The Cheapest AGI-9 Company: Ethereum Treasury + AI Infrastructure at a 32% Discount
Analysis Date: March 12, 2026 | Data Sources: SEC 10-K (FY2024), Q3 2025 Earnings, XBRL, yfinance, Investor Presentation Feb 2026
What you get for $525M: You pay $525M and receive $323M of ETH + a $453M stake in a publicly-traded AI infrastructure company + $179M cash, minus $187M in debt. That is $768M of identifiable assets for $525M -- a 32% discount to sum-of-parts NAV.
The catch: The company has reinvented itself three times in four years (Bitcoin miner to AI cloud to Ethereum treasury). The ETH treasury strategy is MicroStrategy-for-Ethereum -- levered exposure to a volatile asset. The WhiteFiber (WYFI) spinoff creates value but also complexity. Management execution is unproven at scale. Only 54 employees.
Floor confidence: LOW. The assets are real and verifiable (ETH on-chain, WYFI publicly traded), but the floor depends on ETH price which can drop 80% in a crypto winter. This is not a Buffett-style floor -- it is a sum-of-parts discount on volatile assets.
10x plausibility: MODERATE. $525M to $5.2B requires WYFI scaling to multi-GW AI infrastructure AND ETH price appreciation. WYFI alone at $5B would put BTBT's 70% stake at $3.5B. Not impossible if they execute on the 1.6GW pipeline, but the path requires many things to go right.
Bit Digital is not one business -- it is a holding company with three distinct operations that have evolved dramatically over time:
| Period | Identity | Main Revenue |
|---|---|---|
| 2020-2023 | Bitcoin Miner | BTC mining (100% of revenue) |
| 2024 | Bitcoin Miner + AI Cloud | Mining 54%, Cloud 42%, Other 4% |
| Q3 2025 | ETH Treasury + AI Infra | Cloud 59%, Mining 24%, Staking 10%, Colo 6% |
| 2026+ | Strategic Asset Company (SAC) | ETH yield + WYFI stake |
As of early 2026, BTBT has reorganized into two pillars:
| CEO | Sam Tabar (age 53) -- former co-founder of Meridio (acquired by Consensys), Wall Street background |
| CFO | Erke Huang (age 36) -- also a Director |
| Employees | 54 (extremely lean for a $525M company) |
| HQ | 31 Hudson Yards, Floor 11, New York, NY |
| Incorporation | Cayman Islands (originally Chinese company "Golden Bull Limited", renamed 2020) |
| Auditor | Audit Alliance LLP (small firm -- yellow flag) |
| Governance Risk | Overall Risk Score: 10/10 (highest risk per yfinance) -- Board risk 9, Comp risk 10, Audit risk 9 |
| Insider Ownership | 2.04% (low) |
| Institutional Ownership | 53.7% |
| Short Interest | 18.6% of float (significant bearish sentiment) |
This is the central question for BTBT. The company trades at $525M but holds identifiable assets worth substantially more. Let us decompose exactly what you are buying.
