2026-11-11 · 2026-11 / week-2
Rackspace's AI deals add demand, but the short fails 10/5
Rackspace's AI deals add demand, but the short fails 10/5
Classification: Reject / no trade
Summary: Rackspace has moved beyond an AI memorandum of understanding: it disclosed a multiyear Together AI infrastructure agreement, a separate Blacksmith capacity agreement, and a $66 million stock placement. The Together AI GPU deployment still needs separate financing, cash conversion remains thin, and the stock's current risk cannot be bounded tightly enough for a short.
Research time: 2026-11-11 21:15 Asia/Singapore (08:15 ET), pre-market. Model reference: Nov. 10 regular close. Scope: U.S.-listed short screens only.
Opportunity Ranking
| Rank | Candidate | Direction | Mispricing under review | Evidence freshness | Catalyst window | Positioning | 10/5 status | Tradeability | Main rejection risk |
|---|---|---|---|---|---|---|---|---|---|
| 1 | Rackspace Technology (RXT) | Short | Signed AI demand versus financing, deployment and cash conversion | Nov. 9 10-Q; Nov. 10 agreements and placement | Nov. 11 regular-session acceptance; 2027 deployments; next results | 24.51M shares short at Oct. 30 settlement; float percentages conflict by provider and predate 24M restricted shares | Reject | High turnover, but current borrow, spread, depth and exit capacity are unverified | New capital, signed agreements, $850M RPO and management's 12-month liquidity view |
| 2 | Lucid Group (LCID) | Short | Vehicle cash burn versus new PIF-backed liquidity | Nov. 9 Q3 release | Quarterly cash-burn and delivery updates; company says liquidity reaches Q4 2027 | 101.06M short, 66.32% of reported float at Oct. 30 | Reject | Liquid enough to research; locate, borrow and exit conditions are unknown | 30% sequential reduction in cash use, $1.75B additional facility and extreme squeeze exposure |
| 3 | D-Wave Quantum (QBTS) | Short | Small recognized revenue and lower Q3 bookings versus quantum optionality | Nov. 10 Q3 release | RPO conversion and technical milestones | 82.46M short, 22.58% of reported float at Oct. 30 per MarketBeat | Reject | High reported volume, but borrow and current depth are unknown | $503.4M cash and securities, $38.3M RPO and analysts' targets far above the quote |
| 4 | SPDR S&P 500 ETF (SPY), control | Short | No issuer-specific disagreement found | Nov. 10 close | No single dated catalyst identified for this short screen | Broad-market instrument; not used as issuer positioning evidence | No candidate | Very liquid market reference; no isolated short thesis | A broad-market hedge cannot replace a company-specific mispricing |
Selected opportunity: RXT for a no-trade screen with the highest information value, not as the best executable short.
Why this one now: The evidence boundary changed since the Desk's May RXT note. There are now signed AI deployment agreements and new equity funding, but the company says the principal GPU financing is not yet committed. The fresh 10-Q also exposes the cash, capex and 2028 debt bridge.
What should surprise the reader: A signed contract is meaningful progress, yet the deployment can still fail its return test if financing, power, equipment, delivery or customer demand falls short.
Why This Can Move More Than 5% Soon
RXT's Nov. 10 session ranged from $2.52 to $3.22 and closed at $2.68, down 1.47%, on 13.14 million shares. Before the Nov. 11 open, Finviz showed $2.81 at 08:02 ET, up 4.69% from the prior close; MarketBeat showed $2.82 at 07:50 ET, and StockAnalysis showed $2.83 at 07:40 ET. These pre-market observations differ and are not averaged. They are reference marks only. No Nov. 11 regular-session acceptance had occurred by this article's cutoff. RXT price history, Finviz quote and short-interest page, MarketBeat RXT short-interest page
The Nov. 10 announcements and fresh financing create a live repricing path in either direction. A pre-market rise is evidence of volatility, not proof that the market has accepted the new contract economics.
10/5 Asymmetry Gate
The scenario map below is a 12-month sensitivity from the verified $2.68 regular close, not from a live short entry. The model uses a current-share estimate assembled from the latest filings, the Q3 debt and cash balances, and a secondary three-analyst target range. It is not an independent fair-value estimate.
