRAM reseller's pricing environment remains supply-tight but its single-supplier concentration has materially worsened. The conditional: if Netlist's supply-risk framing holds for the broader reseller tier, MU's direct-customer pricing holds even as competitor HBM supply ramps; if SK hynix widens the SK hynix–Netlist terms, the broader DRAM/HBM supply picture rebalances faster than the 10-Q's "late 2027 or 2028" timing implies, and the MU gross margin glide path flattens.

The net read across the adjacent tape is two-sided. The demand floor. Nvidia +92% Y/Y data center, Broadcom +143% Y/Y AI semis, TSMC +33.7% Y/Y leading-edge, hyperscaler capex +69% to +88%, and the equipment vendors' sequential steps up: corroborates MU's supply-tight framing and the HBM TAM CAGR. The risk ceiling, SK hynix's P&T7 capacity and its US$26.5B ADR-funded balance sheet expansion, is the mirror that could narrow MU's pricing window into CY2027–CY2028. The bear case needs the SK hynix ramp to actually clear the cleanroom acceleration the board cited; the bull case needs Nvidia's Blackwell/Vera Rubin conversion to keep absorbing HBM4 at the $91B Q2 FY27 run rate and Broadcom's $30B+ bookings to repeat.

The balance sheet has shifted from stretched to its strongest ever in three quarters

Micron Technology Inc balance sheet, as of May 28, 2026 (10-Q filed Jun 25, 2026)

Micron's balance sheet at the end of fiscal Q3 2026 carries $25.0B in cash and cash equivalents, $26.9B in accounts receivable, $8.6B in inventory, and $66.7B in total current assets, against $19.5B in total current liabilities and $33.4B in total liabilities, a latest 10-Q that puts total assets at $134.1B and stockholders' equity at $100.7B. The working-capital shift is the cleanest signal of the cycle reset: cash and equivalents of $25.0B against short-term debt of $582M and $19.5B of current liabilities is a current ratio the company has not carried at any point in the prior two years, and the operating-cash-flow line in the KPI block, $25.39B in Q3 2026 against $0.024B in Q3 2023, is the income-statement counterpart. Common stock at $128M and retained earnings at $94.7B on a $100.7B equity base means the book value is being rebuilt almost entirely out of earnings, not out of new share issuance: the share count is essentially flat, treasury stock is $8.5B, and accumulated other comprehensive income is a small negative at −$26M.

Murphy was explicit about the timing of the improvement on the 2026 Q3 earnings call, telling investors that the company had reduced total debt by over $5B in the last three quarters, reached the strongest net cash position ever, and earned two credit upgrades to a solid triple-B. The same prepared remarks said FQ2 ended with record cash and investments of $16.7B, $20B of liquidity once the untapped credit facility is included, $10.1B of debt, and a net cash position of $6.5B after $1.6B of debt paydown in the quarter including redemption of 2029 and 2030 senior notes. The trajectory implied by the KPI block, cash, marketable investments, and restricted cash rising from $9.22B in Q3 2024 to $30.2B in Q3 2026, is the same story in dollar terms: the balance sheet has moved from thin to a record buffer in nine reported quarters.

Capital return has tracked that swing in slow motion rather than at speed. The fiscal-year cash outflows show buybacks running at $2.7B in FY2019, then $176M in FY2020, $1.2B in FY2021, $2.4B in FY2022, before falling to $425M in FY2023, $300M in FY2024, and zero in FY2025; dividends were zero in FY2020 and FY2021, then $461M, $504M, $513M, and $522M across FY2022–FY2025. The total returned to shareholders each year was $929M (FY2023), $813M (FY2024), and $522M (FY2025), a declining pattern even as earnings reset, because the CHIPS Act direct-funding agreement capped buybacks at a time when management chose to delever rather than repurchase. Murphy described that priority on the 2026 Q3 earnings call as reinvesting in R&D, capex, and other strategic investments first, maintaining a strong balance sheet second, and returning cash to shareholders third through a combination of offsetting stock-comp dilution and opportunistic repurchases. The dividend line just turned. The board declared a 30% increase in the quarterly dividend to $0.15 per share on the 2026 Q3 earnings call; the same $0.15 was reaffirmed in the 8-K released June 24, 2026 for the July 21, 2026 payment, after eight consecutive $0.115 quarterly payouts from July 2024 through March 2026. Indicated dividend yield at the $959.48 close lands at 0.1%, a small absolute number, but the direction is the first real signal that capital return is becoming a thing the board will defend, not just maintain. Murphy was direct on the same call that the company is generating return on capital over 30%, headed towards 50%, and expects significant capacity for returning cash to shareholders "through repurchase, a combination of offsetting dilution from stock comp and then opportunistically repurchasing."