| Asset | Quantity | Price/Value | Total | Notes |
|---|---|---|---|---|
| ETH Holdings | 155,277 ETH | $2,079 | $323M | On-chain verifiable; staked for ~3% yield |
| WhiteFiber (WYFI) Stake | 27.04M shares (70.4%) | $16.77 | $453M | Publicly traded, mark-to-market |
| Cash & Equivalents | -- | -- | $179M | As of Q3 2025 |
| BTC Mining Assets | ~1.8 EH/s | winding down | ~$10M | Being sold/converted; minimal residual value |
| Total Gross Assets | $965M | |||
| Convertible Notes (4%, 2030) | -- | $4.16 conversion | -$150M | Converts at $4.16/share; dilutive above that price |
| Other Liabilities | -- | -- | -$37M | Est. non-convert debt/liabilities |
| Net Asset Value | $768M | $2.37/share | ||
| Market Cap | $525M | $1.62/share | ||
| Discount to NAV | 32% | You pay 68 cents per $1 of assets |
| Segment | FY2022 | FY2023 | FY2024 | Q3 2025 (Quarterly) | Q3 Ann. | Trend |
|---|---|---|---|---|---|---|
| Digital Asset Mining | $32.3M | $44.2M | $58.6M | $7.4M | $29.6M | WINDING DOWN |
| Cloud Services (WYFI) | -- | -- | $45.7M | $18.0M | $72.0M | GROWING +48% YoY |
| Colocation (WYFI/Enovum) | -- | -- | $1.4M | $1.7M | $6.8M | RAMPING |
| ETH Staking | $0.03M | $0.7M | $1.8M | $2.9M | $11.6M | +542% YoY |
| Other | -- | -- | $0.6M | $0.5M | $2.0M | |
| Total Revenue | $32.3M | $44.9M | $108.1M | $30.5M | $122.0M | +141% FY24 YoY |
| Metric | FY2022 | FY2023 | FY2024 | Notes |
|---|---|---|---|---|
| Gross Profit | $11.9M | $15.4M | $101.4M | Explosive improvement |
| Gross Margin | 37% | 34% | 62% | Cloud services at ~57% gross margin |
| Operating Income | -$107.2M | -$16.6M | $27.6M | First profitable year operationally |
| Net Income | -$105.3M | -$13.9M | $28.3M | Includes $55.7M digital asset gains |
| Adjusted EBITDA | $54.8M | -$22.3M | $60.7M | Normalized (excl. one-time items) |
| Operating Cash Flow | -$8.5M | $1.1M | -$13.0M | Negative due to working capital timing |
| Free Cash Flow | -$27.8M | -$65.6M | -$107.0M | Heavy CapEx for data centers + GPUs |
BTBT has never generated positive free cash flow. Cumulative FCF from 2021-2024: -$264M. The company has funded itself entirely through equity issuance ($242.9M in FY2024 alone) and now convertible debt ($150M in Oct 2025).
This is not inherently disqualifying -- high-growth infrastructure companies often burn cash while building capacity. But the magnitude is significant: BTBT issued more in equity ($243M) than its entire FY2024 revenue ($108M). Shareholders are paying for growth via dilution.
Share count trajectory: 55M (2021) --> 79M (2022) --> 108M (2023) --> 179M (2024) --> 324M (2025). A 5.9x dilution in 4 years.
| Item | FY2023 | FY2024 | Change |
|---|---|---|---|
| Cash & Equivalents | $16.9M | $95.2M | +$78.3M |
| Digital Assets | $41.0M | $161.4M | +$120.4M |
| Current Assets | $83.3M | $296.9M | +$213.6M |
| PP&E (Net) | $91.9M | $161.3M | +$69.4M |
| Goodwill | $0 | $19.4M | New (Enovum acquisition) |
| Intangibles | $0 | $13.0M | New (Enovum acquisition) |
| Investments | $4.4M | $37.6M | +$33.2M |
| Total Assets | $189.3M | $538.2M | +$348.9M |
| Current Liabilities | $27.1M | $55.0M | +$27.9M |
| - Deferred Revenue | $13.1M | $30.7M | Customer prepayments (good) |
| Non-Current Liabilities | $9.5M | $19.7M | Mostly leases |
| Total Liabilities | $36.6M | $74.8M | +$38.2M |
| Stockholders' Equity | $152.7M | $463.5M | +$310.8M |
| Current Ratio | 3.1x | 5.4x | Very strong |
Note: By Q3 2025, total assets had grown to $1,113M and liabilities to $87M (including the $150M convertible notes issued in October 2025). The balance sheet is substantially larger than the FY2024 figures above.
| Asset | Quantity | Cost Basis | Fair Value (Dec 31, 2024) |
|---|---|---|---|
| Bitcoin (BTC) | 741.9 BTC | $43.9M | $69.3M |
| Ethereum (ETH) | 27,623 ETH | $68.1M | $92.1M |
| Total Digital Assets | $112.0M | $161.4M |
By October 2025, ETH holdings had exploded to 155,277 ETH (up from 27,623 at year-end 2024). This was funded by the $150M convertible notes, ATM equity sales, and BTC-to-ETH conversions. The company has deliberately built the largest ETH treasury of any public company outside Ethereum Foundation.
