Riot Platforms RIOT

At a glance

One-line thesis. Riot is a Texas/Kentucky Bitcoin miner with a small power-engineering arm whose stock nearly doubled in twelve months on the sector's power-to-AI/HPC re-rating narrative — but the reported business earns a negative gross margin, burned $774M of free cash flow in FY25, pays for itself by issuing stock (share count ~3.6× in four years) and debt, and just missed consensus EPS by more than a dollar in back-to-back quarters; at 14× EV/sales and a 3.8 beta this is a Watch — a trading vehicle on Bitcoin with embedded power-asset optionality, not an investible business at this price.

1. What it is

Riot Platforms (Nasdaq: RIOT) is one of the largest US-listed Bitcoin miners. It develops and operates large-scale mining infrastructure at sites in Rockdale and Navarro counties, Texas, and two facilities in Paducah, Kentucky, plus an Engineering segment that designs and manufactures power-distribution equipment and custom electrical solutions for data-center, utility, industrial, and renewable-energy customers. Founded 2000, headquartered in Castle Rock, Colorado; CEO Jason Les; ~783 employees. FMP tags it "Financial Services / Capital Markets" — a quirk of crypto classification; operationally it is an energy-and-compute industrial.

Revenue mix (FY2025 segment data, from filings):

• Bitcoin Mining: $576.3M (vs $321.0M FY24, +80%)
• Engineering: $94.4M (vs $46.5M FY24, +103%)
• Data caveat: the segment lines sum to ~$670.7M against reported total revenue of $647.4M — a small reconciliation gap in the FMP segment feed (likely eliminations); we quote the income-statement total as authoritative. No geographic segment data is provided (effectively all-US).

The story the market is paying for is not in these numbers: it is the sector-wide narrative of miners converting large contracted power positions into AI/HPC data-center hosting — visible in the analyst estimates (revenue ~4× by 2030) and in peers like Hut 8, TeraWulf and Cipher re-rating alongside. That optionality is real as a category; its Riot-specific economics are not yet in the reported financials, and this note will not pretend otherwise.

2. The expert thesis — honesty first (traceable)

No expert-panel coverage of RIOT specifically — this note is fundamentals-driven. A KB sweep for RIOT returns zero company-specific, dated claims. Two tangential, sector-level mentions exist and are cited for completeness, not conviction:

• Lyn Alden (skill 1.1, conviction 60, undated — context indicates 2021 vintage): publicly traded miners are "interesting opportunities if they deploy contracted miners," with the operative example being Marathon's planned 5× miner expansion into Q1 2022 (lyn_alden-xKwH_t_5C_M). Riot is named only in the entity list, and the claim predates the entire current cycle. Weight: ~zero.
• Arthur Hayes (skill 1.2, conviction 55, undated): institutional exposure to Bitcoin miners (Riot named among MARA, CleanSpark, Strategy) "signals long-term directional conviction on the whole ecosystem" (arthur_hayes-szIcXCCTXpo). A sector-flows observation, not a Riot thesis. Weight: ~zero.

There is no high-skill voice underwriting this name, no countervailing short thesis either, and no management guidance ingested in the KB. Everything below is built from the FMP fundamentals, estimates, and technicals in the data file — which is the honest house standard for a screen-surfaced name.

3. Synthos scores & the Bull / Base / Bear cases

The one-glance judgment — three scores, 0–10, each anchored to real metrics:

The three cases (assumptions shown; ~12–18-month fair values; no probability blend — the base case is the expected path):

Synthos fair value = ~$27 (+18%), Street-anchored and low-confidence by construction. The honest signal is the shape, not the midpoint: a −52%/+66% spread on a 3.8-beta name means position sizing, not price targeting, is the real risk control. This is a tracked call — the Forecaster Scorecard grades it once it matures.