The refinancing picture is light. Short-term debt of $582M is the only current debt line on the balance sheet, with no long-term tranche disclosed in the table beyond the $654M of operating lease liabilities, and the FQ2 prepared remarks flagged the redemption of senior notes maturing in 2029 and 2030 alongside $1.6B of debt reduction in that quarter. Total debt of roughly $10.1B at FQ2 end against $16.7B of cash and investments at that point already put the company in net cash, and the FQ3 quarter's $25.39B of operating cash flow plus the KPI block's $18.3B of adjusted free cash flow widens that cushion further. The 10-Q filed June 25, 2026 lists the sources available to fund capex and operations: cash and investments, operating cash flow, CHIPS Act direct funding, customer deposits under Strategic Customer Agreements, and available financing, and concludes they will be sufficient at least through the next 12 months and thereafter for the foreseeable future.

The operating trade-off is the real story, not the headline capital-return number. With the company guiding fiscal 2026 capex above $25B per Murphy on the 2026 Q3 earnings call, and KPI-block capex at $7.1B in Q3 2026 alone against $4.93B in Q4 2025 and $4.5B in Q1 2026, the swing in free cash flow from $0.857B in Q2 2025 to $18.3B in Q3 2026 has been absorbed alongside a step-change in build spend rather than redistributed to shareholders. The test the next four quarters run on is whether the FQ3 free-cash-flow reset holds while the Tongluo, Gujarat, Hiroshima, Singapore advanced-packaging, and New York sites all draw on that same balance sheet, and whether the $10B authorization, the $7.84B already used, and the new $0.15 dividend are the floor of a larger return program or the ceiling of a CHIPS-constrained one.

What needs to go right

The bull case rests on three load-bearing assumptions, each tied to evidence already in the piece: DRAM and NAND pricing stays at the mid-sixties and high-seventies percent range Murphy flagged for FQ2 the 2026 Q3 earnings call; HBM4 volume shipments to MU's lead customer keep absorbing the HBM bit ramp Mehrotra sized at roughly $35B (2025) to around $100B (2028); and the Strategic Customer Agreement framework, of which one has been signed and others are in discussion, translates the supply-tight environment into multi-year revenue visibility before SK hynix's incremental packaging comes online. Each has to clear four specific gates, and each gate has a counterpart risk that would break the thesis.

Gate one: hyperscaler capex holds the trajectory already on file. Alphabet's Q2 FY26 capex of $44.924B against $35.674B in Q1, Microsoft's nine-month FY26 capex of $80.146B (up roughly 69% Y/Y from $47.472B), Meta's raised 2026 capex guidance of $125–$145B citing "higher component pricing this year," and Amazon's $25B notes sale for general corporate purposes Alphabet's Q2 FY26 release, Microsoft's Q3 FY26 release, Meta's Q1 2026 release, Amazon's notes sale filing are the demand-side floor. The assumption breaks if any one of the four re-anchors lower, because MU's CMBU segment at 36% of FY2025 revenue sits directly on that capex line. The bear's read is that memory-component pricing already being cited as a binding constraint on hyperscaler build is itself the leading indicator of demand exhaustion: if capex escalates into 2027 because component prices are rising, the same capex escalates down if prices break. Netlist's 10-Q the most recent filing slipped the timing band for incremental industry fabrication capacity from "late 2026 or 2027" in its FY2025 10-K filed earlier to "late 2027 or 2028" in Q1 2026, a one-year push that mirrors what a hyperscaler would do if its capex assumptions rolled forward into 2028. The gate holds if the four biggest cloud platforms keep buying at the +69% to +88% Y/Y pace already booked; it breaks if any of the four guides 2027 capex down by more than the typical mid-cycle revision. The 86% FQ4-26 gross margin guide assumes the HBM4 mix and pricing Murphy described hold into FQ4, a load-bearing assumption given the SK hynix P&T7 acceleration and TSMC's flagged 2nm and overseas-fab dilution on the Q2 CY26 release, both of which directly bear on the HBM cost and pricing line.

