CBRSCerebras Systems Inc.
Industry AI semis Updated 24 Sep 2026
The idea
~62x EV/LTM revenue with no earnings; the base case models below the price and the bull case needs AWS to scale as the term sheet contemplates and customer concentration to broaden.
Events to watch
- The Q2 2026 prepared remarks (only a UBS restatement of 'revenue up more than 3x next year' is in hand)
- 36.1m customer-warrant shares vesting
- first S-3 eligibility
How we got the target
Dossier: base below the traded price; bull $5.50 EPS × 55x on $8bn FY28 revenue. Time frame: FY2028.
Target: from the research notes. Prices: market data get_latest_stock_price, close of 2026-09-22 (fetched 2026-09-23)
What would prove it wrong
6 warning signsThe remaining performance obligation stops growing.
Details
$25.0bn at 2026-03-31, $25.4bn at 2026-06-30. It is disclosed every quarter in the revenue note and it is the OpenAI relationship made visible. A flat or falling RPO is the earliest sign of an MRA renegotiation, and it would arrive quarters before revenue showed it.
The Working Capital Loan's current portion is not drawn down by delivery.
Details
$736.0m of the ~$1.0bn sits in current liabilities at 2026-06-30, and it is repayable in compute rather than cash as capacity is delivered. If it stays in current liabilities across two quarters, capacity is not being delivered on the contracted schedule — and non-delivery is precisely what gives OpenAI the right to terminate tranches.
Concentration transfers instead of breaking.
Details
G42 + MBZUAI were 86% of FY2025 revenue and are related parties to each other. If FY2027 discloses OpenAI at a similar share with no third customer above 10%, the risk profile has not improved at all — only the counterparty's name has changed.
The material weaknesses are not remediated in the FY2026 10-K.
Details
They cover revenue recognition, inventory costing, data-centre asset accounting and equity administration — the four areas that produce every number above. A second year unremediated, in the first annual report as a public company, is a different fact from an IPO-stage disclosure.
Customer-warrant contra-revenue outruns revenue growth.
Details
$2.0m in Q1 2026, $44.3m in Q2, with up to ~$2.74bn of OpenAI warrant value still to amortise. If reported revenue keeps falling sequentially while gross bookings rise, the company is buying its backlog with its own stock and the reported top line will never show the growth the backlog implies.
The "core" yardstick moves again, or the warrant add-back outgrows it.
Details
The bridge went from two add-backs in FY2025 to five in Q2 2026, and the customer-warrant add-back was 21% of Q2 core revenue ($44.3m of $209.9m, EX99-Q2 reconciliation). A sixth add-back, a redefinition of an existing one, or the warrant add-back above 25% of core revenue in any quarter means the company is beating a guide on a number it keeps redrawing — S8's "conservative" would then be re-scored on GAAP.
How far along
7 of 10 steps done · 3 done in a group review
Next step
the Q3 2026 release (expected mid-November 2026). It is the second guide cycle S8 needs: score Q3 core revenue against $214–216m, core GM against 38–40% and core op margin against (25)–(23)%, check whether a sixth add-back appears in the bridge, and read DSO a second time. (The 2026-09-21 gap — the "3x" guide known only from an analyst's paraphrase — was closed on 2026-09-24: it is in the CEO's and CFO's prepared remarks.)
Needs your decision
S6 product test / expert call, S7 conviction. open Open (time): S8's FY2026 rows (first guide $855–865m core revenue, 38–41% GM, (28)–(32)% op margin) until the FY2026 release; FY2027 "more than triple" until FY2027.
Failed checks
none. S0/S4/S7 are not closed by design (cluster stages, carried forward). S3 closed at gate-minimum on the recent-IPO path — see the note there. ⚠️ Threads to pull (dodge log): customer concentration dodged on both calls (Barclays Q1, UBS Q2) and AWS deal size/economics dodged on both (Needham Q1, TD Cowen Q2) — the FY2026 10-K's significant- customer table and any AWS definitive-agreement exhibit settle them. ⚠️ EQ: 4 flags (S3 line); the "core" framework presents revenue before the warrant concessions paid to the customers who make it.
Who runs it
The Front OfficeStep by step
- S0How it does as AI gets cheapDone in a group review
carry-forward (cluster: AI-capex complex — cluster pass, 2026-09-20)
cluster verdict is.
⚠️ SPLIT. That file's Baker lens names CBRS as the one member that passes "different AND hard" outright (wafer-scale is the exemplar it cites by name). Owner checks dropped 2026-09-21 (README §Stage 0 override).
- S1Where the idea came fromDone
watchlist (owner, Apple Stocks, 2026-09-20), note "recent IPO; S-1 is the primer"; no Tier 0 promotion.
- S2Quick deal-breaker checkDone
FY25 rev $510.0m +76%, GM 39%, GAAP net income $237.8m but non-GAAP net LOSS $75.7m. Net cash ~$7.0bn against a ~$100m quarterly cash operating loss. Survives on the balance sheet, not on the business.
⚠️ Q2 2026 revenue fell 7% sequentially ($193.4m → $180.1m) and reported GM collapsed to 14.2% — but $377.0m of IPO stock comp and $44.3m of customer- warrant contra-revenue explain almost all of it; ex-both, GM ~38%. Those adjustments are derived in the dossier, not reported by the company. Capital allocation: NEUTRAL — no acquisitions, no goodwill, no buyback programme, no dividend; ~$9.4bn of equity and customer debt since 2024 went into own-cloud capex ($382.7m FY25, $548.9m H1 26) and $416.7m of IPO RSU tax withholding. Only repurchase: Dec-2025 employee tender at $36.23. ROIC negative on every endpoint (GAAP op loss $101.4m FY24 → $145.9m FY25), incremental return n.m. SBC $386.6m in H1 26 with nothing offsetting it; $1.13bn of customer-warrant assets is equity paid to customers (dossier-2026-09-24.md §S2).
- S3Reading the company's reportsDone
(gate-minimum, recent-IPO variant)
424B4 risk factors + use of proceeds + MD&A overview replace the 10-K (there is none). Q2 2026 10-Q revenue, warrant, loan and SBC notes; one transcript (Q2 2026, turns 2-11 of 31). Three disclosed risks named with locators. No proxy exists.
Full notes
Prepared remarks not read, so the "3x next year" guide is an analyst paraphrase; no efts.sec.gov sweep. (Closed 2026-09-24: both calls read in full; the 3x is in CEO and CFO prepared remarks.) EQ: 4 flags — bridge grew from 2 add-backs (FY25) to 5 (Q2 26) incl. $22.8m of cash payroll tax, and the customer-warrant add-back is 21% of Q2 core revenue. SBC 104% of H1 26 revenue with FCF negative. FY25 accruals 14.4% of assets on a $363.3m non-cash gain; customer-warrant asset $152m → $1,128m in six months. Watch: DSO 29 → 62 days in one quarter (dossier-2026-09-24.md §Earnings quality).
- S4Rivals and suppliersDone in a group review
carry-forward (cluster: AI-capex complex — cluster pass, 2026-09-20)
- S5Who owns the stockDone
first-ever 13F quarter (297 filers, 2026-06-30), 13D/G census, Form 4 and Form 144 cadence against the lock-up, FINRA short interest, and the absence of any shelf, ATM or buyback.
