Riot Platforms, Inc.
RIOT is trying to re-rate from a volatile bitcoin miner into an AI data-center landlord, but at $23.75 the stock already capitalizes a lot of that future while current operations still lose money heavily and the key value inflection does not start paying rent until December 2027.
RIOT's problem is not the absence of a story. It has one of the better stories in crypto-adjacent infrastructure right now. The problem is the price already reflects a lot of that story before the economics arrive. At $23.75, investors are paying 12.23x FY2025 sales for a business that still posted -$1.95 of EPS, -62.0% operating margin, -86.9% net margin, and -$0.2B of free cash flow. That is hard to justify when the key AI lease does not start rent until December 2027.
The bullish case is real. Management landed a 191 MW lease with a leading frontier AI lab, worth $9.1B of contract revenue over 20 years and $7.3B-$8.2B of NOI at an 80%-90% margin. AMD is ramping, recurring lease revenue hit $4.9M in Q2, contracted data-center revenue is now $9.8B, and the engineering business has a $177.1M backlog at 27%+ margins. Those facts support the idea that RIOT can eventually become more digital infrastructure landlord than bitcoin miner. If that transition works, the stock may ultimately deserve a much higher quality multiple than commodity mining peers.
But the case depends on three things that are not yet delivered. First, execution has to improve materially, because the last three quarters missed EPS by -822.7%, -336.4%, and -124.2%. Second, the company has to bridge a long cash-flow gap without destroying value, since Q2 adjusted EBITDA was -$70M and GAAP net loss was $237M, even with $1.2B of liquidity and $573M of interim financing support. Third, Corsicana has to move from a nonbinding 1 GW LOI to a signed agreement before the market can underwrite the next leg of the AI thesis.
The stock is also not coming from a capitulation setup. It is only 21.7% below its 52-week high, up 25.59% in the last month, and sits 25.1% above its 200-day average. That is the wrong setup for paying forward for a still-unproven model transition. The AI pivot deserves attention and could work, but at this price the better stance is to avoid the shares until either fundamentals catch up or the valuation resets materially.
Rockdale hits schedule, AMD expands on time, Corsicana converts to a binding lease, and the market fully values RIOT as an AI infrastructure platform rather than a bitcoin miner.
AI contracts remain intact but near-term losses, execution misses, and the long delay to rent commencement keep valuation closer to the engine's sales-quality blend.
Mining losses persist, Rockdale timing slips beyond December 2027, and Corsicana's 1 GW LOI fails to convert, forcing investors back to valuing RIOT on a weak operating base.
Quant call (SELL) diverges from analyst consensus (Strong Buy: 11 buy / 0 hold / 0 sell) — resolve explicitly.
In the short run, the market is a voting machine, but in the long run, it is a weighing machine.
Forward earnings power is the entire debate here, because trailing earnings are not investable and the stock has no meaningful P/E against a peer average of 12.1x. If Rockdale and AMD ramp as planned, RIOT could move from negative EBITDA today to a much more stable infrastructure cash-flow model by 2028, but that path still has a long duration and requires timely delivery of 96 MW in late 2027 and full 191 MW by June 2028. The moat is improving now because approved power, land, and development capability matter more in AI infrastructure than in mining, but 3 to 5 years out that moat is only durable if Riot proves it can convert power access into repeat leases faster than better-capitalized digital infrastructure peers. AI is a real secular tailwind here, while crypto regulation and mining economics become less central over time, but the market is already discounting a large share of that transition before the earnings power arrives.
- Valuation is stretched at 12.23x P/S versus 3.4x peers, while the quant engine's $12.25 target implies -48.4% downside.
- Core profitability deteriorated sharply, with FY2025 revenue up 63.8% to $0.6B but EPS at -$1.95, operating margin at -62.0%, and net margin at -86.9%.
- The AI pivot is real but back-end loaded: the new Rockdale lease covers 191 MW, yet first rent on the initial 96 MW does not begin until December 2027.
- Riot signed a $9.1B initial 20-year AI lab lease with expected NOI of $7.3B-$8.2B and NOI margin of 80%-90%, a far better long-term economic model than pure bitcoin mining.
- Contracted data-center revenue reached $9.8B across AMD and the new AI lab lease, giving unusually strong revenue visibility for a company that historically depended on commodity-like mining economics.
- Liquidity was $1.2B at Q2 2026 and Morgan Stanley provided $573M of interim financing, which materially reduces near-term funding risk for early development.