4. Exponential Potential

Synthos separates compounders from exponentials. RIOT is neither in the clean sense — it is an optionality vehicle:

• Forward growth (consensus): revenue $665.7M (2026E) → $808.1M (2027E) → $959.0M (2028E) → $1.91B (2029E) → $2.80B (2030E) — a ~43% CAGR with the hockey-stick bending in 2029, which is where analysts evidently locate the power/AI-HPC monetization.
• But the quality of that exponential is poor: the same estimate set has EBITDA negative in every single year (2026E −$24M through 2030E −$99M) while net income turns positive in 2029 ($715M avg) — an internal inconsistency that tells you these long-horizon models are few (5–6 analysts), messy, and probably driven by BTC mark-to-market assumptions rather than operating earnings.
• Per-share reality: every prior growth phase was funded by issuance — $764M of stock sold in 2023, $965M in 2024, $208M in 2025. If the 2029–2030 ramp is funded the same way, the company can 4× while the share does far less.
• Room to run: an $8.6B market cap against a multi-gigawatt Texas power position is the real option. Peers (HUT $11.7B, WULF $11.0B, CIFR $8.9B) have re-rated on the same theme, so RIOT is not even the cheapest expression of it.

Exponential Potential: Moderate (5/10). Big TAM story, thin and contradictory analyst support, and a capital structure that historically converts enterprise growth into shareholder dilution.

5. Financials (real numbers — FMP annual/quarterly)

• Revenue: FY25 $647.4M, +71.9% (FY24 $376.7M, +34.2% on FY23 $280.7M). TTM ~$653M.
• Quarterly trajectory: Q1'25 $161.4M → Q2 $153.0M → Q3 $180.2M → Q4 $152.8M → Q1'26 $167.2M (+3.6% YoY). Choppy, not compounding — mining revenue tracks BTC price and halving economics, not customer demand.
• Margins — the core problem: FY25 gross profit −$101.3M (gross margin −15.6%), because cost of revenue ($748.7M) exceeds revenue once the heavy D&A of mining rigs ($346.8M FY25) flows through. Even on a cash basis the picture fails: EBITDA was −$53.2M — the ~38% "cash gross margin" is fully consumed by $298.5M of SG&A (46% of revenue, including $125.7M of stock comp). Operating income −$400.0M; net income −$663.2M (EPS −$1.95).
• Q1 2026: net loss −$500.5M (EPS −$1.44), of which −$357.8M sits in non-operating items — in a miner's P&L this line is dominated by fair-value remeasurement of the bitcoin held, i.e., the balance sheet's BTC marks now swing the income statement violently both ways (Q3'25 was positive $188M; Q4'25 negative $556M).
• Cash flow: FY25 operating cash flow −$572.9M, capex −$201.4M, FCF −$774.3M. Funded by $251.9M net debt issuance and $203.4M net stock issuance. FY24 was worse on capex (FCF −$1.52B incl. the big build-out). There is no year in the file with meaningfully positive FCF.
• Balance sheet: cash $233.5M vs total debt $866.8M ($260.2M short-term) → net debt $633.2M; current ratio 1.08. PP&E $1.56B; "other non-current assets" $1.78B is where the bitcoin hoard sits — the real asset story. Equity $2.86B (book value/share $6.89; tangible $5.99), retained earnings −$1.35B of accumulated lifetime losses.
• Dilution (the tell): weighted shares 93.5M (2021) → 139.4M (2022) → 175.0M (2023) → 276.0M (2024) → 340.7M (2025) → 347.6M (Q1'26). ~3.6× in four years.

6. Valuation — priced in or room?

There is no earnings-based valuation to do: TTM EPS is −$2.49, TTM FCF yield −11.8%, and consensus EPS stays negative until 2029. What the market is paying: EV/sales 14.3× on a negative-gross-margin revenue stream, 3.3× book on a balance sheet whose main asset is bitcoin plus mining rigs that depreciate at ~$347M/yr. FMP's quant rating is C− (overall 1/5; DCF, ROE, ROA and P/E scores all 1/5). Against that, the sell side is near-unanimous: 17 Buy, 1 Hold, 0 Sell, consensus target $27.25 in a tight $24–31 band — a striking disconnect between quant fundamentals and narrative-driven coverage, and honestly the tight band looks like herding around the same power-optionality story rather than independent models (the long-horizon estimates' internal inconsistencies in §4 support that read). If the AI/HPC option converts, today's price is defensible; if RIOT stays a pure miner, 14× EV/S for negative-EBITDA revenue has no fundamental floor above book (~$6.89/share). That asymmetry is the whole valuation section.