Full notes
Dilution is 36.1m customer-warrant shares plus a 42.65m-share plan reserve, not a shelf.
⚠️ Class B carries 20 votes and ~99.2% of the voting power.
- S6What workers and customers sayDone
carry-forward (cluster: AI-capex complex — cluster pass 2: S6 for 41 names, S0/S4/S7 for three, 2026-09-21)
and this pass CONFIRMS its CBRS row from the primary document: the 62.0% / 24.0% / 86% concentration figures that file carried as third-party search-summary evidence are the issuer's own. Product test / expert call open OPEN (user)
- S7The case for it, and pricesDone in a group review
carry-forward (cluster: AI-capex complex — cluster pass, 2026-09-20)
conviction open OPEN (user) · valuation (bear/base/bull; moved from S8 2026-09-23): ~62x EV/LTM revenue (~73x fully diluted) at $208.315 (2026-09-21); no earnings multiple exists.
Full notes
Base case models BELOW the traded price (~$177); bear ~$55, bull ~$300.
- S8Do the bosses keep their word?Done
8 rows: 4 beat, 4 met, 0 missed, 0 withdrawn
conservative, on ONE guide cycle (first guide 2026-06-23 → Q2 result 2026-08-12) and on "core" metrics introduced the same day (GAAP revenue fell 7% q/q in the "beat" quarter)
Full notes
Current team made 8 of 8 (Feldman CEO, Komin CFO; no 8-K Item 5.02). FY26 guide raised on all three metrics, scored at FY26 release against $855–865m. 7 dodges in 2 calls; customer concentration and AWS deal size/economics dodged in both. Provisional — re-test on the Q3 release (Nov 2026).
- S9What to watch nextDone
trigger written. Tier 0 EDGAR sweep covers the feed (cik 2021728)
Sources and notes
Sources 6 documents
- Full notes, 24 Sep 2026
- Full notes, 21 Sep 2026
- AI-capex complex — cluster pass, 2026-09-20
- AI-capex complex — cluster pass 2: S6 for 41 names, S0/S4/S7 for three, 2026-09-21
- EDGAR filings CIK 2021728
| Short name | Document | Where it came from | Read on |
|---|---|---|---|
| 424B4-2026-05-14 | Cerebras IPO prospectus (Rule 424(b)(4)), filed 2026-05-14 — 30,000,000 Class A shares at $185.00 | https://www.sec.gov/Archives/edgar/data/2021728/000162828026035214/cerebras-424b4.htm | 2026-09-21 |
| 10Q-Q2-26 | Cerebras 10-Q for the quarter ended 2026-06-30, filed 2026-08-12 | https://www.sec.gov/Archives/edgar/data/2021728/000162828026056357/cbrs-20260630.htm | 2026-09-21 |
| CALL-Q2-26 | Cerebras Q2 2026 earnings call, 2026-08-12 (speaker-labelled transcript; first call as a public company) | Equibles GetEarningsCallTranscript CBRS — https://equibles.com/stocks/cbrs/earnings-calls | 2026-09-21 |
| 13F-Q2-26 | Institutional holders by manager, 13F report date 2026-06-30 — the first quarter with any (297 filers) | python ledger/research/tools/threespread.py holders CBRS | 2026-09-21 |
| 13D-2026-08-18 | Schedule 13D/A filed 2026-08-18 by Eclipse Continuity GP I, LLC (event date 2026-08-14) | https://www.sec.gov/Archives/edgar/data/2021728/000119312526355477/0001193125-26-355477-index.htm | 2026-09-21 |
| EDGAR-INDEX | Cerebras EDGAR filing index (CIK 2021728) — the whole history is 2 S-1s, 1 424B4, 2 10-Qs, 3 8-Ks | https://data.sec.gov/submissions/CIK0002021728.json | 2026-09-21 |
Links to other companies 18 links
| From | Link | To | Date | Where it says so |
|---|---|---|---|---|
| Robert Komin | officer of | CBRS | 2026-08-12 | 424B4-2026-05-14 — Management — Executive Officers and Employee Director (biography); role confirmed by 10Q-Q2-26 Signatures |
| Andrew Feldman | officer of | CBRS | 2026-08-12 | CALL-Q2-26 — opening remarks and first Q&A exchange |
| Mohamed bin Zayed University of Artificial Intelligence (MBZUAI) | channel concentration | CBRS | 2025-12-31 | 424B4-2026-05-14 — Risk Factors — 'A substantial portion of our revenue is driven by a limited number of customers' |
| Group 42 Holding Ltd (G42) | channel concentration | CBRS | 2025-12-31 | 424B4-2026-05-14 — Risk Factors — customer concentration |
| Group 42 Holding Ltd (G42) | context | Mohamed bin Zayed University of Artificial Intelligence (MBZUAI) | 2025-12-31 | 424B4-2026-05-14 — Risk Factors — customer concentration |
| OpenAI OpCo, LLC | customer of | CBRS | 2026-06-30 | 10Q-Q2-26 — Note 3 - Revenue |
| OpenAI OpCo, LLC | financing partner | CBRS | 2026-01-31 | 10Q-Q2-26 — Working Capital Loan note |
| OpenAI Working Capital Loan — ~$1.0bn secured promissory note, 6% | litigation watch | CBRS | 2026-06-30 | 424B4-2026-05-14 — Risk Factors — customer concentration, Working Capital Loan paragraph |
| CBRS | warrant holder | OpenAI Warrant — up to 33,445,026 Class N shares at $0.00001 | 2026-06-30 | 10Q-Q2-26 — Warrants note, OpenAI Warrant paragraph |
| Amazon Web Services / Amazon.com NV Investment Holdings LLC | customer of | CBRS | 2026-08-12 | CALL-Q2-26 — Q&A, Joshua Buchalter (TD Cowen) exchange |
| CBRS | warrant holder | AWS Warrant — up to 2,696,678 Class N shares at $100.00 | 2026-06-30 | 10Q-Q2-26 — Warrants note, AWS paragraph |
| Advanced Micro Devices | technology partner | CBRS | 2026-08-12 | CALL-Q2-26 — Q&A, TD Cowen exchange on the AMD engagement |
| Eclipse Continuity GP I, LLC / Eclipse Operations, LLC | holds | CBRS | 2026-08-18 | 13D-2026-08-18 — EDGAR filing index, 'Eclipse Continuity GP I, LLC (Filed by)' |
| Coatue Management LLC | holds | CBRS | 2026-06-30 | 13F-Q2-26 — row 3, COATUE MANAGEMENT LLC; https://www.sec.gov/Archives/edgar/data/1135730/000091957426005478/ |
| Atreides Management, LP | holds | CBRS | 2026-06-30 | 13F-Q2-26 — row 5, Atreides Management, LP; https://www.sec.gov/Archives/edgar/data/1777813/000177781326000009/ |
| Tiger Global Management LLC | holds | CBRS | 2026-06-30 | 13F-Q2-26 — row 7, TIGER GLOBAL MANAGEMENT LLC; https://www.sec.gov/Archives/edgar/data/1167483/000091957426005427/ |
| CBRS | governance structure | CBRS | 2026-05-14 | 424B4-2026-05-14 — prospectus cover page |
| CBRS | event | CBRS | 2026-05-14 | 424B4-2026-05-14 — Risk Factors — internal control over financial reporting |
History 3 entries
- 2026-09-21 — first pass, slot
names-ai-capex-3, model-drafted unattended by the cloud research routine. S1, S2, S3 (gate-minimum, recent-IPO variant), S5 and S9 closed; S0/S4/S7 carried forward from the 2026-09-20 cluster file and S6 from the 2026-09-21 one; S8 modelled and left open. Evidence and access limits in dossier-2026-09-21.md. Nothing here is human-verified. - 2026-09-23 — stage redefinition, owner decision (WATCHLIST_PLAN.md §25): S8 is now Management credibility and is not yet run for this name; the old S8 valuation figures moved to the
valuation:line under S7, and any proposed buy price or position size was removed (the pipeline no longer drafts either). No research was redone. S2 capital allocation and S3 earnings quality are also pending, in the same backfill slot. - 2026-09-24 — slot
backfill-2026-09-23-3, model-drafted by a manually started cloud session; nothing here is human-verified. Wrote S8 (8 rows, conservative on one guide cycle, ✅), S2 capital allocation (neutral) and S3 earnings quality (4 flags); glyphs of S2 and S3 unchanged. Both market data transcripts read in full in §3.1 order and the dodge log started (7 dodges, two topics in both quarters). Added a kill criterion on the non-GAAP yardstick and an officer_of edge for the CFO. S7 valuation unchanged (no over-promise, so no haircut). Evidence in dossier-2026-09-24.md; Equibles calls: 2.
