Score: a real company having the best year in its history — and a stock priced like the good times are the new normal. Memory has never once worked that way for long.
Micron is the last American memory maker standing, it just printed the biggest quarter in its 48-year history, and it's pouring roughly $50 billion into two new fabs a few miles from where it started in the basement of a Boise dentist's office. The stock is up about 295% in a year. It's also 27% below its record close, its executives have sold roughly $116 million of stock in 90 days without buying a single share, and the whole investment case rests on a shortage that management itself says will end someday — it just can't say when. Here's the whole story: past, present, future, the power bill, and the split question.
Price $884.74 (Oct 30, 2026 close) · Market cap ~$1.0–1.1 trillion · +295% over 1 year · NASDAQ: MU
Micron, 1-year price. Source: Yahoo Finance price history, checked Oct 31, 2026.
The past: four guys, a dentist's basement, and J.R. Simplot's money
Micron was founded on October 5, 1978, in Boise, Idaho, by twin brothers Ward and Joe Parkinson, Dennis Wilson, and Doug Pitman. Three were engineers from Mostek, then a leading memory maker; Joe Parkinson was a lawyer. The first office was the basement under a Boise dentist's office — Joe Parkinson later recalled the nitrous oxide drifting down from the practice above: "We got used to smelling happy gas." Seed money was about $300,000 raised from local businessmen, then $1 million from potato magnate J.R. Simplot, who at one point owned roughly 40% of the company.
The first fab went up in Boise in 1981, turning out a 64K DRAM with the smallest die in the industry — more chips per wafer than anyone else. Micron went public in 1984. Then came the war that killed almost everyone else: between 1978 and 1986, the US share of the DRAM market collapsed from about 70% to about 20% as Japanese makers took over. Intel, Texas Instruments, Motorola — the American giants all exited memory. Micron survived on cheap Idaho land and power and relentless die-size engineering.
The cycles kept trying to kill it. CEO Steve Appleton, who ran Micron from 1994, died in 2012 when his experimental plane crashed after takeoff from Boise Airport. In the 2008–09 crash Micron cut 15% of its workforce, then about 2,000 more jobs. It bought Numonyx for $1.27 billion in stock in 2010, and in 2013 it closed the deal that reshaped the industry: Elpida, Japan's last DRAM maker, bought out of bankruptcy for roughly $2.5 billion all-in, adding about 45% to Micron's manufacturing capacity overnight. By the end of the consolidation, the entire global memory industry was three companies — Samsung, SK hynix, and a company from Boise, Idaho. In December 2025 Micron even walked away from ordinary consumers, announcing the end of its Crucial retail brand to aim everything at AI data centers. That's where the money is, and Micron knows it.
The present: the biggest quarter in company history
Fiscal Q4 2026, reported September 30, 2026: revenue of $54.23 billion, up 379% year over year, non-GAAP earnings of $33.42 a share, and a gross margin of 87% — in an industry that was posting negative gross margins three years ago. Full fiscal 2026 revenue: $133.19 billion. Guidance for the current quarter is $61.5 billion, plus or minus $1.5 billion. Management says more than 75% of planned fiscal 2027 shipments are already committed, backed by 26 strategic customer agreements carrying roughly $32 billion in customer financial commitments, mostly cash deposits. Micron took the #1 spot in our Top 5 Growth Stocks for 2027 on exactly this math.
Two honest footnotes. First, the stock's record close was $1,213.56 on June 25, 2026 — so even after that monster quarter, shares sit about 27% below the high. The market is already arguing about what comes after the boom. Second, the trailing price-to-earnings ratio is only about 12×, which looks cheap until you remember that's 12× peak earnings in the most cyclical industry in tech. With memory, a low P/E at the top is the oldest trap in the book.
Who's behind it
Sanjay Mehrotra. Photo: Micron Technology, CC BY-SA 4.0, via Wikimedia Commons.
The boss. Chairman and CEO Sanjay Mehrotra has run Micron since May 2017 and added the chairman title in January 2025. He co-founded SanDisk in 1988, ran it as CEO from 2011, and sold it to Western Digital for about $19 billion in 2016 — a clean exit, and in this check no fraud, blowup, or settlement history attaches to him personally. He holds 70+ patents and is, by any fair read, one of the two or three most credible operators in memory.
Skin in the game: thin, and getting thinner. Insiders own just 0.24% of the company; institutions hold about 80% (BlackRock ~9.3%, Vanguard entities ~9.5%). And over the last 90 days, insiders sold roughly 123,545 shares for about $115.8 million — with zero open-market buys. Mehrotra himself sold about 200,000 shares across May–August 2026 for roughly $172–180 million. Most of it is pre-planned selling into a historic run, which is what any sane person would do with that much paper wealth. But nobody at the top is adding. Filed under: noted.