- Engineering backlog rose to $177.1M, with margins above 27% and data-center exposure around 90%, showing the services business is becoming more relevant and more profitable.
- Recent execution is poor: RIOT missed EPS by -822.7%, -336.4%, and -124.2% in its last three reported quarters, which cuts against taking long-dated guidance at face value.
- Current operations are deeply loss-making, with Q2 2026 EBITDA of -$0.14B and -78.5% margin, plus a GAAP net loss of $237M.
- The biggest incremental upside asset after Rockdale is still uncommitted, because Corsicana's full 1 GW opportunity remains only a nonbinding LOI with timing and terms uncertain.
- The stock setup is not washed out: shares are only 21.7% below the 52-week high after a 25.59% one-month rally, despite fundamentals that have worsened through 2026.
| Metric | Value | Context |
|---|---|---|
| Current price | $23.75 | Reference point for target and scenario analysis |
| Price target | $12.25 | Ground-truth target, -48.4% vs current, Low confidence, sales_quality_blend |
| FY2025 revenue | $0.6B | 63.8% YoY growth, but quality of earnings remains weak |
| FY2025 EPS | -$1.95 | -315.3% YoY, no meaningful P/E |
| FY2025 operating margin | -62.0% | Collapsed from 40.8% in the prior fiscal year |
| FY2025 net margin | -86.9% | Shows severe earnings dilution from the transition phase |
| Valuation | 12.23x P/S | Versus peer average 3.4x |
| Balance sheet | 0.13 debt/equity, 1.56 current ratio, $0.3B total debt | Leverage is manageable, liquidity not the immediate issue |
| Free cash flow | -$0.2B | Development spend is outrunning internal cash generation |
| Q2 2026 EBITDA | -$0.14B | -78.5% margin, down 5,140 bps QoQ and 6,940 bps YoY |
| AI and data-center contracts | $9.8B contracted revenue | Includes the new $9.1B AI lab lease |
| Price action | -21.7% vs 52-week high, +25.1% vs 200-day average | Not a washed-out setup despite weak earnings delivery |
The biggest risk to the bearish call is successful execution of the AI pivot, because a $9.1B lease with 80%-90% NOI margins can justify much higher value if Riot delivers the first 96 MW on time by December 2027 and converts Corsicana's 1 GW LOI into a binding contract.
The next catalyst is October 29, 2026 earnings, especially any update on AI project financing, AMD expansion timing, and whether Corsicana's 1 GW LOI converts into a signed lease.
Ownership is concentrated enough to matter, with the top three institutions holding 49.5%, while insider ownership is only 3.9%. Short interest is elevated at 14.87%, or 51.9M shares and about 3 days-to-cover, which can fuel squeezes around AI headlines, and the positive near-dated options gamma regime helps explain why the stock can run hard on incremental news. That positioning can support tactical upside, but it does not fix the mismatch between current losses and current valuation.
Still the economic base today, but profitability has worsened sharply as the company funds the data-center transition and reduced bitcoin holdings.
Recurring lease revenue rose 400%+ QoQ as AMD's first 25 MW came online, and this segment now carries the strategic valuation case through the $9.8B contracted pipeline.
Margins expanded to above 27%, with roughly 90% data-center exposure, making it a useful internal capability and a growing earnings support leg.
Rockdale and Corsicana in Texas are the critical assets, with the investment case tied to domestic power availability and permitting rather than international expansion.
Riot's key takeaway is a $9.1B AI lab lease that accelerates its data-center pivot.
Execution has been uneven at best. On the 2024 Q3 call, management cut FY2025 exit hash-rate guidance to 46.7 EH/s from 56 EH/s and FY2024 to 35 EH/s from 36 EH/s, citing delays, while guiding capex of about $663M through FY2025 and a 1 GW Corsicana site by 2026. Since then, the strategy has clearly shifted from mining growth to AI and data centers, which may prove smart, but the beat-miss record has worsened: after beats in late 2025, the last three quarters all missed badly at -822.7%, -336.4%, and -124.2% versus street EPS. Current guidance is more ambitious in dollar terms than anything management laid out two years ago, including $9.1B of AI lab contract revenue and 80%-90% NOI margins, so credibility now rests less on narrative and more on hitting concrete milestones like AMD's 10 MW in November 2026, the remaining 15 MW in May 2027, and first rent on 96 MW in December 2027.