Full notes, 24 Sep 2026 18 min read
CBRS — dossier, 2026-09-24 (backfill: S8, S2 capital allocation, S3 earnings quality)
Slot backfill-2026-09-23-3. Model-drafted by a manually started cloud session; nothing here is human-verified. This file holds only the three checks added on 2026-09-23 (WATCHLIST_PLAN.md §25) and the transcript reading they needed; every other stage stands as written in dossier-2026-09-21.md. Anything from background knowledge is marked [background]. Figures marked derived are computed here from the cited lines, not reported by the company.
Sources read this run (all EDGAR unless stated; User-Agent: ddboard-research <owner email>):
| Key | Document | Locator style |
|---|---|---|
| EX99-Q1 | 8-K Item 2.02, Ex. 99.1, filed 2026-06-23 (acc. 0001628280-26-044941, cbrsannouncesfinancialresu.htm) — Q1 2026 results and the first guide | section heading |
| EX99-Q2 | 8-K Item 2.02, Ex. 99.1, filed 2026-08-12 (acc. 0001628280-26-056186) — Q2 2026 results and the raised guide | section heading / table name |
| 424B4 | Prospectus filed 2026-05-14 (acc. 0001628280-26-035214) | section heading |
| 10Q-Q2 | 10-Q for 2026-06-30, filed 2026-08-12 (acc. 0001628280-26-056357) | note / item |
| XBRL | data.sec.gov/api/xbrl/companyfacts/CIK0002021728.json (10-Q facts only — the prospectus years are not tagged) | tag, period |
| CALL-Q1 | ledger/research/_transcripts/CBRS/FY2026-Q1 (market data, call 2026-06-23, sha256 in _transcripts/index.json) | call date + speaker + exchange |
| CALL-Q2 | ledger/research/_transcripts/CBRS/FY2026-Q2 (market data, call 2026-08-12) | call date + speaker + exchange |
| EQB | Equibles GetGuidance and GetNonGaapBridge (2 calls) | tool output |
market data transcripts label turns by name only. Roles are from EDGAR: Andrew D. Feldman, Chief Executive Officer, President, and Director, and Robert Komin, Chief Financial Officer (424B4 Management table; both sign the Q2 10-Q on 2026-08-12, 10Q-Q2 Signatures). Every quotation below was located by exact substring match in the file or filing named beside it and checked by script before it was written. Filing text was extracted from the HTML with whitespace collapsed.
S8 — management credibility
The record is one guidance cycle long. Cerebras has held two earnings calls. The first guide was given on 2026-06-23 (CALL-Q1; EX99-Q1 "Q2 2026 Financial Outlook" and "Full Year Fiscal 2026 Financial Outlook"); the only guided period that has since closed is Q2 2026, reported 2026-08-12. The prospectus gives no financial outlook (a search of the 424B4 for "preliminary" and "recent developments" returns nothing); its checkable commitments are the use of proceeds and the RSU net settlement. Rows below are scored against the first guide.
The yardstick was chosen the same day as the guide. The Q1 call introduced "core" metrics — GAAP less data-centre pass-through revenue and cost, plus the customer-warrant contra-revenue, less SBC and "onetime items" — and the CFO said "I will only be discussing these core metrics today." (CALL-Q1, Komin, prepared remarks). Every financial promise below is on that basis. On GAAP, revenue fell 7% sequentially in the quarter the company "beat on all metrics" ($193.4m → $180.1m, XBRL RevenueFromContractWithCustomerExcludingAssessedTax, 10-Q Q1 and Q2).
| # | Promise (metric, period, figure) | Made where / by whom | Result and locator | Score |
|---|---|---|---|---|
| 1 | Q2 2026 core revenue ~$194.0m — "In our core business in Q2, we expect core revenue of approximately $194 million, representing year-over-year growth of 88%." | CALL-Q1 2026-06-23, Robert Komin (CFO); EX99-Q1 "Q2 2026 Financial Outlook" | $209.9m core ($180.1m GAAP) — EX99-Q2, "Reconciliation of GAAP Revenue to Core Revenue" | beat (+8.2%) |
| 2 | Q2 2026 core gross margin 36–38% | CALL-Q1, Komin; EX99-Q1 outlook | 40.6% — EX99-Q2, "Reconciliation of GAAP Gross Margin to Core Gross Margin" (GAAP 14.2%) | beat |
| 3 | Q2 2026 core operating margin (30)–(32)% | CALL-Q1, Komin; EX99-Q1 outlook | (16)% — EX99-Q2, "Reconciliation of GAAP Operating Margin to Core Operating Margin" (GAAP (265)%) | beat |
| 4 | Core hardware revenue to fall — "We plan to see decreasing hardware revenue for the next few quarters" | CALL-Q1, Komin, prepared remarks | Core hardware $111.6m (Q1) → $82.1m (Q2) — EX99-Q1 and EX99-Q2 core revenue reconciliations | met |
| 5 | Core cloud margin to fall on rented capacity — "We expect the impact to be a decrease of 10 to 15 margin points" from Q1's 52.9% | CALL-Q1, Komin, prepared remarks | 52.9% → 41.8%, down 11.1 points — EX99-Q2 core gross margin reconciliation | met (inside the range) |
| 6 | Core hardware margin to "return to the low 30s as this contract pricing normalizes" | CALL-Q1, Komin, prepared remarks | 42.0% (Q1) → 38.8% (Q2) — EX99-Q2 core gross margin reconciliation | beat (fell less than guided) |
| 7 | "G&A expense was $9.9 million and will continue to step up significantly next quarter" | CALL-Q1, Komin, prepared remarks | Core G&A $9.9m → $25.3m — EX99-Q2, "Reconciliation of GAAP Operating Expenses to Core Operating Expenses" | met |
| 8 | IPO: net proceeds ~$5,408.5m ($6,221.4m with the over-allotment) and "we will use approximately $416.1 million to satisfy our tax withholding and remittance obligations related to the vesting of such RSUs." | 424B4 "Use of Proceeds", company (written) | Over-allotment taken; "Proceeds from initial public offering, net of underwriting discounts and commissions" $6,232.5m and "Tax withholding from tender offer and initial public offering" $416.7m (the latter also carries the tender-offer withholding) — EX99-Q2 cash-flow statement, six months to 2026-06-30 | met |
Open promises — recorded now, scored when their period closes (not counted above):
| Promise | Made where / by whom | Path so far |
|---|---|---|