The skeletons, because we always check. The big one predates Mehrotra by about 15 years: Micron was a participant in the 1999–2002 DRAM price-fixing cartel. Samsung paid a $300 million criminal fine, Hynix $185 million, Infineon $160 million, and rival executives went to prison. Micron paid no criminal fine — it was the cooperating leniency applicant, and CEO Steve Appleton admitted on the record that the DOJ found evidence of price fixing by Micron employees. Micron later took an ~$80 million charge to settle the civil class action. There's also a live securities class action (filed over 2023–24 disclosures; current status unverified in this check) and a standing geopolitical wound: since May 2023, China has barred Micron products from its critical-infrastructure operators. On the other side of the ledger, Micron was the victim in the decade's biggest memory IP-theft case — Taiwan's UMC pleaded guilty in 2020 to stealing Micron's DRAM secrets for China's Fujian Jinhua.
Investor relations: investors.micron.com
The Boise bet: $15 billion became $50 billion
Micron's Boise campus at sunset, ID1 rising behind the headquarters buildings. Photo: Micron Technology (official video still), editorial use.
In September 2022 Micron announced a $15 billion leading-edge fab next to its Boise headquarters — the first new US memory fab in over 20 years. By June 2025, under an expanded ~$200 billion US investment plan, Boise's share had grown to roughly $50 billion across two fabs (ID1 and ID2, about $25 billion each), with about $100 billion more slated for Clay, New York. The federal CHIPS award, finalized December 10, 2024, is up to $6.165 billion shared across the Idaho and New York projects — and about $1.2 billion of support was redirected from New York to Idaho's second fab after the New York timeline slipped.
The scale is hard to overstate. Each Boise fab carries about 600,000 square feet of cleanroom; ID1 is a four-story structure more than a third of a mile long, running EUV lithography. Micron swapped general contractors mid-build (Hoffman replaced Exyte in late 2025) and announced another ~$10 billion Boise research lab in August 2026.
ID1 steel going up, with a crane lifting the fab's 160-foot box truss. Photo: Micron Technology, editorial use (cropped).
Now the BogusAPY part: the timeline slipped. At announcement, DRAM output from ID1 was expected in 2025. The current target is first wafers in mid-2027, with ID2 following in late 2028. The budget more than tripled while the first-wafer date moved out about two years. None of that makes the project fake — the cranes in these photos are real — but when you read "American memory renaissance" headlines, price in the delay.
What ID1 is meant to look like finished. Rendering: Micron Technology, via NIST/CHIPS.
The power bill: 445 megawatts looking for a grid
A fab is a machine for turning electricity into memory chips, and ID1 is a very large machine. Idaho Public Utilities Commission staff have estimated the Boise fab load at roughly 445 average megawatts — about 495 MW at a 90% load factor, on a take-or-pay basis. For scale: that's about 72% of Idaho Power's entire large-load growth, and large loads are roughly 88% of all the system load growth Idaho Power expects through 2030. One customer is effectively redrawing the state's power map.
Idaho Power's special contract with Micron was approved on May 8, 2026, under the official principle that "growth pays for growth" — the new load funds its own generation and wires. Intervenors in the case, including irrigators, the City of Boise, and Micron itself, are still fighting over exactly how the costs get allocated. Residential rates rose 9.74% in January 2026, but that was a general rate case, not a charge attributed to the fab — anyone who tells you your power bill already went up because of Micron is ahead of the evidence. Water is the quieter strain: the Boise operation uses roughly 4.7 million gallons a day today, headed toward about 10.2 million with ID1 (single-source figure — treat as approximate). The honest summary: the grid strain is real, planned for, and argued about in public by people who will have to pay for getting it wrong.
The global memory game
A Micron DRAM chip up close. Photo: Raimond Spekking, CC BY-SA 4.0, via Wikimedia Commons.
DRAM revenue share, Q2 2026 (TrendForce): Samsung 39.4%, SK hynix 24.9%, Micron 23.3% — just 1.6 points out of second place — with China's CXMT already near 9.5% and climbing (CXMT progress headlines knocked about 5% off Micron's stock in a single October session). In high-bandwidth memory, the AI gold, the ranking is SK hynix ~50%, Samsung 33%, Micron 18% (Counterpoint) — and note the direction: Micron's HBM share fell from 21% in Q1 while Samsung's jumped. Micron's HBM4 is qualified for NVIDIA's Vera Rubin platform and its 2027 HBM output is largely contracted, but the scoreboard says it's running third in the race that matters most.