| Measure | Result | Basis |
|---|---|---|
| Verified reference | $2.68 | Nov. 10 regular close, 16:00 ET |
| Favorable base-case decline | 0.0% | Base holds the current approximate EV / FY26 Adjusted EBITDA multiple |
| Credible adverse stress | +97.8% underlying rise | $5.30 high of the three-analyst, 12-month target range |
| Top-to-bottom gross reward / adverse risk | 0.74:1 | $0.74 top versus $5.30 bottom, before costs |
| Subjective probability-weighted gross short return | -6.3% | Price-only, before borrow, fees, slippage and carrying costs |
| Status | Reject / no trade | Base fails the 10% decline hurdle; adverse stress exceeds 5%; ratio is below 2:1 |
The $0.74 top is a mechanical 10x sensitivity on the $290 million midpoint of FY26 Adjusted EBITDA guidance, using only Sept. 30 cash and assuming the $66 million placement proceeds have been spent. If all proceeds remain as cash, the same multiple produces about $0.97. Neither is a floor or fair value. The $5.30 bottom is an outside analyst stress, not a price ceiling. The 25% / 50% / 25% weights are Desk judgments, not observed frequencies or analyst probabilities.
The Setup
Rackspace operates Public Cloud and Private Cloud businesses. Its new AI agreements are real contractual steps, but the capital plan is incomplete. The short question is whether the common equity price gives enough compensation for that execution and debt risk after the stock has already fallen and short interest has eased.
The prior May 29 Desk note used a $4.85 reference and focused on the then non-binding AMD MOU. It is retired, not carried forward. During that note's six-week window, RXT closed at $7.53 on June 17 with an $8.60 intraday high, crossing its stated $6 invalidation. No order or execution record exists, so this is a price-path audit, not trading performance. The present screen tests new Q3, financing and deployment evidence. Prior RXT note, RXT historical prices
The Market Price
The model reference is the Nov. 10 regular close at $2.68. The more recent $2.81 pre-market indication at 08:02 ET is recorded separately and cannot be treated as an executable entry. The current regular session had not opened.
The 10-Q reports 262.3 million common shares outstanding at Sept. 30. Rackspace then issued 3.8 million shares to Palantir on Oct. 1 and 23,999,997 shares in the Nov. 9 private placement. That yields an approximate 290.1 million shares before any later employee-plan or ATM issuance. At $2.68, that implies about $777 million of equity value. The share count is a derived estimate from rounded filing figures, not a transfer-agent confirmation. The company also has a $250 million ATM facility; it sold 2.8 million shares under it during Q3 for $11.1 million net of all offering costs. Future use is unknown. RXT Form 10-Q, RXT private-placement Form 8-K
At Sept. 30, total reported debt was $2.793 billion and cash was $108 million. The private placement adds $66 million cash if still unspent. On that assumption, simple enterprise value is about $3.40 billion, or 11.7x the $290 million midpoint of FY26 Adjusted EBITDA guidance. If all placement proceeds have been spent, it is about $3.46 billion, or 11.9x. This simple bridge excludes $500 million of operating and finance lease arrangements disclosed in the 10-Q, so it is not comparable with a lease-adjusted peer multiple. It also does not convert Adjusted EBITDA into cash available for debt service.
Three S&P Global analysts have a $4.00 to $5.30 12-month target range and a $4.77 mean. It is a small sample of external forecasts, not market-implied value or a valuation model. Every target is above the $2.68 close, which is an adverse counterparty signal for this short. RXT analyst forecast
The Mispricing
The potential disagreement is not “AI demand is fake.” Rackspace has signed multiyear arrangements with Together AI and Blacksmith. The gap is between those announcements and the conditions needed to turn them into cash after equipment, power, financing, operation and debt service.