The lease we are announcing today is contracted directly with one of the world's leading frontier AI labs for 191 megawatts of critical IT capacity at our Rockdale campus and marks the beginning of what we expect to be a long and significant partnership.
We are currently targeting delivery for the first 96 megawatts with rent commencing in December 2027.
For the second quarter of 2026, Riot reported total revenue of $174 million, a 14% increase year-over-year, driven by growth across our Data Center and Engineering segments.
With that capacity delivered in May, the financing process has advanced to late-stage discussions with multiple banks, and we expect to close before the end of the third quarter.
While an LOI is not a signed lease, this milestone reflects the depth of demand for large-scale fully approved power in today's market and the confidence prospective tenants have in Riot's ability to deliver.
Riot's edge is no longer just mining scale. It is control of large, already-developed power sites in Texas, especially Rockdale and the 1 GW Corsicana campus, plus an engineering arm that can support build-out and raise switching costs for tenants. That moat is real today because frontier AI demand is chasing scarce powered land, but its durability is still unproven because hyperscale and private infrastructure competitors can also bring capital and relationships if Riot misses delivery timelines.
Capital allocation has shifted from maximizing bitcoin exposure to funding AI infrastructure. That is strategically sensible, but current cash generation is weak, with FY2025 FCF at -$0.2B, no buybacks, no dividend, and a need for external financing such as the $573M Morgan Stanley interim facility. Discipline looks mixed: management avoided issuing common stock in Q2 and kept leverage modest at 0.13 debt/equity, but the company is still spending ahead of rent commencement and has been reducing bitcoin holdings to fund development.
RIOT rates a SELL with Low conviction. The AI pivot is credible enough to prevent a high-conviction bearish call, but the stock still looks too expensive for a company with current margins this weak, free cash flow still negative, and major rent streams delayed until late 2027. The setup is especially awkward after a sharp recent rally and with shares still trading well above the 200-day average. A move to HOLD or BUY would require either a much lower entry price or clear proof that the AI strategy is converting into on-time delivery, signed Corsicana capacity, and visible earnings power rather than just contracted backlog.
≣ Full research note tap to expand
Executive Summary
SELL, Low conviction. Rated SELL, Low conviction. The stock has a credible new AI data-center angle, anchored by a $9.1B 20-year lease and $9.8B of contracted data-center revenue, but the present business is still weak: FY2025 operating margin was -62.0%, net margin -86.9%, and Q2 2026 adjusted EBITDA was -$70M. With shares at 12.23x sales versus a 3.4x peer average and still 25.1% above the 200-day average, the market is paying today for earnings power that does not show up in cash flow for another 15 months.
Investment Thesis
RIOT's problem is not the absence of a story. It has one of the better stories in crypto-adjacent infrastructure right now. The problem is the price already reflects a lot of that story before the economics arrive. At $23.75, investors are paying 12.23x FY2025 sales for a business that still posted -$1.95 of EPS, -62.0% operating margin, -86.9% net margin, and -$0.2B of free cash flow. That is hard to justify when the key AI lease does not start rent until December 2027.
The bullish case is real. Management landed a 191 MW lease with a leading frontier AI lab, worth $9.1B of contract revenue over 20 years and $7.3B-$8.2B of NOI at an 80%-90% margin. AMD is ramping, recurring lease revenue hit $4.9M in Q2, contracted data-center revenue is now $9.8B, and the engineering business has a $177.1M backlog at 27%+ margins. Those facts support the idea that RIOT can eventually become more digital infrastructure landlord than bitcoin miner. If that transition works, the stock may ultimately deserve a much higher quality multiple than commodity mining peers.
But the case depends on three things that are not yet delivered. First, execution has to improve materially, because the last three quarters missed EPS by -822.7%, -336.4%, and -124.2%. Second, the company has to bridge a long cash-flow gap without destroying value, since Q2 adjusted EBITDA was -$70M and GAAP net loss was $237M, even with $1.2B of liquidity and $573M of interim financing support. Third, Corsicana has to move from a nonbinding 1 GW LOI to a signed agreement before the market can underwrite the next leg of the AI thesis.
The stock is also not coming from a capitulation setup. It is only 21.7% below its 52-week high, up 25.59% in the last month, and sits 25.1% above its 200-day average. That is the wrong setup for paying forward for a still-unproven model transition. The AI pivot deserves attention and could work, but at this price the better stance is to avoid the shares until either fundamentals catch up or the valuation resets materially.