Gate two: SK hynix's packaging build does not arrive faster than MU's HBM4E ramp. The risk mirror is the SK hynix board's approval of a Won 7,093,100,000,000 (5.88% of equity as of 2025-12-31) increase in the Cheongju P&T7 advanced-packaging plant investment, with the stated rationale "to accommodate increased production needs and in connection with the acceleration of the opening schedule of the cleanroom" SK hynix's 6-K filed 2026-07-22, paired with a Won 39,890,534,790,000 (US$26,507,100,000) ADR allotment completed July 14, 2026 SK hynix's 6-K filed 2026-07-15. The gate holds if MU's HBM4E CY2027 volume production, anchored in the FY2025 results and the FY2026 capex budget above $25B Murphy guided on the call, reaches scale before P&T7's cleanroom opening accelerates into MU's pricing window. The gate breaks if SK hynix's cleanroom opening moves forward of CY2027: the HBM4 high-volume shipment base MU is building on the HBM4E roadmap has a two-quarter pricing cushion, not a two-year one. The equipment-vendor tape reads both ways. AMAT's NEXX acquisition from ASMPT, disclosed on the Q2 FY26 call, is a panel-level advanced-packaging deal targeting the same bottleneck. Lam Research's June-quarter guide of $6.60B ± $400M on the Q3 FY26 call, KLA's TTM FY26 capex of $387M against $296M TTM FY25 KLA's 10-K, and FormFactor's Q2 guide of $240M ± $5M FormFactor's 8-K filed 2026-04-29 collectively corroborate that advanced-packaging capacity is the constraint; the same signal also tells the bear case that the constraint is being relieved.

Gate three: the customer base does not consolidate into a single bottleneck. MU's FY2025 10-K the latest annual report discloses one customer at 17% of FY2025 revenue, primarily CMBU, against 10% in FY2024 and no customer at 10% in FY2023. The FY2025 results confirm "HBM4, built on 1-beta DRAM technology, is in high-volume shipments for our lead customer's platform" but do not name the customer. Nvidia's Q1 FY27 data center revenue of $75.246B (+92% Y/Y) and Q2 FY27 guide of $91.0B ± 2% Nvidia's 8-K filed 2026-05-20 puts the most-likely identity of that customer at the GPU buyer absorbing HBM4 at scale, with Blackwell framed as the fastest product ramp in our company's history by CEO Jensen Huang the 2027 Q1 earnings call. The gate holds if Nvidia's Blackwell and Vera Rubin conversion tracks Kress's commentary (Vera Rubin ramping from Q3 FY27 forward) and if MU's five-year Strategic Customer Agreement framework, of which one has been signed and others are in discussion per Mehrotra on the call, spreads the volume risk across multiple chip-customer IP programs. The gate breaks if Broadcom's $30B+ AI semiconductor bookings quarter Broadcom's call commentary, its FY2027 AI line "in excess of $100B," and its Apple custom-ASIC extension through 2031 Broadcom's 8-K filed 2026-07-06 push the ASIC adjacency toward single-customer concentration faster than MU's SCA roster diversifies it. The mirror risk is the converse: if the GPU/XPU chip customer cohort consolidates (AMD's Q1 FY26 data center revenue of $5.775B, +57% Y/Y AMD's 8-K filed 2026-05-05, is the next-largest adjacent name), MU's customer-mix chart compresses even as absolute volumes rise.