| Facility | Capacity | Location | Status | Details |
|---|---|---|---|---|
| MTL-1 | 4 MW | Montreal, Canada | OPERATIONAL | Tier-3, fully leased, 5,000+ GPUs (H200/H100), 14 customers, ~30mo avg lease term |
| MTL-3 | 7 MW | Montreal, Canada | OPERATIONAL | Fully operational per Feb 2026 presentation |
| MTL-2 | 5 MW | Pointe-Claire, QC | ON HOLD | 160K sqft former manufacturing, development paused |
| NC-1 | 24-99 MW | North Carolina, USA | IN DEVELOPMENT | Initial 24MW, expandable to 99MW over 4 years. Cerebras contract. |
| Iceland | 6 MW | Blonduos, Iceland | OPERATIONAL | Colocation agreement, 100% renewable hydro, cloud services GPU hosting |
| Total Online (2025) | ~11 MW | |||
| Target YE 2026 | ~76 MW | Includes NC-1 ramp | ||
| Pipeline | ~1.6 GW | Sites under evaluation |
Total Contract Value: $865M (announced April 2025). This is the single largest contract in BTBT/WYFI history. Cerebras, an AI chip company, contracted for colocation space at the NC-1 facility. This is the kind of validation the AI pivot thesis requires -- a real AI customer making a real financial commitment.
$275M total contract value, 4,096 GPUs over two three-year periods (~$92M annualized). This customer drove most of the $45.7M cloud revenue in FY2024. However, the customer requested delays in July 2024 to evaluate newer GPUs, creating execution risk. The customer made a $30M non-refundable prepayment in August 2024.
Master Services Agreement for GPU-as-a-service for cloud gaming. Initial orders: 300 GPUs ($4.6M/5yr) + 200 GPUs ($3.2M/5yr). Option to expand up to 50,000 servers ($700M potential). Still early but diversifies beyond pure AI workloads.
| CapEx per MW | $7-9M |
| Debt financing | 70-75% |
| Revenue per MW | $1.7-2.5M/yr |
| Contract term | 4-12 years |
| Profit margin | 75-85% |
| Power source | Hydro-Quebec (renewable) |
| Build time | ~6 months (retrofit) |
| GPU CapEx | $30-50K/GPU |
| Contract term | ~36 months |
| Revenue per card | $18-30K/yr |
| Profit margin | ~85% |
| Current fleet | ~4,500 GPUs |
| GPU types | H200, B200, GB200 |
| Partners | NVIDIA NCP, SuperMicro, Dell, HPE |
In June 2025, BTBT announced a strategic pivot to become a "pure-play Ethereum treasury and staking company." This mirrors MicroStrategy's Bitcoin treasury strategy but applied to Ethereum. The key differences:
| Feature | MicroStrategy (BTC) | Bit Digital (ETH) |
|---|---|---|
| Asset | Bitcoin | Ethereum |
| Holdings | ~500K+ BTC (~$35B+) | 155K ETH (~$323M) |
| Market Cap | ~$70B+ | $525M |
| Premium to NAV | Historically 1.5-3x | 0.68x (discount!) |
| Yield on Asset | 0% (BTC doesn't stake) | ~3% (ETH staking) |
| Funding | Convertible notes + equity | Convertible notes + equity + ATM |
| Other Business | Legacy software (~$500M rev) | WYFI AI infra (~$80M+ rev) |
| Date | ETH Held | Value (at date) | Action |
|---|---|---|---|
| Dec 31, 2024 | 27,623 | $92M | Pre-strategy; held from mining/staking |
| Jun 30, 2025 | 30,663 | -- | Strategy announced; beginning accumulation |
| Sep 30, 2025 | 122,187 | ~$350M | Aggressive buying via equity + convertible |
| Oct 31, 2025 | 153,547 | $591M | Used $150M convertible proceeds to buy ~31K ETH |
| Dec 31, 2025 | 155,277 | $461M | ETH price dropped from ~$3.85K to ~$2.97K |
| Mar 12, 2026 (est.) | ~155K+ | ~$323M | ETH at ~$2,079 (further decline) |
BTBT bought most of its ETH at significantly higher prices. At Oct 31, 2025 when they had 153K ETH valued at $591M, ETH was ~$3,850. Current ETH price: $2,079. That is a 46% decline.
Estimated unrealized loss on ETH position: ~$268M (from peak acquisition prices). This is a massive impairment to book value that will show in Q4 2025 / Q1 2026 financials.