Key Metrics
| Metric | Value | Context |
|---|---|---|
| Current price | $23.75 | Reference point for target and scenario analysis |
| Price target | $12.25 | Ground-truth target, -48.4% vs current, Low confidence, sales_quality_blend |
| FY2025 revenue | $0.6B | 63.8% YoY growth, but quality of earnings remains weak |
| FY2025 EPS | -$1.95 | -315.3% YoY, no meaningful P/E |
| FY2025 operating margin | -62.0% | Collapsed from 40.8% in the prior fiscal year |
| FY2025 net margin | -86.9% | Shows severe earnings dilution from the transition phase |
| Valuation | 12.23x P/S | Versus peer average 3.4x |
| Balance sheet | 0.13 debt/equity, 1.56 current ratio, $0.3B total debt | Leverage is manageable, liquidity not the immediate issue |
| Free cash flow | -$0.2B | Development spend is outrunning internal cash generation |
| Q2 2026 EBITDA | -$0.14B | -78.5% margin, down 5,140 bps QoQ and 6,940 bps YoY |
| AI and data-center contracts | $9.8B contracted revenue | Includes the new $9.1B AI lab lease |
| Price action | -21.7% vs 52-week high, +25.1% vs 200-day average | Not a washed-out setup despite weak earnings delivery |
Financial Strength
The balance sheet is not the immediate problem. Debt is modest at $0.3B with 0.13 debt/equity, the current ratio is 1.56, and management cited $1.2B of liquidity plus $573M of interim financing support. The issue is economic conversion, not solvency: RIOT is carrying enough liquidity to build, but current operations are consuming value, with sharply negative EBITDA and free cash flow while bitcoin holdings are being reduced to fund development. Margin trajectory is therefore bifurcated, bad today and potentially excellent later, but the bridge period remains expensive and execution-sensitive.
Competitive Position
Riot's edge is no longer just mining scale. It is control of large, already-developed power sites in Texas, especially Rockdale and the 1 GW Corsicana campus, plus an engineering arm that can support build-out and raise switching costs for tenants. That moat is real today because frontier AI demand is chasing scarce powered land, but its durability is still unproven because hyperscale and private infrastructure competitors can also bring capital and relationships if Riot misses delivery timelines.
Management & Guidance
Management's current guidance is strategically attractive and financially back-end loaded. The company is guiding to a 191 MW AI lab deployment at Rockdale, with the first 96 MW delivered by December 2027 and full deployment by June 2028, plus project capex of $11M-$12M per IT MW and lease economics that imply 80%-90% NOI margins. That guidance could transform the business, but credibility is not clean because the recent earnings track record is poor and the company has previously cut operating targets, including hash-rate guidance on the 2024 Q3 call. The milestones to trust are physical and near-term: AMD's 10 MW in November 2026, 15 MW in May 2027, financing close timing, and any conversion of Corsicana's 1 GW LOI into a binding lease.
Segment Analysis
- Bitcoin mining and related activity (Majority of FY2025 revenue, exact split not disclosed in provided data): Still the economic base today, but profitability has worsened sharply as the company funds the data-center transition and reduced bitcoin holdings.
- Data Center ($4.9M recurring lease revenue in Q2 2026): Recurring lease revenue rose 400%+ QoQ as AMD's first 25 MW came online, and this segment now carries the strategic valuation case through the $9.8B contracted pipeline.
- Engineering ($177.1M backlog): Margins expanded to above 27%, with roughly 90% data-center exposure, making it a useful internal capability and a growing earnings support leg.
- Geography (Predominantly U.S.): Rockdale and Corsicana in Texas are the critical assets, with the investment case tied to domestic power availability and permitting rather than international expansion.
Capital Allocation
Capital allocation has shifted from maximizing bitcoin exposure to funding AI infrastructure. That is strategically sensible, but current cash generation is weak, with FY2025 FCF at -$0.2B, no buybacks, no dividend, and a need for external financing such as the $573M Morgan Stanley interim facility. Discipline looks mixed: management avoided issuing common stock in Q2 and kept leverage modest at 0.13 debt/equity, but the company is still spending ahead of rent commencement and has been reducing bitcoin holdings to fund development.