Gate four: capex stays funded through the next twenty-four months without diluting equity. The 10-Q filed 2026-06-25 puts cash and equivalents at $25.0B against short-term debt of $582M and total liabilities of $33.4B, with stockholders' equity at $100.7B; the same filing states that cash, investments, cash flow from operations, CHIPS Act direct funding, customer deposits under SCAs, and available financing will be sufficient at least through the next 12 months. The KPI block shows capex stepping from $4.93B in Q4 2025 to $7.1B in Q3 2026, with management guiding fiscal 2026 capex above $25B the call. The gate holds if operating cash flow stays at the Q3 2026 pace of $25.39B (against the FY2025 cash flow line of $5.73B in Q4 2025) and if the $10B share-repurchase authorization, $7.84B of which has been used through May 28, 2026 the 10-Q, is not converted into a defensive equity raise to fund the New York, Singapore, Tongluo, Gujarat, and Hiroshima sites all drawing at the same time. The gate breaks if any one of the fab sites (the Tongluo $1.8B acquisition closed in March 2026 with shipments expected mid-calendar 2027, the New York groundbreaking January 2026 with supply in 2030 and beyond, the Singapore advanced packaging groundbreaking January 2025 with capacity in the first half of calendar 2027, per the 10-Q) overruns its schedule by more than two quarters; capex of that scale does not tolerate project slippage the way a $5B/year capex profile would.

The bear case has the same shape as the bull case, just inverted. The first break is pricing: if DRAM and NAND pricing moves reverse even half of the mid-sixties and high-seventies percent range Murphy described for FQ2 the call, the FQ4-26 guidance of $50.0B ± $1.0B and ~86% gross margin becomes the high-water mark and FY2027 opens with comp pressure against it. The second break is the customer base: the FY2025 results and the call commentary on SCAs together imply a multi-year lock-in, but the contractual visibility Broadcom described on the call (visibility runs all the way to 2028) and Nvidia's Q2 FY27 guide of $91.0B ± 2% are not the same as MU's contractual visibility; the SCAs "have specific commitments," Murphy said, but only one has been signed. The third break is the supply side: Netlist's Q1 2026 10-Q filed 2026-05-12 slipped the broader memory industry's capacity timing by a year even as its own SK hynix supply term expired in April 2026 with purchases now on a purchase-order basis, a small-cap read that the larger DRAM/HBM tier has not yet echoed but that an investor should weight against the P&T7 acceleration. The fourth break is the macro: TSMC's full-year 2026 USD revenue growth was lifted to "slightly above 40%" in the Q2 CY26 release from the "above 30%" given on the prior call, the second upward revision in two quarters, and the most recent deck flags 2nm ramp dilution of 2–3% to gross margin and overseas-fab dilution widening to 3–4% in latter stages; if leading-edge logic margins compress, HBM base-die cost-of-goods moves with them, and MU's gross margin runway narrows even at flat pricing.

Risk/reward

The setup tilts bull. The FQ4-26 guidance of $50.0B ± $1.0B in revenue, ~86% gross margin, and GAAP diluted EPS of $30.73 ± $1.00 at the midpoint is a single-quarter print larger than every full fiscal year Micron has reported through FY2024, and the FQ3-26 8-K reports revenue of $41.46B and GAAP net income of $28.24B ($24.67 per diluted share) at the prior step. Free cash flow reached $18.3B in Q3 2026 against $0.803B in Q4 2025, and operating cash flow of $25.39B in the same quarter is the income-statement counterpart. The balance sheet at the end of that quarter carries $25.0B in cash and equivalents against $582M in short-term debt and $33.4B in total liabilities, with stockholders' equity at $100.7B; the company has reduced total debt by over $5B in the last three quarters, reached the strongest net cash position ever, and earned two credit upgrades to a solid triple-B the 2026 Q3 earnings call. The dividend, raised 30% to $0.15 per share on the same call, is the first real signal that capital return is becoming something the board will defend. Indicated yield of 0.1% is small in absolute terms, but the direction is the read. The drawdown is the cushion: the stock closed at $959.48, off its $1,213.56 high by 20.9%, against Nvidia's −10.0% and Netlist's −26.7% from peak. An investor who buys at $959.48 has already absorbed the bulk of the cycle's first shake-out without the tape having to make a fresh call on demand.