7. Technicals (from the tech block)

• Trend: intact but cooling. $22.87 is below the 50-DMA ($24.44) and well above the 200-DMA ($18.38); the longer trend is up, the shorter trend has rolled over.
• Location: −20.3% off the 52-week high ($28.69) and +107% off the 52-week low ($11.03) (the quote feed shows a slightly wider $10.59–$30.32 range; we use the technicals block consistently). Historical max drawdown from peak: −70.6% — the realized-volatility warning label.
• Momentum: RSI(14) 34 — soft, approaching but not yet in oversold territory. MACD −0.04, essentially flat-negative: momentum has stalled.
• Relative strength: +87.5% 12-mo vs SPY +21.1% / QQQ +31.2%; +77.8% 3-mo vs SPY +14.6% / QQQ +23.6%. Enormous outperformance — all of it delivered in the last two quarters, which is exactly what a narrative re-rating looks like.
• Read: a huge run that is now correcting below its 50-DMA with fading momentum. For a Watch name, the technically interesting zone is a retest of the rising 200-DMA (~$18) that holds — that is where the trigger in our strategic line lives.

8. Moat & competitive position

Bitcoin mining has no moat in the classic sense: the product is a perfect commodity, the network difficulty adjusts to squeeze margins toward the marginal producer, and every four years the halving cuts gross revenue per unit of work. What miners can own is cheap contracted power and sited, energized infrastructure — Riot's Texas/Navarro position and its Engineering segment (power-distribution manufacturing) are genuine assets in a power-constrained AI build-out, and that is the entire re-rating thesis. But it is a shared thesis: Hut 8 ($11.7B), TeraWulf ($11.0B), Cipher ($8.9B), Bitmine ($8.9B), and MARA ($4.9B) are the FMP-supplied peers chasing the same conversion, several with announced hosting deals — the data file gives us no basis to claim Riot leads that race. Peer-list caveat: the FMP set also includes irrelevant names (Cullen/Frost and Old National — regional banks — plus UWM, XP, StepStone), an artifact of the "Capital Markets" industry tag; judge RIOT only against the miner cohort. ROIC of −14% says whatever moat exists has never yet earned its cost of capital.

9. Management, capital allocation & guidance

• Capital allocation — the record is the verdict: cumulative net stock issuance of ~$1.93B across 2022–2025 ($295M + $764M + $965M + $203M — the FY22 figure per the cash-flow statement) against −$1.35B of accumulated retained losses. Capital is raised from shareholders and consumed by operations and capex; no dividend, no meaningful buyback ($4.3M repurchased FY25, likely tax-withholding). In FY25 the company also added $252M of net debt, including a $260M short-term slug — a new fragility given negative operating cash flow.
• Insider activity (2026-07-01 filings): routine — four directors received identical 8,347-share RSU awards; a senior advisor converted RSUs; SVP/CAO Ryan Werner had 13,869 shares withheld for taxes at $23.96. No discretionary open-market buys — nobody with the best information is reaching for stock at this price — and no alarming discretionary sales either.
• Guidance: none ingested — no management claims in the KB and no guidance fields in the data file. What we can verify is the estimate track record: the last two prints missed consensus EPS by −$1.11 (Q1'26: −$1.44 vs −$0.33) and −$1.81 (Q4'25: −$2.03 vs −$0.22). BTC mark-to-market drives much of that, but it means the "Street EPS" line for this name carries little information.

10. Catalysts & what to watch

• Next earnings: 2026-07-30 (Q2 2026; Street EPS est −$0.21, revenue est ~$155.5M). Given the miss pattern, watch the cash items, not the EPS line: operating cash flow, cost-to-mine, and any change in the issuance run-rate.
• A signed, disclosed AI/HPC hosting deal — the single event that would move this from narrative to underwritable. Watch the Engineering/hosting revenue lines for third-party data-center economics actually landing.
• Bitcoin price — the dominant variable for revenue, the balance-sheet marks, and the cost of equity capital. Not a catalyst so much as the weather.
• Financing actions: the $260M short-term debt maturity/refinance, and the pace of ATM stock issuance (each 10% of new shares is a direct tax on the thesis).
• Peer read-through: hosting economics disclosed by HUT/WULF/CIFR reprice the whole cohort, RIOT included.

Thesis tripwires (what would change the call): a disclosed hosting contract with real revenue attached (upgrade trigger); a hold of the ~$18 200-DMA zone on a BTC-stable pullback (entry trigger); accelerating dilution or a BTC bear leg (downgrade to Avoid).