| FY2026 core revenue $855–865m, core GM 38–41%, core op margin (28)–(32)% | CALL-Q1, Komin; EX99-Q1 "Full Year Fiscal 2026 Financial Outlook" | Raised on all three on 2026-08-12 to $880–890m, 41–43%, (19)–(17)% (EX99-Q2 "Full Year 2026 Financial Outlook": "Core Non-GAAP Financial Outlook has been raised for all metrics"). Komin on the first call: "So I'm being conservative for the second half as we're still pretty early in the year" (CALL-Q1, Buchalter exchange). Scored against $855–865m at the FY2026 release (~Feb 2027 [background]). |
| Q3 2026 core revenue $214–216m, core GM 38–40%, core op margin (25)–(23)% | CALL-Q2, Komin; EX99-Q2 "Third Quarter 2026 Financial Outlook" | Due ~mid-November 2026. Komin: "we expect Q3 to be the low point for core gross margin before improving significantly in Q4 '26" (CALL-Q2). |
| FY2027 core revenue more than 3x FY2026 — "we expect to more than triple our core revenues in '27" | CALL-Q2, Andrew Feldman (CEO), prepared remarks; repeated by Komin | The dossier-2026-09-21 note that this was only an analyst's paraphrase is closed: it is in both executives' prepared remarks. On the raised FY26 midpoint it implies FY27 core revenue above ~$2.66bn (derived). |
| AWS — "I think you should expect to see AWS' impact in 2027." | CALL-Q1, Feldman, Arcuri exchange | CALL-Q2 and EX99-Q2: Bedrock availability in Q1 2027. Consistent, not yet due. |
| Disaggregated inference "which we expect to begin to deliver in the second half of this year" | CALL-Q1, Komin | CALL-Q2, Feldman: "I think it will be deployed and available in Q4." EX99-Q2 says AMD solution "in production in Q4 2026". Narrowed inside the promised window, not moved. |
| OpenAI committed capacity — "We expect to deploy the Committed Capacity in tranches during 2026 through 2028" | 424B4 "Business" / related-party section, company (written) | Tested by the kill criteria on the Working Capital Loan's current portion and the RPO. |
Departed check. No promiser has left. The EDGAR submissions index for CIK 2021728 holds three 8-Ks — Items 5.03/8.01/9.01 (2026-05-15), 2.02/9.01 (2026-06-23) and 2.02/9.01 (2026-08-12) — and no Item 5.02. Both promisers signed the Q2 10-Q on 2026-08-12. Current team made 8 of 8 scored rows. Context, not a flag: every executive officer adopted a Rule 10b5-1 sale plan in the IPO quarter — Feldman 1,500,000 shares (2026-05-20), Sean Lie (CTO) 1,872,000, Dhiraj Mallick (COO) 851,139 and Komin 128,000 (2026-06-30) — the week after the first guide for the last two (10Q-Q2, Item 5, "Rule 10b5-1 Trading Arrangements").
Dodge roll-up (detail in the transcript section below): 7 dodges in two calls (4 on 2026-06-23, 3 on 2026-08-12). Two topics were dodged in both quarters:
- Customer concentration — Barclays on the first call (who is above 10%), UBS on the second (share of FY2027 revenue from OpenAI + AWS). Management has never guided a concentration figure, so this is not a dodge of its own guidance; the 10-Q answers it after the fact (Customer A 34%, B 32%, C 10% of Q2 revenue; A and D are related parties — 10Q-Q2, Note 3 "Significant Customers").
- AWS deal size and economics — Needham on the first call (MW scale), TD Cowen on the second (deal economics). Management did guide AWS timing ("impact in 2027"), and kept to it, but has twice declined the size behind it — while the CEO's own Q2 remarks note the $25bn RPO carries no AWS backlog.
Verdict — conservative, on a thin record. 8 scored rows: 4 beat, 4 met, 0 missed, 0 withdrawn. Every first guide was beaten or met and the full-year guide was raised on all three metrics after one quarter. Two caveats keep this provisional: the whole record is one guide cycle, the classic place for a newly listed company to set a low first bar, and it is scored on a metric set the company introduced the same day. Re-test on the Q3 release (November 2026) and the FY2026 release against the $855–865m first guide. No valuation haircut follows from this verdict.
S2 — capital allocation
Five years do not exist. Audited years are FY2024 and FY2025 (424B4 consolidated statements of cash flows); H1 2026 is from EX99-Q2's cash-flow statement. $m.
| FY2024 | FY2025 | H1 2026 | |
|---|---|---|---|
| Operating cash flow | 452.0 | (10.1) | (47.5) |
| Capex (purchases of property and equipment) | (23.4) | (382.7) | (548.9) |
| FCF (derived) | 428.5 | (392.8) | (596.4) |
| Acquisitions (cash) | — | — | — |
| Share repurchase / tender offer | — | (21.4) repurchase + (49.3) option settlement "in connection with the Tender Offer" | — |
| RSU tax withholding (net settlement) | — | (3.0) | (416.7) |
| Dividends | — | — | — |
| Equity raised | 85.0 preferred | 1,100.0 preferred (Series G) | 1,014.2 Series H + 6,232.5 IPO (net of underwriting) |
| Debt raised | — | — | 1,004.6 OpenAI Working Capital Loan (repayable in compute); $850m revolver undrawn (CALL-Q2, Komin) |
| Stock-based compensation | 58.6 | 49.8 | 386.6 |
| Customer warrant assets, period end (non-cash, equity paid to customers) | — | 152.4 | 1,128.4 |
Locators: 424B4 "Consolidated Statements of Cash Flows" (FY24/FY25) and balance sheet; EX99-Q2 "Condensed Consolidated Statements of Cash Flows" and balance sheet (current + non-current customer warrants: 60.9 + 91.4 at 2025-12-31, 167.8 + 960.6 at 2026-06-30). XBRL agrees for H1 2026 (PaymentsToAcquirePropertyPlantAndEquipment 548,873k; ShareBasedCompensation 386,601k; PaymentsRelatedToTaxWithholdingForShareBasedCompensation 416,660k).
Where the cash went. Nothing was spent on acquisitions — no Goodwill, GoodwillImpairmentLoss or PaymentsToAcquireBusinessesNetOfCashAcquired fact exists and the prospectus names none — so there is no goodwill to impair. FY2024's positive cash flow was customer money, not profit: a $640.3m increase in customer deposits (424B4 MD&A "Operating Activities"). Since then the cash has gone into the company's own cloud: capex of $382.7m in FY2025 and $548.9m in six months of 2026, property and equipment net $43.2m → $437.4m → $986.8m (424B4 balance sheet; EX99-Q2 balance sheet). The CFO says a "meaningful portion of the remaining CapEx for data center fit-out" is reimbursed by the largest customer as pass-through (CALL-Q2, Komin; exact span: "we are reimbursed for a meaningful portion of the remaining CapEx for data center fit-out as data center pass-through cost reimbursement from our largest customer").