If ETH drops to $1,500 (2022 bear market levels), the 155K ETH would be worth $233M -- versus $591M when purchased. The equity dilution taken to buy this ETH would have destroyed enormous shareholder value.
With 155K ETH representing ~42% of NAV, a 50% drop in ETH (which has happened multiple times historically) would destroy ~$160M of value -- roughly 30% of the current market cap. ETH dropped 82% from peak in 2022. The company has concentrated its balance sheet in a single volatile asset with no hedge.
Probability: HIGH (30-40% over 2 years) | Impact: SEVERE
Share count has gone from 55M (2021) to 324M (2026) -- a 5.9x increase. The company still has $202M available under its ATM registration statement. The convertible notes add another potential 36M shares. Management's primary capital allocation tool is selling equity to buy crypto. This is a transfer of value from existing shareholders to the ETH position.
Probability: VERY HIGH (near certain) | Impact: MODERATE-SEVERE
Cayman Islands incorporation. Overall governance risk score: 10/10 (worst possible). Insider ownership only 2%. Small, unfamiliar auditor. Board member Brock Pierce has a controversial background. The company was originally a Chinese company called "Golden Bull Limited." Compensation risk score: 10/10. These are red flags that could indicate poor alignment with minority shareholders.
Probability: ONGOING | Impact: MODERATE
WhiteFiber needs to scale from 11MW to 76MW by YE2026. The NC-1 (99MW) facility is the centerpiece and must be built, powered, and occupied. The Cerebras contract provides demand but construction/execution risk remains. If WYFI stumbles, BTBT loses its growth engine.
Probability: MODERATE (20-30%) | Impact: HIGH
Cloud services revenue ($45.7M in FY2024) was largely from a SINGLE initial customer. The Cerebras colocation contract is the second major customer. If either relationship deteriorates, revenue impact is severe. 20+ customers reported but revenue is heavily concentrated.
Probability: MODERATE | Impact: HIGH
$150M of 4% convertible notes due 2030, converting at $4.16/share. If BTBT rises above $4.16, these convert into ~36M new shares (11% dilution). If BTBT stays below $4.16, the company must repay $150M + interest in 2030. At current price ($1.62), this is a debt obligation, not equity -- which is actually the worse scenario.
Probability: MODERATE | Impact: MODERATE
BTBT is classified under SIC 6199 (Finance Services) and "Capital Markets" industry. ETH's regulatory status is uncertain -- potential classification as a security could impact staking operations and holdings value. Additionally, crypto-focused companies face banking and compliance headwinds.
Probability: LOW-MODERATE | Impact: MODERATE-HIGH
| Risk | Probability | Impact | Can We Model It? |
|---|---|---|---|
| ETH price crash | High | Severe | Yes -- sensitivity analysis |
| Continued dilution | Very High | Moderate-Severe | Yes -- ATM + convert math |
| Governance failure | Ongoing | Moderate | No -- binary event |
| WYFI execution miss | Moderate | High | Partially -- milestone tracking |
| Customer loss | Moderate | High | Partially |
| Regulatory action on ETH | Low-Moderate | High | No |
| Bitcoin mining irrelevance | Already happening | Low (winding down) | N/A |
| ETH Price | ETH Value (155K) | WYFI Stake (current) | Cash Net Debt | NAV | NAV/Share | vs $1.62 |
|---|---|---|---|---|---|---|
| $1,000 (bear) | $155M | $453M | -$8M | $600M | $1.85 | -12% discount |
| $1,500 | $233M | $453M | -$8M | $678M | $2.09 | -23% discount |
| $2,079 (current) | $323M | $453M | -$8M | $768M | $2.37 | -32% discount |
| $3,000 | $466M | $453M | -$8M | $911M | $2.81 | -42% discount |
| $4,000 | $621M | $453M | -$8M | $1,066M | $3.29 | -51% discount |
| $5,000 | $776M | $453M | -$8M | $1,221M | $3.77 | -57% discount |
