Applied Digital (APLD)
Next-gen AI data center operator purpose-built for HPC GPU clusters, emerging as one of the most structurally tight AI infrastructure plays in the market.
1. Core Product / Service
Applied Digital (NASDAQ: APLD) designs, builds, and operates high-performance computing (HPC) data centers purpose-built for AI training and inference workloads. Unlike traditional colocation providers (e.g. digital-realty, Equinix) that serve general enterprise IT, Applied Digital constructs facilities specifically optimized for NVIDIA GPU clusters with:
- High-density power: 50-100+ kW per rack, compared to 5-10 kW for traditional data centers
- Direct liquid cooling: Warm-water and rear-door heat exchanger cooling for H100/B200 GPU clusters
- Fiber-dense networking: Redundant low-latency backbone connections for distributed AI training
- Large-scale campus designs: 100 MW+ greenfield data center campuses on owned land
Applied Digital operates three business segments:
- HPC Data Centers — The core AI infrastructure business. Purpose-built facilities designed for large-scale GPU deployments. This segment is growing rapidly as hyperscalers and AI labs seek new capacity.
- Cloud Services — Managed GPU cloud services offering on-demand H100/B200 compute for AI workloads.
- Bitcoin Mining (legacy) — The company originated as a cryptocurrency miner and maintains some mining operations, but this segment is being de-emphasized as the HPC AI business takes center stage.
The company's flagship project is the Ellendale HPC Data Center campus in North Dakota (initially planned as a Bitcoin mining site, converted to HPC AI), alongside facilities in Garden City, South Dakota and other locations.
2. Target Users & Pain Points
Primary users:
- AI hyperscalers (Microsoft, Google, Amazon) needing to rapidly expand GPU capacity beyond their owned footprints
- Large AI labs (deepseek, Mistral, Anthropic, xAI) requiring massive GPU clusters on short timelines
- Enterprise AI teams needing HPC capacity in regions where hyperscaler supply is constrained
Pain points solved:
- Power availability crisis: New AI data centers require 100-500 MW, but interconnection queues for the US grid are 4-7 years long. Applied Digital pre-secures power and land, offering "shovel-ready" sites
- GPU supply constraints: When NVIDIA GPU allocations are scarce, Applied Digital provides turnkey clusters
- Construction speed: Traditional data center construction takes 3-5 years; Applied Digital can deploy modular AI data centers in 12-18 months
- Cooling complexity: Most legacy data centers lack liquid cooling infrastructure for H100/B200; Applied Digital builds for it from day one
3. Competitive Landscape
| Company | Strategy | Capacity (GW pipeline) | Key Differentiator |
|---|---|---|---|
| Applied Digital (APLD) | Greenfield HPC DC builder | ~1.2 GW pipeline | Speed to power, convert mining sites |
| CoreWeave | Pure-play GPU cloud | ~1.0 GW | Kubernetes-first, fastest provisioning |
| Crusoe Energy | Gas-flare + modular DC | ~500 MW | Stranded energy + modular construction |
| Digital Realty | Global colocation | ~5 GW+ | Largest scale, but legacy infrastructure |
| Equinix | Global colocation | ~4 GW+ | Enterprise connectivity, not AI-optimized |
| Lancium | Power-flexible DC | ~500 MW | Grid-responsive load management |
Applied Digital's key competitive moat is its ability to secure power and land in markets where interconnection is the primary bottleneck. The company also benefits from its "HPC-first" design philosophy versus providers retrofitting legacy space.
4. Unique Observations
Applied Digital is the most structurally tight AI infrastructure play in the market today. Multiple structural drivers converge: (1) US power interconnection queues hit 2.5 TW in 2025 (per LBL Grid Integration Study) making new site development extremely time-constrained; (2) existing GPU service providers (coreweave, Lambda) are running at near-100% utilization; (3) no new greenfield supply is coming online at scale until 2027-2028 due to transformer and switchgear lead times.
In the AI token supply chain, Applied Digital plays at the infrastructure layer — providing the physical substrate for AI compute. Specifically:
- Physical infrastructure: Land, building, power, cooling, networking for GPU clusters
- Capacity intermediary: Connects power utilities (constrained supply) with AI labs (massive demand)
- GPU fleet operator: Manages NVIDIA GPU clusters for hyperscaler tenants
The 13F analysis context is critical: The fund added 18.87% to its APLD position, bringing the total to $319.98M (13,478,438 shares, 2.34% of book). This identifies APLD as one of the stocks most worth continuing to research. FY26 Q3 results emphasized HPC AI DC demand, with management reporting unprecedented leasing interest. The thesis is: AI GPU demand is structurally supply-constrained; Applied Digital is one of the few publicly-traded ways to bet on this physical scarcity.
Key risk: Execution risk on construction timelines and financing. Building multi-hundred-MW campuses requires significant capital; APLD carries relatively high debt and has been dilutive to equity holders. If AI demand growth decelerates, Applied Digital's pre-leased model becomes vulnerable to tenant defaults.
5. Financials / Funding
13F Position (Q1 2026) — Aschenbrenner Fund
- LONG: $319.98M — 13,478,438 shares — 2.34% of book — +ADD 18.87% (increased significantly from Q4 2025)
- Implied price per share: ~$23.74 (based on position value / share count)
- Market cap at entry: ~$10.18B (per Yahoo Finance, Feb 2026)
- Conviction: HIGH — one of the stocks identified as "most worth continuing to research"
- Thesis: Purpose-built GPU DC operator with fastest-to-market power solutions in a structurally supply-constrained market
Company Financials
- Market cap: ~$10B+ (May 2026, up significantly from earlier estimates as AI DC thesis plays out)
- Revenue (FY26 Q3, quarter ending November 2025): $60M+, growing >100% YoY driven by HPC data center ramp
- Net income: Not yet profitable on a GAAP basis; heavy capex depreciation and interest burden
- Debt: ~$500M+ in project-level and corporate debt financing
- Equity raises: Multiple at-the-market (ATM) offerings and private placements to fund construction; significant dilution to early shareholders
- Major funding rounds:
- 2022: Raised $20M+ in Series A/B for Bitcoin mining pivot to HPC
- 2023-2024: Multiple debt facilities totaling >$400M secured against data center assets
- 2025: Raised additional capital via ATM offers and strategic investments to fund Ellendale campus
- Project pipeline: ~1.2 GW of total potential capacity across owned sites
Sources: Applied Digital earnings releases at https://ir.applieddigital.com/ (2026-05-21); company filings; SEC 13F Q1 2026 (Accession 0002045724-26-000008); DaveManuel.com 13F analysis 2026-05-19; EliteCurrenSea 13F update May 2026.
6. People & Relationships
- Wes Cummins — CEO and Chairman. Founded the company; former investment banker. Driving the pivot from Bitcoin mining to HPC AI data centers.
- David Gershowitz — CFO. Managing capital structure and project finance for the AI data center buildout.
- NVIDIA partnership — Applied Digital is a preferred partner for NVIDIA GPU data center deployments; access to preferred GPU allocation.
- Microsoft — Reported to be a tenant for some HPC capacity; Microsoft's AI infrastructure demand creates a natural customer pipeline.
- Core Scientific — Comparable competitor that also pivoted from Bitcoin mining to AI HPC hosting (emerged from Chapter 11 in 2024).