Burry shorted Micron. The headlines say he's fighting $100 billion in contracts. The filings say something more interesting.

Financial research, not investment advice. Built from Aether search over SEC filings; evidence labels are inline.

Run your own SEC filing analysis with Aether: aether.evidinvest.com maps filings into evidence labels so you can separate confirmed facts from inferred relationships.

The trial nobody is running

If you hold Micron, or you've been watching it climb all year, this week put you in an uncomfortable spot. The most famous short seller alive disclosed a bet against the most-watched stock in America — press reports had MU as the most-searched US ticker in both May and June, and the single most-traded US stock on July 2, the day after Michael Burry's disclosure landed (press-reported, Benzinga). And immediately, both sides started shouting numbers at each other.

The bears quote Burry: memory is in "bubble territory," the cycle always ends, he's positioned for a 30–41% decline in semiconductors. The bulls quote the counter-headline: Micron has 16 Strategic Customer Agreements, roughly $100 billion of minimum contracted revenue through 2030, and $22 billion of prepayments — how do you short a company whose revenue is contractually locked?

Here's the problem. One side of that argument is quoting a Substack. The other side is quoting articles about filings. Almost nobody in the past 72 hours has quoted the filings themselves — and when you do, the picture changes in both directions. This piece is the fair trial: what Micron actually filed with the SEC, what it didn't, and where each side's numbers come from. We ran the primary-source search on Aether; every filed figure below links to sec.gov.

We are not going to tell you whether Burry wins. The verdict here is on which numbers are real.

Exhibit A: what Micron actually filed about the contracts

Micron's fiscal Q3 2026 10-Q (period ended May 28, 2026, filed June 25) is the primary document, and its Note 14 is the most important half-page in this whole fight. Here is what it discloses, in the filing's own structure (10-Q, 2026-06-25, Note 14 — Revenue and Customer Contract Liabilities). SEC-filed:

• Micron "recently executed certain strategic customer agreements, including agreements executed subsequent to May 28, 2026."
• The agreements are "structured as take-or-pay agreements, with binding commitments for specific volumes over the multi-year contract terms" and "include contractually enforceable volumes."
• "Pricing for most agreements is either fixed, or is subject to minimum and maximum pricing. A minority of the agreements do not have any fixed pricing or price bands, as pricing for those agreements is subject to market conditions."
• "As of May 28, 2026, the transaction price allocated to our remaining performance obligations was approximately $5 billion, of which $422 million has been recognized as contract liabilities."
• "Approximately one-third of the remaining performance obligations as of May 28, 2026 are expected to be recognized as revenue over the next twelve months."

The MD&A adds the pricing architecture: "The largest agreements generally have a ceiling price for existing products that approximates the market price in the second calendar quarter of 2026, and a floor price through the term of the agreement," and management expects gross margins from the price-banded agreements, "even at floor pricing levels, to yield gross margins well above our peak quarterly margins in any past cycle" (10-Q, 2026-06-25, Item 2). SEC-filed — though the margin sentence is a forward-looking expectation, not a realized result.

And the money: "In connection with these strategic customer agreements, we expect to receive cash deposits and related financial commitments of $22 billion for agreements concluded to date. Approximately $18 billion of these commitments will be in the form of cash deposits" (10-Q, 2026-06-25, Item 2 — Liquidity). SEC-filed — note the tense: expect to receive, not received.

The verdict on the headline numbers

Now line the press claims up against the documents:

We verified this directly, at the source. Micron's EDGAR feed shows no 8-K after the June 24 earnings release — every subsequent filing through July 2 is an insider Form 4 or 144 (EDGAR, CIK 0000723125) — and the June 24 8-K's only exhibit is the press release, which contains neither the $100 billion nor the agreement count. Both live one layer down, in the call's prepared remarks: 16 SCAs signed, 14 of them totaling roughly $100 billion of cumulative minimum-price revenue, and a remaining-performance-obligation figure of approximately $100 billion once the agreements executed after quarter-end are included. That last clause is the whole reconciliation: the filed 10-Q shows ~$5B because it measures May 28, and most of the wall was signed in the four weeks after. The $100 billion is real, spoken by management — but until the FY2026 10-K lands, it is a statement, not a filed disclosure with liability attached. If you are sizing the bull case, that distinction is the trade.