Buybacks, and at what price. There is no buyback programme and none is authorised (S5, dossier-2026-09-21). The only repurchases are pre-IPO employee liquidity: the December 2025 tender bought 2,156,765 Class B shares at $36.2324 ($78.1m gross; 424B4 "Tender Offer"), against a $185.00 IPO price and a $205.24 close on 2026-09-23 (market data get_latest_stock_price, NASDAQ:CBRS). The IPO net settlement is the other one: $416.7m of withholding at the $185.00 price is equivalent to ~2.25m shares retired (derived) — roughly 8% of the raise paying the tax on the company's own IPO compensation event.
SBC against dilution. SBC was $386.6m in H1 2026 (104% of GAAP revenue (derived)), driven by the IPO liquidity condition on 2016-Plan RSUs. Nothing offsets it: no buyback exists, and the share base will grow by the 36.1m customer-warrant shares and the 42.65m-share 2026 Plan reserve (dossier-2026-09-21 S2/S5). What the top pay is on: the CEO's 5,700,000 and CTO's 3,300,000 PRSUs vest on market capitalisation — "The PRSUs vest in three equal tranches based on the achievement of market capitalization thresholds of $75 billion, $150 billion and $250 billion, respectively." (424B4, notes to the financial statements, stock-based compensation) — measured on the Class A share count only (same note). Pay on market cap rather than on EPS; no buyback, so no EPS-plus-buyback flag.
Returns. Negative on every endpoint, so the incremental return is n.m.:
| FY2024 | FY2025 | H1 2026 | |
|---|---|---|---|
| GAAP operating loss (NOPAT proxy, no tax benefit) | (101.4) | (145.9) | (492.3) |
| Non-GAAP operating loss (company's pre-2026 definition; core from 2026) | (42.9) | (96.1) | (37.1) core |
| Operating invested capital proxy: PP&E + inventories + receivables, period end (derived) | 355.1 | 551.4 | 1,142.8 (2,271.2 incl. customer warrant assets) |
GAAP operating losses are the sums of the quarterly figures in the 424B4 "Quarterly Trends in Non-GAAP Financial Measures" table; the FY non-GAAP operating losses are the same table's totals ($42,874k for FY2024 matches the prospectus's own annual figure). Incremental FY2024 → FY2025: ΔNOPAT −$44.4m over ΔIC +$196.3m (derived) — a negative number that means only that the capital has not produced a profit yet. ROIC is negative in every period on any definition; for a company at this stage it is not informative, and the report says so rather than scoring it.
Verdict — neutral. No acquisitions, no goodwill, no buyback, no dividend: the capital raised (~$9.4bn of equity and customer debt since 2024) went into the cloud capacity that the $25.4bn RPO is supposed to fill, and into a large tax bill on the IPO. It cannot be called compounds while every return is negative, and nothing has yet been wasted to call it destroys. What bears watching is the currency used to buy customers: $1.13bn of customer-warrant assets on the balance sheet at 2026-06-30, to be charged against revenue "through October 2031" (10Q-Q2 Note 3, exact span: "The remaining balance of the customer warrant assets as of June 30, 2026 will be recorded as a reduction in revenue in the consolidated statement of operations through October 2031."). No ⚠️ destroyer line and no new destroyer kill criterion.
Earnings quality
Clean it is not, but most of the noise is the IPO. Checked in the skill's order.
- The non-GAAP bridge — two add-backs in 2025, five in 2026. Flag. The prospectus's non-GAAP net loss removed SBC and the forward-contract fair-value change only (424B4 MD&A "Non-GAAP Net Loss": FY2025 +$49.8m SBC, −$363.3m forward contract; FY2024 +$58.6m, +$401.3m). The Q1 2026 release introduced "core" metrics that also remove pass-through revenue and cost and add back customer-warrant amortisation; Q2 added a fifth, employer payroll tax on IPO SBC — $22.8m of cash added back to core operating loss (EX99-Q2, "Reconciliation of GAAP Loss from Operations to Core Operating Loss"; the release: "Employer payroll tax related to stock-based compensation from IPO is presented as an adjustment in calculating core net loss."). The customer-warrant add-back went $2.1m → $44.3m in one quarter and was 21% of Q2 core revenue ($44.3m / $209.9m (derived)); it is recurring by construction, amortised in proportion to the revenue it relates to until 2031. Equibles
GetNonGaapBridgereturns the same lines for both quarters (Q2 10-Q nets pass-through revenue and cost into one −$428k line). - SBC — flag. SBC / revenue: FY2024 20.2%, FY2025 9.8%, Q1 2026 5.0%, Q2 2026 209%, H1 2026 104% (derived; SBC from the cash-flow statements, revenue GAAP). FCF is negative in FY2025 and H1 2026, so SBC against FCF is n.m.; FCF after SBC is below zero in both.
- Cash conversion — n.m., flag on FY2025. OCF / net income: FY2024 +$452.0m / −$481.6m; FY2025 −$10.1m / +$237.8m; H1 2026 −$47.5m / −$464.5m — the sign flips every period because of the forward-contract swing ($401.3m loss, then $363.3m gain) and the IPO SBC. Accruals ((NI − OCF) / average total assets, derived): FY2025 +14.4% (NI inflated by the $363.3m non-cash gain), H1 2026 −6.0% (half-year, not annualised). FY2024 not computed (FY2023 total assets not read).
- Working capital — watch, not yet a flag. DSO (derived, GAAP revenue): FY2024 173 days, FY2025 36, Q1 2026 29, Q2 2026 62 — up one period while GAAP revenue fell; Customer B is 57% of receivables at 2026-06-30 (10Q-Q2 Note 3). Inventory $174.5m → $63.6m → $89.0m → $32.7m (424B4; XBRL
InventoryNet) as systems move into the company's own cloud (FY2025 transfer to property and equipment out of inventories $67.3m, 424B4 cash-flow supplemental). A second quarter of rising DSO would make it a flag. - Costs moved onto the balance sheet — flag. No capitalised software tag. The asset that grew is the customer-warrant asset: $152.4m → $1,128.4m in six months (7.4x) while revenue grew ~1.8x year on year; its amortisation runs through revenue, not expense. H1 2026 also carries $38.6m of non-cash interest expense (EX99-Q2 cash-flow statement), consistent with the imputed cost of the OpenAI loan [background inference; the note was not read this run].
- Customer concentration inside the bridge. Two of the five add-backs belong to the largest customers: pass-through is data-centre cost the OpenAI contract lets the company pass through at a markup (CALL-Q1, Komin), and the warrant add-back is the OpenAI and AWS warrants. Q2 revenue was Customer A 34%, Customer B 32%, Customer C 10% (10Q-Q2 Note 3; "Customer A and Customer D are considered related parties with respect to each other as defined by ASC 850, Related Party Disclosures."). The "core" framework therefore presents the top line before the price concessions made to win the customers that make it.
Result: EQ 4 flags — bridge grew from 2 to 5 add-backs incl. $22.8m cash payroll tax, and the warrant add-back is 21% of core revenue; SBC 104% of H1 revenue with FCF negative; FY2025 accruals 14.4% of assets on a non-cash gain; customer-warrant asset 7.4x in six months. Watch: DSO 29 → 62 days in one quarter. Most of the size is the IPO; the structure (warrant add-back, pass-through) is permanent.