For BTBT to reach the bull case of $5-10B, you need a combination of:
| Scenario | ETH Value | WYFI Stake Value | Other | Total | Plausibility |
|---|---|---|---|---|---|
| Bull Case A: ETH + WYFI both moon | $1.6B ETH at $10K |
$3.5B WYFI at $130 (5x) |
$200M | $5.3B | POSSIBLE 10-15% |
| Bull Case B: WYFI becomes major infra | $775M ETH at $5K |
$7B WYFI at $260 (16x) |
$200M | $8B | UNLIKELY 3-5% |
| Bull Case C: ETH treasury premium (like MSTR) | $1.6B at 2x NAV ETH $5K + 2x premium |
$2.3B WYFI at $85 (5x) |
$200M | $5.7B | POSSIBLE 8-12% |
| Target | Price | Market Cap | Methodology | Confidence |
|---|---|---|---|---|
| Very Safe (Floor) | $0.70-1.00 | $225-325M | ETH at $1,000 (bear case) + WYFI at 50% of current + cash. Assumes crypto winter, WYFI struggles. This is NOT a hard floor because ETH can go lower. | LOW |
| Fair Value | $2.00-2.50 | $650-810M | NAV at current ETH/WYFI prices with modest discount (10-15%) for holding company structure. Closing the NAV gap. | MODERATE |
| Best Case 5yr | $8-15 | $2.6-4.9B | ETH at $5-10K, WYFI scales to 200+ MW and $3-5B market cap, BTBT trades at NAV or small premium. Requires ETH bull market + WYFI execution. | LOW |
| Company | Market Cap | Power (MW) | $/MW | AI/HPC Revenue % | AGI Score | AI Pivot Stage |
|---|---|---|---|---|---|---|
| Core Scientific (CORZ) | $5.1B | ~1,400 | $3.6M | ~25% | 9 | 590MW contracted to CoreWeave |
| Iris Energy (IREN) | $3.0B | ~1,000 | $3.0M | ~15% | 8 | Building GPU cloud |
| Riot Platforms (RIOT) | $4.0B | ~1,000 | $4.0M | <5% | 7 | Mostly mining still |
| Bit Digital (BTBT) | $525M | 11 (76 target) | $48M* | ~65% | 9 | Furthest AI pivot % + ETH treasury |
| WhiteFiber (WYFI) | $642M | 11 (76 target) | $58M* | 100% | N/A | Pure AI infra spinoff from BTBT |
*$/MW for BTBT and WYFI is extremely high relative to current capacity but reflects heavy growth expectations. At 76MW target, WYFI = $8.4M/MW -- in line with "transitioning" peers.
| Condition | Status | Notes |
|---|---|---|
| NAV discount > 40% | MET (32%) | Close but not quite 40%. Gets there if ETH stabilizes and WYFI holds. |
| ETH at or near cycle low | APPROACHING | ETH at $2,079 vs. $4,800+ peak. Could go lower. |
| WYFI execution milestones hit | PARTIAL | Cerebras signed, MTL-1 operational, NC-1 in development. But only 11MW live. |
| Dilution slows or stops | NOT MET | $202M ATM still available. Management has shown no restraint. |
| Governance improves | NOT MET | No signs of improvement. Still 10/10 risk score. |
Bottom line: BTBT is a WATCHLIST position, not a conviction buy. The NAV discount and 10x math are compelling, but the governance and dilution risks mean you cannot size this like a high-conviction position. If ETH drops to $1,500 and the NAV discount widens to 50%+, it becomes more interesting as a small, speculative position. The real catalyst would be WYFI's NC-1 facility coming online and BTBT stopping equity dilution.
| Demand Boost | 10/10 | AGI needs exactly what WYFI provides: GPU compute and data center capacity |
| Margin Expansion | 4/10 | Hardware/hosting margins are constrained by physics and competition |
| Strategic Assets | 8/10 | Power contracts, Tier-3 facilities, Cerebras relationship, GPU fleet |
| Disruption Risk | 2/10 | Physical infrastructure is hard to disrupt -- AGI needs it, doesn't replace it |
| Innovation Risk | 3/10 | Low risk of obsolescence -- data centers and GPUs are the compute layer |
BTBT/WYFI is positioned at the physical infrastructure layer of the AI stack. AGI increases demand for compute, which increases demand for data centers, which increases demand for power and cooling. As long as intelligence runs on silicon and silicon needs electricity, companies like WYFI benefit. The ETH strategy adds a separate, uncorrelated bet on programmable money infrastructure.