Management Execution & Track Record
Execution has been uneven at best. On the 2024 Q3 call, management cut FY2025 exit hash-rate guidance to 46.7 EH/s from 56 EH/s and FY2024 to 35 EH/s from 36 EH/s, citing delays, while guiding capex of about $663M through FY2025 and a 1 GW Corsicana site by 2026. Since then, the strategy has clearly shifted from mining growth to AI and data centers, which may prove smart, but the beat-miss record has worsened: after beats in late 2025, the last three quarters all missed badly at -822.7%, -336.4%, and -124.2% versus street EPS. Current guidance is more ambitious in dollar terms than anything management laid out two years ago, including $9.1B of AI lab contract revenue and 80%-90% NOI margins, so credibility now rests less on narrative and more on hitting concrete milestones like AMD's 10 MW in November 2026, the remaining 15 MW in May 2027, and first rent on 96 MW in December 2027.
Positioning & Flows
Ownership is concentrated enough to matter, with the top three institutions holding 49.5%, while insider ownership is only 3.9%. Short interest is elevated at 14.87%, or 51.9M shares and about 3 days-to-cover, which can fuel squeezes around AI headlines, and the positive near-dated options gamma regime helps explain why the stock can run hard on incremental news. That positioning can support tactical upside, but it does not fix the mismatch between current losses and current valuation.
Bull Case
- Riot signed a $9.1B initial 20-year AI lab lease with expected NOI of $7.3B-$8.2B and NOI margin of 80%-90%, a far better long-term economic model than pure bitcoin mining.
- Contracted data-center revenue reached $9.8B across AMD and the new AI lab lease, giving unusually strong revenue visibility for a company that historically depended on commodity-like mining economics.
- Liquidity was $1.2B at Q2 2026 and Morgan Stanley provided $573M of interim financing, which materially reduces near-term funding risk for early development.
- Engineering backlog rose to $177.1M, with margins above 27% and data-center exposure around 90%, showing the services business is becoming more relevant and more profitable.
Bear Case
- Recent execution is poor: RIOT missed EPS by -822.7%, -336.4%, and -124.2% in its last three reported quarters, which cuts against taking long-dated guidance at face value.
- Current operations are deeply loss-making, with Q2 2026 EBITDA of -$0.14B and -78.5% margin, plus a GAAP net loss of $237M.
- The biggest incremental upside asset after Rockdale is still uncommitted, because Corsicana's full 1 GW opportunity remains only a nonbinding LOI with timing and terms uncertain.
- The stock setup is not washed out: shares are only 21.7% below the 52-week high after a 25.59% one-month rally, despite fundamentals that have worsened through 2026.
Valuation & Price Target
Engine price target $12.25 (-48.4% vs current), Low confidence.
- P/S Multiple: $4.2 (27% weight)
- Quality-adjusted: $23.18 (20% weight)
This note aligns with the quant baseline of SELL, and the reason is straightforward: the valuation is already rich relative to current fundamentals. The engine's $12.25 target, based on a sales-quality blend, fits a company with a 47/100 quality score, 57/100 risk score, 12.23x sales versus 3.4x peers, and worsening earnings execution. The one strong counterpoint is the AI lease portfolio, especially the $9.1B Rockdale agreement and $9.8B contracted data-center revenue, which is why conviction stays low rather than high. Sell-side consensus sits at Strong Buy, 11 buy / 0 hold / 0 sell, but that is low-signal context here and looks anchored to the strategic upside more than the current income statement or the delayed start of rent.
Risk Assessment
Risk score 57/100 (Moderate). The biggest risk to the bearish call is successful execution of the AI pivot, because a $9.1B lease with 80%-90% NOI margins can justify much higher value if Riot delivers the first 96 MW on time by December 2027 and converts Corsicana's 1 GW LOI into a binding contract.
The Bottom Line
RIOT rates a SELL with Low conviction. The AI pivot is credible enough to prevent a high-conviction bearish call, but the stock still looks too expensive for a company with current margins this weak, free cash flow still negative, and major rent streams delayed until late 2027. The setup is especially awkward after a sharp recent rally and with shares still trading well above the 200-day average. A move to HOLD or BUY would require either a much lower entry price or clear proof that the AI strategy is converting into on-time delivery, signed Corsicana capacity, and visible earnings power rather than just contracted backlog.
Go deeper in the app
Live charts, options analytics, the full earnings-call breakdown, price alerts, and every other stock — updated in real time.
Get Tickr