Transcripts — Q1 2026 (2026-06-23, first read) and Q2 2026 (2026-08-12, prepared remarks and turns 12–31 first read)
dossier-2026-09-21.md read the Q2 call from Equibles, turns 2–11 only. This run read both market data files in full, in the §3.1 order. The safe-harbour paragraphs are skipped.
Guidance changes (CFO sections)
- Q1 call — the framework and the first guide. Komin introduced the core metrics ("So we're adjusting our GAAP numbers to exclude the impact of these items and a few other common ones like stock-based compensation and onetime items") and flagged that pass-through can vanish because "OpenAI has the option to choose whether to receive its future committed amounts in our cloud or in its own data centers". Guided Q2 and FY2026 (S8 rows 1–3 and the open table); hardware revenue down, cloud margin down 10–15 points on rented-back systems, hardware margin to the low 30s, G&A up. Long-term target restated as ~60% gross, ~40% operating margin. Liquidity $3.3bn before the IPO; $850m revolver added in April.
- Q2 call — raised. Q3 guided below the FY run-rate on margin (38–40% GM; Q3 the low point); FY2026 raised on all three metrics; FY2027 "more than triple". Capex framed as lower per MW than peers because the company deploys its own hardware at bill-of-materials cost and is partly reimbursed by its largest customer. Manufacturing capacity >10x in 2026, "We're going to end the year well over 10x our manufacturing capacity." (Komin, Arcuri exchange). Cash >$8.6bn; the revolver unused.
What analysts asked, by topic
- Customer concentration and the customer list — Q1: O'Malley (Barclays), 10% customers; Malik (Citi), OpenAI's choice between cloud and on-prem. Q2: Arcuri (UBS), FY2027 share of OpenAI + AWS and talks with Google and Microsoft; Rakesh (Mizuho), neocloud pipeline.
- AWS — Q1: Arcuri, timing and supply; Bolton (Needham), deal size in MW. Q2: Buchalter (TD Cowen), how the economics work; Bleustein for O'Malley, hosting Trainium in Cerebras's cloud.
- Disaggregated inference and AMD — Q1: Bryson (Wedbush), value split with Trainium. Q2: Buchalter, AMD go-to-market and AMD's acquisition; Bleustein, revenue share; Bolton, extra revenue inside the OpenAI contract and whether 20x throughput removes the need; Moore (Morgan Stanley), commercial readiness and NVIDIA installed base.
- Capacity and the guide's shape — Q1: Moore, the binding constraint (answer: data centres); Moore, margin dynamics; Buchalter, pricing vs OpenAI in the second-half ramp; Bryson, whether Q2 is the margin trough; Rakesh, the monthly MW ramp. Q2: Arcuri, 2027 manufacturing capacity; Rakesh, whether 600 MW accelerates 2027.
- TAM and large models — Q1: O'Malley, Bolton (token premium), Buchalter (the Kimi demo specs), Shannon (Craig-Hallum) on GPT-5.4 and Trainium sold externally.
Dodges
| Call | Topic | Analyst | Question in one line | What the answer addressed instead | Locator |
|---|---|---|---|---|---|
| Q1 2026 | Customer concentration | Tom O'Malley (Barclays) | Any 10% customers, and how large? | Declined and pointed to the filings: "I don't think we should share on the call. I think you'll see in the filings." | CALL-Q1 2026-06-23, Feldman, O'Malley second question |
| Q1 2026 | AWS deal size | Quinn Bolton (Needham) | Is AWS tens of MW, hundreds, a gigawatt? | Declined: "I don't think we're going to -- we're sharing that at this time." | CALL-Q1, Feldman, Bolton second question |
| Q1 2026 | Rent-back share of the FY guide | Atif Malik (Citi), asked of the CFO | How much of the FY outlook is new data centres vs renting back from G42? | The CEO took it and described the data-centre search: "And so I think the answer is all of the above." No split given | CALL-Q1, Feldman, Malik follow-up |
| Q1 2026 | Value capture in disaggregation | Matt Bryson (Wedbush) | What share of a Trainium + CS-3 solution's value is Cerebras's? | Explained when disaggregation works: "Not exactly. Let me share maybe a different crack at the problem." | CALL-Q1, Feldman, Bryson first question |
| Q2 2026 | Customer concentration | Timothy Arcuri (UBS) | Will OpenAI + AWS be ~2/3 of FY2027 revenue? | A history of past concentration worries, then "But I think you can expect for next year them to still be a meaningful portion of our revenue." No share given | CALL-Q2 2026-08-12, Feldman, Arcuri first question |
| Q2 2026 | AWS economics | Joshua Buchalter (TD Cowen) | How will the economics of the Amazon deal work? | Deployment mechanics and timing: "It will be delivered through Amazon's API service, Bedrock. And we are in the process right now of organizing deployments." | CALL-Q2, Feldman, Buchalter first question |
| Q2 2026 | OpenAI contract economics | Quinn Bolton (Needham) | Do AMD racks serving OpenAI add revenue under the contract? | Throughput economics in general: "Now I don't want to go into the specifics of our relationship with OpenAI." | CALL-Q2, Feldman, Bolton first question |
Not counted: Buchalter's Q1 question on pricing vs OpenAI in the second-half ramp got half an answer ("A lot of the improvement is going to come from OpenAI being deployed in our cloud, and it's back-end loaded.", Komin) — the OpenAI half, not the pricing half. Rakesh's Q1 "50 MW a month" premise was disputed rather than dodged.
Full notes, 21 Sep 2026 12 min read
CBRS — dossier, 2026-09-21
Slot names-ai-capex-3. Model-drafted unattended by the cloud research routine; nothing here is human-verified. Sources and locators are in relationships.yaml; figures below carry theirs inline. Anything from background knowledge is marked [background].
Price reference: $208.315 close, 2026-09-21 (market data get_latest_stock_price NASDAQ:CBRS; volume 5,000,300 shares, so roughly $1.04bn traded that day). IPO priced at $185.00 on 2026-05-14 and closed 2026-05-15 with the over-allotment exercised in full (34,500,000 Class A shares); the stock is +12.6% on its issue price four months later.
Shares at 2026-08-05 (10-Q cover): 112,247,109 Class A + 111,601,424 Class B + 13,715,508 Class N = 237,564,041, so ~$49.5bn of market capitalisation. Add the two customer warrants (33,445,026 Class N at $0.00001, 2,696,678 at $100.00) and it is ~273.7m shares, ~$57.0bn.
This name is a recent IPO under the skill's §3 rule, so the prospectus risk factors and use of proceeds replace the 10-K read. There is no 10-K, no proxy and no DEF 14A to skim: the entire EDGAR history is two S-1s, one 424B4, two 10-Qs and three 8-Ks.
S1 — Source
watchlist (owner, Apple Stocks, 2026-09-20), cluster ai_semis, coverage sec_domestic, cik 2021728, carrying the watchlist note "recent IPO; S-1 is the primer". No Tier 0 promotion.