The bear case has more legs than the headline reads. The supply-side mirror is SK hynix's Won 7,093,100,000,000 (5.88% of equity as of December 31, 2025) increase in the Cheongju P&T7 advanced-packaging facility investment, with the board's stated rationale to accommodate increased production needs and in connection with the acceleration of the opening schedule of the cleanroom SK hynix's 6-K filed 2026-07-22, paired with a Won 39,890,534,790,000 (US$26,507,100,000) ADR allotment completed July 14, 2026 SK hynix's 6-K filed 2026-07-15. Netlist's Q1 2026 10-Q slipped the timing band for incremental industry fabrication capacity from "late 2026 or 2027" in the FY2025 10-K to "late 2027 or 2028," and reports that the SK hynix Supply Agreement supply term expired in April 2026 with purchases now on a purchase-order basis. Customer concentration has tightened too: one customer was 17% of FY2025 revenue, primarily CMBU, against 10% in FY2024 and no customer at 10% in FY2023 the FY2025 results, and only one five-year Strategic Customer Agreement has been signed even as others are in discussion the 2026 Q3 earnings call. The 86% gross margin guide is the single most contestable line on the page: SK hynix's packaging build, TSMC's lifted full-year 2026 USD revenue growth to "slightly above 40%" with 2nm ramp dilution of 2–3% to gross margin and overseas-fab dilution widening to 3–4% the Q2 CY26 release, and Broadcom's $30B+ AI semiconductor bookings quarter all point at the same bottleneck from opposite ends.

The risk/reward reads two-sided at the same numbers. On the bull case, the FQ4-26 guided band ($50.0B revenue at midpoint, 86% gross margin) plus the free-cash-flow run rate of $18.3B in Q3 2026 imply the FQ4-26 guidance can clear on the multiple of operating cash flow the KPI block already shows. On the bear case, the SK hynix P&T7 acceleration arrives on the same timeline as MU's HBM4E CY2027 ramp, and the FQ4-26 guidance becomes the high-water mark the FY2027 numbers have to comp against. The valuation table puts the P/E at 21.7× against Nvidia's 32.5× and Broadcom's 66.0×, with EV/Revenue of 11.8× against Nvidia's 20.1× and Broadcom's 25.6×, leaving a real cushion if pricing stays in the mid-sixties and high-seventies percent range Murphy described for FQ2 the 2026 Q3 earnings call; the same cushion erodes fast if DRAM and NAND pricing reverse even half of that move. Short interest of 2.8% is well below Netlist's 11.0% and roughly in line with FormFactor's 2.6% and Applied Materials' 2.5%, so the bearish side of the tape is not heavily armed.

What an investor should monitor next is narrow. The FQ4-26 print is the first test of whether the FQ4-26 guidance is a floor or a ceiling: a clean beat at or above the $30.73 GAAP diluted EPS midpoint and the 86% gross margin would imply the FQ4-26 guidance was conservative and the FY2027 opens with comp cushion; a print at the midpoint or below would imply the HBM4E ramp is absorbing pricing earlier than expected. The second watch item is SK hynix's Q2 2026 earnings call scheduled for July 29, 2026 SK hynix's 6-K filed 2026-07-15, the first post-P&T7 announcement opportunity for the company to quantify the cleanroom opening timeline. The third is the next round of hyperscaler capex disclosures: Alphabet's $44.924B Q2 FY26 print Alphabet's 8-K filed 2026-07-22 is the largest single-quarter hyperscaler spend on file, and any FY2026 capex revision from Microsoft, Meta, or Amazon would shift the demand floor the bull case sits on. The fourth is TSMC's monthly revenue cadence and Q3 CY26 print against the US$44.6–$45.8B guide the Q2 CY26 release, the foundry-side read on whether the HBM4 base-die substrate stays tight. The fifth is Micron's own next capex update, given that fiscal 2026 capex is already guided above $25B the 2026 Q3 earnings call and the Tongluo $1.8B acquisition closed in March 2026 with shipments expected mid-calendar 2027 the 10-Q filed 2026-06-25.

Data comes from SEC filings, earnings-call transcripts and Equibles' deterministic datasets.

This article is for information only and is not investment advice.