The Together AI agreement calls for dedicated NVIDIA GB300 NVL72 infrastructure, with deployment planned for 2027 and subject to conditions. The company says it has no definitive financing commitment for the GPUs yet and is still speaking with lenders. The related joint go-to-market letter of intent is explicitly non-binding. Blacksmith's agreement covers about 2 MW of AMD-powered compute and storage, with deployment expected in early 2027. Together AI agreement, Blacksmith agreement
The Q3 operating picture is mixed. Revenue fell 9.3% year over year to $609 million. Public Cloud revenue fell 17% to $349 million; Private Cloud rose 4% to $260 million. Q3 operating cash flow was $36 million, and trailing-twelve-month operating cash flow was $69 million. Management raised its 2026 revenue outlook by $35 million at the midpoint and kept its Adjusted EBITDA range at $285 million to $295 million. Those are credible counterweights to the short case. Rackspace Q3 results
For the first nine months, operating cash flow was $9.3 million versus $91.7 million a year earlier, while net cash used in investing was $70.3 million. The company attributes most of the operating-cash decline to $74 million less net cash proceeds from accounts receivable sales, so it would be wrong to call the entire decline an underlying operating collapse. The company separately reports $167.2 million of total capital expenditure, up from $91.6 million, mainly from customer gear and capitalized software. That total is not the same line as cash purchases in the cash-flow statement; do not subtract both from CFO. RPO was $850.3 million, with about 11% expected to be recognized as revenue in the rest of 2026. RPO is not the same as cash collected. RXT Form 10-Q
The Positioning
The latest settlement data available by this cutoff are for Oct. 30: 24,509,918 shares short, down 5.92% from Oct. 15. MarketBeat reports 9.84% of float; Finviz uses a smaller 99.13 million-share float and reports 24.73%. These percentages cannot both describe the same denominator. The Nov. 9 placement issued 24 million restricted shares after the settlement date, adding to shares outstanding without immediately confirming a larger tradable float. Use the raw short count as the observed fact; treat float percentage as unknown within conflicting vendor definitions. Current borrow, locate, fee, recall terms and available-to-short supply were not verified. MarketBeat, Finviz, SEC Form 8-K
The Catalyst
The next tests are observable, but not all are dated. First, the next filed quarter should show whether Public Cloud declines stabilize and whether operating cash generation recovers after the receivables-sale timing effect. Second, the two AI deployments need financing, equipment, power and delivery before they can produce revenue. The company expects Blacksmith deployment in early 2027 and Together AI deployment during 2027. Third, the debt stack has material 2028 maturities. The company says it expects sufficient liquidity for at least 12 months at current operations, but makes no assurance that future operating cash or borrowing capacity will cover all debt and capital needs.
The cheapest falsification test is the first reported deployment and financing update: require funded GPU purchases, installed capacity, customer billing, cash collection and contribution after associated financing and operating costs. A contract headline alone does not pass this test.
The Payoff
| Scenario | Probability | Target / level | Gross short return from $2.68 | Horizon | Conditions | Evidence quality |
|---|---|---|---|---|---|---|
| Top case for the short | 25% | $0.74 | +72.4% | 12 months | The 10x guide-multiple sensitivity applies after the placement proceeds are used and cash conversion misses | Low; mechanical sensitivity, no peer multiple established |
| Base case | 50% | $2.68 | 0.0% | 12 months | The current approximate guide multiple holds as the contracts proceed but do not yet change the cash bridge | Low; price anchor, not a forecast |
| Bottom case for the short | 25% | $5.30 | -97.8% | 12 months | The highest S&P Global target in its three-analyst sample is reached as deployments advance and short covering adds demand | Medium-low; external target, not a cap |
| Invalidation | n/a | n/a | n/a | Immediate | GPU financing closes, both deployments enter service, customer cash receipts and contribution margins are disclosed, and debt falls without further material dilution | High if observed in filings |
Probability-weighted expected value: about -6.3% gross price-only return for the short before borrow, fees, slippage and carrying costs. The 25% / 50% / 25% weights are subjective Desk inputs, not observed frequencies.
Current market level and timestamp: $2.68, Nov. 10 regular close at 16:00 ET. The $2.81 pre-market indication at 08:02 ET Nov. 11 is separate and reference-only.
Primary instrument: Common stock, not proposed for trading.
10/5 favorable base move: 0.0%. Credible adverse stress: +97.8%. Measurement basis: Reference-only, not verified entry. 10/5 status: Reject. Confidence: Low.
The Kill Shot
The strongest counterparty case is substantial: two real agreements; a $66 million cash placement; $850.3 million of RPO; a positive $36 million Q3 operating cash flow; $69 million trailing-twelve-month operating cash flow; and management's view that existing liquidity covers at least twelve months at current operations. The share price is also below a three-analyst target range of $4.00 to $5.30. If the financing closes and deployments produce contracted revenue with cash margin, the short can lose before any 2028 refinancing issue matters.
The load-bearing assumption is that Rackspace must finance GPU capacity before the new contracts generate enough cash to support the capital stack. That can fail if customer payments, equipment finance and deployment timing align better than the short expects.
What Could Go Wrong
The $66 million placement reduces near-term liquidity pressure but dilutes common holders; it is not proof of distress. Management raised the revenue midpoint and maintained EBITDA guidance. The 2027 deployments could be financed on acceptable terms and create a valuable recurring business. The analyst target survey is small, but it points upward. The 2028 maturities may be refinanced well before they become urgent. RXT can rise on contract details, analyst changes or short covering, and gaps can exceed every scenario marker. A short's maximum loss is not capped at the target table.
What Would Prove This Wrong
This short screen fails if future filings show financed deployments, recognized AI revenue, collected customer cash, acceptable margins after equipment and interest costs, stable core-cloud demand, and falling net debt without repeated large share issuance. Keep this record closed as a Reject; those observations require a new valuation and fresh market data.