S2 — Kill test
Annual figures as the company states them in the prospectus MD&A overview; quarterly from the XBRL company facts (data.sec.gov/api/xbrl/companyfacts, CIK 0002021728).
| FY | Revenue $m | Gross margin | GAAP net income / (loss) $m | Non-GAAP net loss $m |
|---|---|---|---|---|
| 2022 | 24.6 | 12% | — | — |
| 2023 | 78.7 | 33% | — | — |
| 2024 | 290.3 | 42% | (481.6) | (21.8) |
| 2025 | 510.0 | 39% | 237.8 | (75.7) |
The FY2025 GAAP profit is not earnings power. The company's own reconciliation removes stock-based compensation and the "change in fair value (extinguishment) of forward contract liability" to get to a non-GAAP net loss of $75.7m. Cerebras has never made money on operations.
Quarterly (10-Q XBRL):
| Quarter | Revenue $m | Gross profit $m | GM % | R&D $m | Op income $m | Net income $m | OCF $m |
|---|---|---|---|---|---|---|---|
| 2025 Q1 | 99.5 | 41.6 | 41.8 | 52.8 | (28.5) | (23.9) | (54.9) |
| 2025 Q2 | 103.3 | 32.1 | 31.1 | 60.8 | (57.2) | 309.5 | (68.9) |
| 2026 Q1 | 193.4 | 86.2 | 44.6 | 75.5 | (15.0) | (14.0) | 12.3 |
| 2026 Q2 | 180.1 | 25.6 | 14.2 | 320.2 | (477.2) | (450.5) | (59.8) |
Q2 2026 looks like a catastrophe and mostly is not. Stock-based compensation in the quarter was $377.0m ($15.4m in cost of revenue, $222.1m in R&D, $71.1m in sales and marketing, $68.5m in G&A) — the IPO triggered the liquidity condition on RSUs granted under the 2016 Plan. A further $44.3m was charged against revenue as amortisation of the OpenAI customer warrant asset. Adding both back: gross revenue ~$224.4m, cost of revenue ex-SBC ~$139.2m, gross margin ~38% — in line with FY2025's 39% — and an operating loss nearer ~$100m than $477m. These adjustments are derived here, not reported by the company; the company gives no non-GAAP figures in the 10-Q.
The real deterioration is sequential revenue: $193.4m → $180.1m, down 7% quarter on quarter, in a company the market prices for triple-digit growth. Part of that is the warrant contra-revenue starting to bite ($2.0m in Q1, $44.3m in Q2). Gross of it, revenue rose ~15% sequentially.
Balance sheet at 2026-06-30: cash $6,742.2m plus available-for-sale securities $1,179.4m = $7.92bn, against the OpenAI Working Capital Loan of $736.0m current and $182.2m long-term. Net cash ~$7.0bn. Total assets $11.63bn; stockholders' equity swung from −$578.7m at 2025-12-31 to +$9.15bn after the IPO and the conversion of preferred into 124,652,775 Class B shares. Equibles GetGoingConcernStatus: no flag in either examined 10-Q.
Cash runway: not the binding constraint. H1 2026 operating cash flow was −$47.5m and the company holds $7.9bn. What the $7.9bn is for is the constraint — this is now a data-centre business, and the prospectus says the cloud offering requires "significant data center capacity and capital investments for which we expect to require significant additional capital".
Share count: no five-year history exists — the company has been public for four months. Diluted weighted-average shares were 150,968,000 for Q2 2026. The dilution that matters is ahead: 36.1m warrant shares to two customers, a 42,650,268-share 2026 Plan reserve, and a 3,554,189-share ESPP that grows 1% a year.
Analyst coverage: at least five firms asked questions on the first call — UBS, TD Cowen, Barclays and two others in the turns read. No count obtained.
Verdict — survives on the balance sheet, not on the business. $7.9bn of cash against a ~$100m quarterly cash operating loss is years of runway, and there is a $25.4bn contracted backlog behind it. It does not die of funding. It dies, if it dies, of one customer.
S3 — Filings, gate-minimum (recent-IPO variant)
Read this run: the 424B4 prospectus (risk factor summary, the customer-concentration, internal- control and export-control risk factors, use of proceeds, MD&A overview, cover-page capital structure), the Q2 2026 10-Q (revenue note, Working Capital Loan note, warrants note, stock-based compensation note, balance sheet), and the Q2 2026 earnings call (turns 2–11 of 31). Not read: the September 2024 S-1, the Q1 2026 10-Q in full, the prepared remarks of the call (the transcript returns the CEO's opening as a single short turn and then jumps to Q&A — the guidance discussed below is the analyst's restatement of remarks this run did not see).
The three biggest risks management itself discloses
- The revenue is three counterparties, and one of them is also the lender. From the risk factors: "Group 42 Holding Ltd (together with its affiliates, "G42") accounted for 24.0% and 85.0% of our total revenue for the years ended December 31, 2025 and 2024, respectively" and "in the year ended December 31, 2025, Mohamed bin Zayed University of Artificial Intelligence ("MBZUAI") accounted for 62.0% of our total revenue." Those two are 86% of FY2025 revenue and the company states "G42 and MBZUAI are considered related parties with respect to each other" — so it is one Abu Dhabi relationship, not two customers. Receivables are worse: MBZUAI was 77.9% of the balance at 2025-12-31, G42 91.0% at 2024-12-31. Going forward the concentration transfers rather than dissolves: the December 2025 OpenAI MRA "represents a substantial portion of our projected revenues over the next several years".
This independently confirms, from the primary document, the third-party teardown figures the cluster pass recorded as search-summary evidence (_clusters/cluster-ai-capex-2026-09-21.md, §S6 row for CBRS: "MBZUAI 62.0% + G42 24.0% = 86% of 2025 revenue"). Those numbers are right.
- Material weaknesses in internal control over financial reporting, unremediated. "we identified certain material weaknesses in our internal control over financial reporting, including most recently for the years ended December 31, 2025 and 2024." They relate to "inadequate or missing resources who possess an appropriate level of expertise to timely review account reconciliations and identify, select, and apply U.S. generally accepted accounting principles ... including revenue recognition, inventory management and costing, data center assets accounting, and equity administration". Every one of those four areas is load-bearing for the numbers in S2 above, and the company is an emerging growth company taking the reduced reporting accommodations.
- Export controls on a business whose historical customers are in the UAE. The offerings are subject to US export controls and "may be exported outside the United States only with the required export license or through a license exception". The prospectus records licences held for export to G42 and MBZUAI with conditions attached. A significant majority of FY2025 revenue came from customers headquartered in the United Arab Emirates, and the risk factors separately name the war in the Middle East, "where our strategic partners G42 and MBZUAI are headquartered".
Use of proceeds
Net proceeds ~$5,408.5m at $185.00 (up to $6,221.4m with the over-allotment, which was taken). Purpose: "general corporate purposes, including working capital, operating expenses and capital expenditures" — no debt to repay, no acquisition named. Of that, ~$416.1m goes straight back out as tax withholding on ~4,799,565 shares of RSUs vesting on the IPO. So roughly 8% of the raise funds the tax bill on its own compensation event.
Latest quarter
Covered in S2. The one forward number obtainable this run came from the Q&A, where a UBS analyst restated management's guide — "you did say that revenue would be up more than 3x next year" — and the CEO did not dispute it, adding that capacity will "end the year well over 10x our manufacturing capacity, and we already have contracted facilities three or four times more for growth in 2027". Treat the 3x as thin: it is an analyst's paraphrase of prepared remarks this run did not read.