Why the filed number is so much smaller — and why that's the real story

The gap between $100 billion and $5 billion is not evidence of fraud on either side. It's evidence of how differently a legal disclosure and a headline are constructed, and the 10-Q tells you exactly how (10-Q, 2026-06-25, Note 14). SEC-filed:

1. The remaining performance obligations figure "is based on minimum committed volumes and minimum pricing" — the floor of the floor.
2. "Agreements without fixed pricing or price bands are not included" — an entire category of the contracts contributes zero to the disclosed number.
3. Contracts with terms of one year or less are excluded as a practical expedient.
4. And Micron says it outright: the disclosure "is not expected to be indicative of future revenue under these contracts."

So the ~$5 billion is not Micron's estimate of what the contracts are worth. It is the smallest number the accounting rules let Micron promise: minimum volumes, at minimum prices, only for agreements with enforceable pricing, only beyond one year. For scale, Micron guided a single quarter to $50.0 billion ± $1.0 billion of revenue (8-K Exhibit 99.1, 2026-06-24). The enforceable, minimum-priced, filed floor of the entire multi-year contract regime is about five weeks of current revenue.

That cuts both ways, and honesty requires saying so. Bulls can fairly argue the filed floor wildly understates the economics — the $22 billion of expected deposits is real filed money that customers don't hand over for a $5 billion relationship. Bears can fairly argue that when you strip the press framing down to what is legally enforceable at minimum pricing, the "locked-in revenue" story shrinks by an order of magnitude. Both readings come from the same paragraph. That paragraph — not the Substack, not the aggregators — is the ground truth.

The case as filed: bulls

To be clear about what Burry is betting against, the filed record is formidable (8-K Exhibit 99.1, 2026-06-24; 10-Q, 2026-06-25). All SEC-filed:

• Fiscal Q3 2026 revenue of $41,456 million, up from $9,301 million a year earlier; GAAP gross margin 84.6%; net income $28,243 million ($24.67 diluted EPS).
• Guidance for fiscal Q4: revenue $50.0 billion ± $1.0 billion, gross margin approximately 86%, GAAP diluted EPS $30.73 ± $1.00.
• HBM4 "is in high-volume shipments for our lead customer's platform", with qualification samples at multiple other customers. The filing does not name the lead customer, and neither will we.
• Nine-month operating cash flow of $45,702 million; quarter-end cash and investments of $30.13 billion; adjusted free cash flow of $18.3 billion in the quarter alone.
• The press release's own headline framing: "Micron executes transformational Strategic Customer Agreements," with CEO Sanjay Mehrotra saying the multi-year agreements "will significantly enhance the durability and predictability of Micron's strong financial performance."

A company printing an 80.4% operating margin, guiding higher, holding $30 billion of cash, and collecting billions in customer deposits is not a fragile short target on its face. That is the filed bull case.

The case as filed: bears

Here is what almost no one quoting the $100 billion headline mentions: the same 10-Q files a substantial bear case, in Micron's own words.

The contracts constrain Micron too. The risk factors state that the strategic agreements "could also constrain our available supply and limit our flexibility to respond to changes in market conditions, and if customers fail to meet their purchase commitments, we may need to enforce our contractual rights, which could result in litigation or disputes," with failure to perform exposing Micron to "contractual damages or other financial consequences" (10-Q, 2026-06-25, Item 1A). SEC-filed. Take-or-pay is only as good as the counterparty's willingness to take or ability to pay — and Micron itself flags counterparty default risk on customer receivables in the same section.

The ceiling is as real as the floor. The largest agreements cap pricing at roughly the market price of the second calendar quarter of 2026 (10-Q, 2026-06-25, Item 2). SEC-filed. In other words, Micron sold its customers protection against exactly the price spike it is currently enjoying. If memory prices keep rising, contracted volumes don't ride the increase; if prices crash, the floor holds only for the price-banded agreements — and a minority of agreements float entirely with the market.

Receivables tripled. The balance sheet shows receivables of $31,025 million as of May 28, 2026, versus $9,265 million at the end of August 2025 (10-Q, 2026-06-25, Balance Sheets). SEC-filed. Roughly $31 billion of the AI boom is money customers still owe. Alongside it, "consideration payable to customers, including pricing adjustments and returns" grew to $3.32 billion from $1.19 billion (Note 14). SEC-filed.

Concentration is filed, direction noted. "Revenue from one customer was 10% and 16% (primarily included in the CMBU segment) of total revenue for the first nine months of 2026 and 2025, respectively" (10-Q, 2026-06-25, Note 17). SEC-filed. The filing does not name the customer, and we won't guess. Concentration actually fell as revenue broadened — a point for the bulls inside a bear exhibit — but a tenth of revenue from one buyer is still a filed fact.