Risk Audit
- The Nov. 11 price indications were pre-market only. No regular-session price acceptance, spread, depth, venue quality or exit capacity was verified.
- Short interest is an Oct. 30 settlement measure. Float percentages differ by provider, and the 24 million restricted placement shares were issued after that settlement.
- The company retains a $250 million ATM facility. Additional use could add liquidity and dilute existing holders; future issuance is not known.
- Current locate, borrow rate, recall risk and carrying cost are unknown.
- The short has an asymmetric tail. New AI financing, early deployments, a customer expansion, improved public-cloud performance or a squeeze can produce a gap beyond the modeled $5.30 stress.
- The $0.74 and $5.30 targets are different kinds of reference: one is a sensitivity calculation, the other is an external analyst estimate. Neither is fair value or a risk limit.
Best Trade Strategy
Reject / no trade. No entry is defined. Keep entry price null and execution disabled. Do not use options, leverage, margin or an unborrowed short. Reconsider only after a regular-session price, financing terms, borrow availability, cash conversion and updated share count are verified and the scenario map is rebuilt.
Sources
| Source | Claim supported |
|---|---|
| Rackspace Q3 2026 Form 10-Q | Revenue, cash flows, RPO, capex, debt, liquidity, lease arrangements and share-count bridge |
| Rackspace Q3 results release | Segment revenue, Q3 / TTM operating cash flow and updated FY26 guidance |
| Rackspace private-placement Form 8-K | 23,999,997 shares issued for approximately $66 million |
| Together AI agreement | Multiyear infrastructure agreement, 2027 deployment plan, conditions and financing status |
| Blacksmith agreement | Approximately 2 MW of capacity and early-2027 deployment expectation |
| StockAnalysis RXT history | Nov. 10 regular close, range, volume and earlier pre-market observation |
| Finviz RXT page | Nov. 11 08:02 ET pre-market price and alternative float denominator |
| MarketBeat RXT short interest | Latest FINRA settlement count and alternative float percentage |
| StockAnalysis RXT analyst forecast | Three-analyst $4.00 / $4.77 / $5.30 12-month target range |
| Lucid Q3 2026 release | Revenue, cash-consumption change and additional PIF-backed liquidity |
| Lucid Nov. 10 history and short-interest data | Reference close and Oct. 30 reported short positioning |
| D-Wave Q3 2026 release | Revenue, bookings, RPO and cash/investments |
| D-Wave Nov. 10 quote and short-interest data | Reference close and Oct. 30 reported short positioning |
| SPY Nov. 10 close | Broad-market control price |
| FinanceCharts RXT history | June 17 path used in the prior-forecast audit |
Research Quality Scorecard
| Criterion | Score | Rationale |
|---|---|---|
| Market disagreement | 3 | The contract/funding gap is clear; the price's implied expectations are not directly observable |
| Evidence base | 4 | Fresh SEC filing and issuer agreements; current post-placement cash and exact share count are not yet reported |
| Positioning and flows | 3 | Raw short count is recent, but vendor float denominators conflict and borrow is unknown |
| Catalyst path | 4 | Financing, installation, revenue recognition and cash collection are observable, though mostly 2027-dated |
| Payoff architecture | 2 | Base decline is zero and adverse risk is far above 5% |
| Invalidation discipline | 4 | Contract funding, deployment, cash margin and debt reduction are monitorable |
| Differentiated insight | 4 | A signed customer contract still leaves a separate equipment-financing and cash-conversion bridge |
| Client value | 5 | The screen prevents a financing-dependent AI headline from being mistaken for a bounded short |
Total: 29 / 40. Reject classification follows the failed 10/5 economics, regardless of the score.
Bottom Line
Rackspace's new AI agreements make the short thesis harder, not easier. The 9M cash bridge and 2028 debt stack deserve scrutiny, but funding and demand are moving forward too. The 10/5 map shows no favorable base case and severe upside risk. Reject the short until new filings establish a cash return that the current share price does not reflect.
AI Illustration Prompt
Create an editorial image of a modern liquid-cooled GPU hall seen through two layers of glass. Above the floor, place signed Together AI and Blacksmith contract sheets beside clean server racks; below it, show a separate, still-open equipment-financing term sheet, a $66 million equity ledger, and a compressed 2028 debt calendar. The image should show that customer demand exists while the funding bridge is not complete. Use charcoal, brushed steel, restrained NVIDIA green, AMD red accents and warm paper. For an institutional finance publication, realistic and forensic, with no stock charts, flying money, cartoon bears or hype. Add a subtle readable watermark: “The Mispricing Desk.”