11. Key risks

• Bitcoin price (the dominant risk): revenue, the coin hoard, the equity currency, and sentiment all fall together — there is no diversification inside this stock.
• Structural unprofitability: negative gross margin after rig depreciation; halving cycles mechanically squeeze mining economics every four years regardless of execution.
• Dilution treadmill: operations are funded by stock issuance; a lower share price makes every dollar raised more dilutive — a reflexive downside loop.
• Leverage, newly added: $867M total debt ($260M short-term) against negative operating cash flow is a fragility the FY23 balance sheet (net cash) did not have.
• Narrative risk: the AI/HPC conversion is a sector story not yet in Riot's numbers; if the market stops paying for the option, 14× EV/S has no earnings floor above ~$6.89 book.
• Estimate quality: thin coverage (5–8 analysts at long horizons) with internally inconsistent models — the consensus anchor under our base case is soft.
• Volatility itself: beta 3.81 and a realized −71% max drawdown; position sizing errors are unrecoverable in names like this.

12. Verdict, position sizing & monitoring

Watch. The honest summary: a +87% twelve-month tape, a near-unanimous Buy-rated sell side, and a genuinely large power-asset option — sitting on top of a business that has never sustainably generated cash, earns a negative gross margin, and finances itself out of its own shareholders' pockets. Synthos does not buy 3.8-beta commodity-levered vehicles at 14× EV/sales on a narrative the reported numbers can't yet see. But we don't dismiss the option either: the power-to-AI/HPC conversion is exactly the kind of forward exponential this house exists to catch when the evidence arrives.

• Sizing: 0% in the flagship. For a speculative sleeve only: <1%, treated explicitly as a BTC-beta trade with a hard stop, never averaged down.
• What flips it to Tactical: a disclosed hosting contract with real economics, or a washout entry near the 200-DMA (~$18) with Bitcoin's trend intact — cheaper optionality, same option.
• What flips it to Avoid: a BTC bear leg, or another year of >10% share-count growth with no hosting conversion.
• Monitoring: re-score at the 2026-07-30 print (cash costs and issuance pace, not EPS); track peer hosting disclosures. This verdict is logged as a tracked Synthos call as of 2026-07-06 at $22.87.
• Single biggest risk: a Bitcoin drawdown that hits revenue, the balance sheet, and the funding model simultaneously.

Provenance & disclosures

• Traceability: zero company-specific KB claims on RIOT. Two tangential, undated sector-level mentions (Lyn Alden lyn_alden-xKwH_t_5C_M; Arthur Hayes arthur_hayes-szIcXCCTXpo) are cited in §2 at ~zero weight. kb_net_conviction is null because there is nothing to aggregate — fabricated conviction is structurally impossible here, and this note claims none.
• Data as-of: fundamentals 2026-03-31 (Q1 2026, filed 2026-04-30) · estimates & prices 2026-07-06 (FMP) · no dated expert claims. Forward figures are analyst consensus (FMP), labeled as estimates.
• Fair-value caveat: the $27 base case is anchored on Street consensus ($27.25) because negative FCF and negative near-term EPS make any in-house DCF false precision; the bull case applies ~18× to the thin 2029 consensus EPS of $2.10, the bear case marks to the 52-week-low region. All three are labeled assumptions, not outputs of a defensible model.
• Estimate-quality caveat: long-horizon consensus is 5–8 analysts and internally inconsistent (negative EBITDA with positive net income in 2029–30; 2029 EPS above 2030). Treat all forward figures as soft.
• Segment caveat: FMP segment revenue ($670.7M) does not reconcile exactly to the income statement ($647.4M); the income statement is authoritative. No geographic segment data supplied.
• Peer caveat: the FMP peer list mixes true comps (HUT, WULF, CIFR, MARA, BMNR) with irrelevant financials (Cullen/Frost, Old National, UWM, XP, StepStone) due to the sector tag; judge only against the miner cohort.
• Not investment advice. Independent research, educational and informational only, never personalized. Hypothetical/forward figures are labeled; the only performance numbers Synthos will headline are the live, real-money Flagship's.
• Version: 2026-07-06. Prior versions available via the deep-dive version dropdown ("based on the info at the time").