Insider pattern
3spread: 173 Form 4 filings since 2025-09-21, essentially all post-IPO, and 11 Form 144 proposed sales since 2026-03-25, dated from 2026-08-21 and 2026-09-04. The IPO closed 2026-05-15; a customary 90-day lock-up would have expired around 2026-08-13 [background], and the first Form 144 lands eight days later. Insiders began filing to sell at the first legal opportunity. Amounts and sellers were not itemised this run.
Red-flag sweep
- Going concern: no flag (Equibles, 10-Q filed 2026-08-12).
- Auditor change / restatement: no NT 10-K, no Item 4.01 8-K in an EDGAR history four months long. The
efts.sec.govfull-text sweep was not run (time). - Executive changes: not checked (Equibles budget reserved for ASML).
- Customer concentration: read from the prospectus directly rather than from the Equibles tool, because the company has no 10-K for the tool to tag.
- Material weaknesses: flagged above. This is the red flag; it is disclosed, not hidden, and it is unremediated.
S5 — Ownership
13F, 2026-06-30 — the first quarter in which anyone reported CBRS at all. 297 filers, 67.2m shares. Every line is a new position by construction.
| Manager | Shares | $m |
|---|---|---|
| Eclipse Operations, LLC | 11,446,270 | 2,529.6 |
| Jane Street Group | 8,111,322 | 1,792.6 |
| Coatue Management | 7,011,028 | 1,549.4 |
| Susquehanna International | 4,698,018 | 1,038.3 |
| Atreides Management | 3,810,086 | 842.0 |
| Citadel Advisors | 3,114,670 | 688.3 |
| Tiger Global Management | 2,999,000 | 662.8 |
| BlackRock | 2,075,168 | 458.6 |
| T. Rowe Price | 2,044,623 | 451.9 |
Two market-makers (Jane Street, Susquehanna) and a multi-strategy (Citadel) in the top six is what a newly listed, heavily optioned name looks like; it is inventory, not conviction. The conviction names are the pre-IPO venture holder (Eclipse) and the crossover funds (Coatue, Tiger, Atreides).
13D/G: six 13Gs filed 2026-08-14 for the 2026-06-30 period, plus a 13D/A filed 2026-08-18 by Eclipse Continuity GP I, LLC — the pre-IPO venture holder's control-group filing, not an activist position.
Insider ownership: no proxy exists yet, so no clean percentage. What the cover page gives instead is control: Class B carries 20 votes and "will represent approximately 99.2% of the voting power of our outstanding capital stock immediately following this offering." Public shareholders own economics, not votes.
Short interest (Equibles, FINRA): 8.14m (2026-06-15) → 13.53m → 10.42m → 12.90m → 15.84m (2026-08-14) → 13.33m (2026-08-31), days to cover 1.4–2.3. Against 112.2m Class A shares — and a tradeable float much smaller than that, since the IPO sold 34.5m — a low-double-digit-million short position is a large fraction of the free float, even though days-to-cover looks benign on the elevated post-IPO volume. The tool's 2026-09-15 estimate (~13.0m) is a model output, not FINRA data.
Shelf / ATM / buyback: none. No S-3 is on file (the company is not yet S-3 eligible [background]), no ATM, no repurchase authority, no dividend. The dilution here does not come from a shelf — it comes from 36.1m warrant shares owed to two customers and a 42.65m-share 2026 Plan reserve.
Who is on the other side: the pre-IPO venture and crossover holders who are now free to sell, and a short base sized against a thin float. The insiders' own Form 144s start eight days after lock-up expiry.
S8 — Valuation and sizing (proposed only; owner-only items OPEN)
At $208.315: market cap ~$49.5bn on 237.6m shares, ~$57.0bn fully diluted for the two customer warrants. Net cash ~$7.0bn → EV ~$42.5bn / ~$50.0bn diluted.
LTM revenue (FY2025 $510.0m − H1 2025 $202.8m + H1 2026 $373.5m) = $680.7m. EV/LTM revenue ≈ 62x basic, ≈ 73x diluted. There is no earnings multiple to compute.
The only forward anchor the filings give is the $25.4bn remaining performance obligation at 2026-06-30 — 37x LTM revenue — of which the OpenAI MRA is the bulk, and against which the unamortised customer-warrant asset (~$1.1bn already, up to ~$2.74bn at the OpenAI warrant's $82.02 grant value) will be charged as contra-revenue as it is recognised.
Out-year model. Revenue is built off the capacity commitments rather than a growth rate, because the growth rate is a backlog conversion question, not a demand question.
| FY2026E | FY2027E | FY2028E | |
|---|---|---|---|
| Revenue $m (net of warrant contra-revenue) | 900 | 2,900 | 5,800 |
| Gross margin | 35% | 38% | 40% |
| Operating margin | (10)% | 8% | 16% |
| Net income $m | (90) | 230 | 930 |
| Diluted shares m | 260 | 280 | 290 |
| EPS | (0.35) | $0.82 | $3.21 |
| Implied P/E at $208.32 | n/m | 254x | 65x |
| Case | Basis | Value |
|---|---|---|
| Bear — the OpenAI MRA is renegotiated, delayed or terminated. The Working Capital Loan accelerates in the same event, the backlog goes, and what remains is a ~$500m revenue business with $7bn of cash. | ~$8bn EV + net cash | ~$55/sh |
| Base — the table above, at 55x FY2028 earnings | $3.21 × 55 | ~$177/sh |
| Bull — AWS scales as the term sheet contemplates, concentration genuinely broadens, FY2028 revenue $8bn at 20% net margin | $5.50 × 55 | ~$300/sh |
The base case is below the traded price. That is the honest read of this model: at $208 the market is already paying for the bull path, and the bear case is not a slow derating but a step-change tied to one contract. The spread between $55 and $300 is the widest of any name in this slot and it is the right shape for the risk.
⏳ Buy price and position size are owner-only and stay OPEN (user). Liquidity is ample today (~$1.04bn traded on 2026-09-21) but that is post-IPO volume on a thin float and should not be assumed durable.
S9 — Monitoring
Next earnings: Q3 2026 results, expected in the second week of November 2026 — derived from the two data points that exist (Q1 10-Q filed 2026-06-24, Q2 10-Q and call 2026-08-12), not from a calendar feed.
Filing feed: the Tier 0 EDGAR sweep covers cik 2021728. For this name the sweep matters more than for most: with no 10-K and no proxy, 8-Ks are the disclosure channel, and the first S-1/S-3 or lock-up-related filing will arrive there.
The leading indicator: the remaining performance obligation, quarter on quarter. $25.0bn at 2026-03-31 → $25.4bn at 2026-06-30. This single number is the OpenAI relationship made visible. It is disclosed in the revenue note every quarter, it leads revenue by years, and it is the only figure that would show an MRA renegotiation before revenue did. Secondary: the customer-warrant asset amortisation charged against revenue (which tells you the rate at which capacity is actually being delivered) and the current portion of the Working Capital Loan.
Trigger that converts this to a buy: two consecutive quarters in which the RPO holds or grows and a customer other than OpenAI, G42 or MBZUAI is disclosed above 10% of revenue — i.e. the backlog is real and the concentration is actually breaking, rather than transferring from Abu Dhabi to San Francisco. AWS revenue appearing as a disclosed line is the specific form that takes. Until then this is a single-contract instrument priced as a platform. The price and the size remain the owner's.