The supply response is funded — by Micron itself. Micron estimates capital expenditures "net of proceeds from government incentives, to be approximately $27 billion in 2026," with $19.6 billion already spent in nine months, versus $10.2 billion a year earlier (10-Q, 2026-06-25, Item 2 and Cash Flows). SEC-filed. And the risk factors concede the classic cycle mechanics: "industry oversupply" from elevated customer inventories, plus "the threat of increasing competition and DRAM and NAND oversupply due to significant investment in the semiconductor industry, including by the Chinese government and various state-owned or affiliated entities, such as CXMT and YMTC" (Item 1A). SEC-filed. Press reports add that Samsung and SK Hynix are part of a roughly $646 billion Korean capex program — that figure is press-reported, not in any SEC filing we searched.

Burry's side of the ledger

Everything specific about Burry's position comes from his Substack and the financial press, not from a regulatory filing — his fund deregistered as an investment adviser, so there is no current 13F to check. Label accordingly. As press-reported: the short was disclosed around July 1, with MU closing at $1,051.87 that day; his broader positioning reportedly includes put options on the SOXX semiconductor ETF expiring March 2027 at strikes implying roughly a 30–41% decline, and bearish positions on Nvidia and Applied Materials; his framing is that AI-driven memory demand has pushed the sector into "bubble territory."

Steelman it, because the honest version is not stupid. Memory has been the most violently cyclical business in technology for four decades, and every cycle has ended the same way: high prices fund capacity, capacity arrives, prices collapse. The filings support the mechanism — Micron's own $27 billion capex plan is the supply response, filed. A year ago Micron earned a 37.7% gross margin; the same company, the same fabs, now files 84.6%. Burry's bet, stripped of drama, is that 85% gross margins in a commodity-adjacent industry are self-extinguishing.

But if the trade rests on the idea that the contracts are fake armor, the record complicates it. The take-or-pay structure, the binding volumes, the floors, and the $22 billion of expected deposits are filed — that is not headline vapor. And the $100 billion is management-stated, in prepared remarks, with specifics attached: 14 of 16 agreements, minimum pricing, terms through calendar 2030. What separates the two is legal weight and timing. As of the filed record, the enforceable minimum is ~$5 billion, because the wall was signed after the quarter closed; the $100 billion version of it exists so far only in remarks that carry no filing liability, from counterparties the company will not name. Burry is fighting a wall that is probably real but not yet auditable — and the prepared remarks add one more detail the bulls rarely quote: the deposits get returned to customers over time, toward the back half of the agreement terms. They are security, not revenue.

What survives, on both sides

Where the short thesis survives contact with the filings: the ceiling on the largest agreements caps contracted upside near Q2-2026 prices; a minority of agreements have no price protection at all; the disclosed enforceable floor is small relative to the revenue run-rate; receivables have tripled; the industry's supply response — Micron's included — is filed and funded; and Micron's own risk factors describe the oversupply scenario in detail.

Where it collides with filed fact: the contracts exist, are binding, are take-or-pay, and carry floors that management expects to hold margins above any previous cycle peak even in a downturn; customers are putting up $18 billion of expected cash deposits; and the company is guiding to $50 billion quarters with $30 billion of cash on hand.

And where both sides are talking past each other: the 16 agreements and the $100 billion are management-stated in prepared remarks — real numbers, spoken with specifics, but not yet in any document filed under securities liability. The filed record says ~$5 billion, measured before most of the agreements were signed. The FY2026 10-K, due this fall, is where the two must reconcile: if the ~$100 billion RPO shows up there, the wall becomes auditable; if it comes in materially lower, the prepared remarks were the high-water mark. Either way, that filing — not this week's shouting — is the verdict.

Headlines move prices. Filings size the exposure.

On July 2, the most-traded stock in America moved on a Substack post and a wave of articles quoting other articles. The actual disclosure — the one with securities-law liability attached — had been sitting on EDGAR for a week, and it says something subtler than either camp's version: the contract regime is real, the filed floor is deliberately tiny, the deposits are expected rather than banked, and the bear case is co-filed in the same document as the bull case.

You don't need to take our word for any of this. Run the search yourself on Aether — query Micron's strategic customer agreements — and read Note 14 of the Q3 FY2026 10-Q. It is half a page. It will tell you more about what's actually promised than everything published this week combined.

Research, not investment advice. We hold no position in MU and make no price forecast. All filed figures trace to the linked sec.gov documents; figures labeled press-reported are not verified against any SEC filing.