Meanwhile the shortage is still tightening: TrendForce has Q4 2026 contract prices up another 10–15% for DRAM, 15–20% for NAND, and 23–28% for enterprise SSDs. That is the entire bull case in one paragraph — every AI server needs stacks of HBM and oceans of conventional memory, three companies make it, and the newest supply (Boise) doesn't ship wafers until mid-2027.
The forecast — labeled as opinion, because it is
Analyst consensus price targets cluster around $1,408–$1,622 (aggregator data, late October 2026), with DA Davidson at a Street-high $3,000 (October 7, 2026) and Morgan Stanley at $1,200 on the cautious end. Consensus earnings estimates run to roughly $167–177 a share for fiscal 2027 and $179–207 for fiscal 2028. The bull case: the shortage is structural this time — long-term customer agreements with cash deposits, HBM sold out years ahead, and two Boise fabs plus New York eventually re-shoring a third of the world's memory supply chain with Micron holding the US keys. If fiscal 2027 delivers anything near $170 a share, the stock is trading at about 5× forward earnings and the targets make sense.
The bear case is the cycle, and the cycle has a body count. In fiscal 2023, Micron's revenue fell 49%, it lost $5.83 billion, and gross margin went to −9.1% — from boom to negative margins in about a year. The stock fell 49% in 2022 and about 57% in the 2018 downcycle. Every memory boom in history has ended the same way: everyone builds, supply arrives all at once, prices crater. The "this time is different" argument (AI demand, disciplined oligopoly, contracted volumes) is the strongest it has ever been. It was also the argument in 2018. Both things are true, and which one you weight decides whether $884 is cheap or a cliff edge.
So — is a stock split coming?
Micron has split its stock exactly three times: 5-for-2 in April 1994, 2-for-1 in May 1995, and 2-for-1 on May 2, 2000. That is 26 years without one, across a run from under $10 to nearly $900. At $884.74, a single share costs real money, options contracts control $88,000 of stock each, and the stock is mechanically awkward for small accounts and employee grants — precisely the conditions under which boards usually reach for a split. As of October 31, 2026, though, there is no announcement and no management comment pointing to one; the Motley Fool's October 2026 read was "probably not this year."
And here's the part split headlines skip: a split changes nothing about the company. A 10-for-1 split turns one $884.74 share into ten $88.47 shares. Same factories, same earnings, same market cap, same slice of the pie — cut into smaller pieces. Splits tend to pop a stock short-term because they feel cheaper and pull in smaller buyers, and they genuinely help options liquidity. That's it. If Micron splits, it's a marketing and plumbing decision, not a raise. The capital event with actual teeth is nearer and quieter: Micron's expanded ~$35 billion buyback is barred by its CHIPS terms from starting until December 9, 2026. Watch that date, not the split rumor.
The bottom line. Micron is a real business doing something historic — the last American memory maker, booked out into 2028, building $50 billion of fabs in its hometown. It's also a cyclical at peak earnings, run by people currently selling stock, in an industry whose booms have always ended in a flood. Both halves are the story. Price anyone quotes you that only includes one half is a sales pitch.
Get posts like this by email: Subscribe to BogusAPY
Follow BogusAPY on X: @BogusAPY
Check it yourself: links and sources
Official: Micron investor relations — fiscal Q4 2026 earnings release (8-K Exhibit 99.1, September 30, 2026) for the revenue, EPS, margin, and guidance figures; NIST CHIPS program — Micron award (up to $6.165B, finalized December 10, 2024) for the federal funding and project scope.
Reporting and data behind the numbers: Idaho Statesman (founding history, "happy gas" quote); TrendForce (Q2 2026 DRAM revenue shares; Q4 2026 contract-price outlook); Counterpoint Research (HBM shares); Idaho Public Utilities Commission proceedings and Idaho Power filings as reported (fab load estimate ~445 average MW; special contract approved May 8, 2026); Finnhub and Yahoo Finance (price history — $884.74 close, Oct 30, 2026; record close $1,213.56, June 25, 2026); Form 4 filings compiled by MarketBeat/InsiderTrades (insider sales, last 90 days to late October 2026); DOJ and SEC records of the 1999–2002 DRAM price-fixing case; analyst targets via MarketBeat/TIKR aggregators (late October 2026) — analyst opinion, not a promise. Full research dossier with per-fact sourcing kept on file at BogusAPY; prices and shares checked October 31, 2026.
Not financial advice.
Comments
Post a Comment