Part of the Technology sector
Core investment principles and frameworks for this industry
Many US AI compute names began as bitcoin miners. Mining gives optionality in weak hosting demand but adds commodity volatility, halving risk, and treasury exposure that should be separated from contracted data-center revenue.
GPU clusters and data centers require heavy upfront capex. A sound thesis must connect financing sources, collateral, customer prepayments, and expected returns before underwriting growth.
Hosting contracts should be judged by duration, take-or-pay terms, credit quality, escalation clauses, and customer concentration. Headline megawatts can overstate value if utilization or counterparty quality is weak.
AI compute infrastructure is constrained by power, land, interconnection queues, and cooling, not just servers. Sites with secured low-cost power and grid approvals have a stronger moat than companies only announcing GPU purchases.
Compute assets compound only when utilization, uptime, and power efficiency remain high. Idle GPUs, delayed energization, or poor fleet management can destroy economics quickly.
Active trends shaping the industry landscape
Access to leading GPUs can drive near-term revenue, but refresh cycles and depreciation can compress returns. Track whether contracts pass through hardware obsolescence risk.
Data-center load growth is drawing scrutiny from utilities, regulators, and local communities. Interconnection delays and rate-case outcomes can become material bottlenecks.
Hyperscalers and AI labs are leasing external capacity to bridge power and data-center constraints. The trend is durable only if external providers can meet reliability, security, and cost requirements.
Demand is moving from training bursts toward persistent inference workloads. This can improve utilization but pressures cost per token, latency, and power efficiency.
Bitcoin miners are repurposing sites for high-performance compute. Conversion success depends on power density, fiber, cooling, customer certification, and balance-sheet capacity.
Events and factors that could trigger significant change
For hybrid miners, bitcoin price and network difficulty can affect cash generation and funding capacity. Separate mining windfalls from repeatable AI infrastructure economics.
Revenue inflects when GPUs are installed, certified, and utilized. Track delivery timing, cluster acceptance, and uptime, not only purchase orders.
A signed AI hosting contract with a credible counterparty can validate capacity economics. The key details are MW committed, ramp schedule, pricing, capex responsibility, and termination rights.
Utility approval, energized capacity, or substation completion can unlock revenue. These catalysts matter more than announced pipeline that lacks grid certainty.
Debt, lease financing, customer prepayments, or joint ventures can de-risk capex. Dilutive equity raises without matched contracted revenue should be treated cautiously.
Critical financial and operational metrics for evaluation
Capex per MW and capex per deployed GPU reveal whether projects can earn adequate returns. Rising build costs can turn strong revenue demand into weak equity returns.
Track contracted MW, energized MW, and revenue-generating MW separately. Pipeline that is neither contracted nor energized should receive a steep probability discount.
For hybrid names, hashrate, realized bitcoin price, energy cost, and network difficulty should be isolated from AI hosting metrics to avoid mixing two different businesses.
Power cost, curtailment exposure, and demand charges are core unit economics. Compare fixed-rate power contracts against merchant exposure and local grid congestion.
GPU utilization, data-center uptime, and customer SLA performance determine realized economics. Low utilization can erase returns even when market demand is strong.
Mawson Infrastructure Group Inc. - Common Stock
NASDAQ:MIGINASDAQ
MIGI
Bitfarms Ltd. - Common Stock
NASDAQ:BITFNASDAQ
BITF
IREN Limited - Ordinary Shares
NASDAQ:IRENNASDAQ
IREN
Hut 8 Corp. - Common Stock
NASDAQ:HUTNASDAQ
HUT
TeraWulf Inc. - Common Stock
NASDAQ:WULFNASDAQ
WULF
Applied Digital Corporation - Common Stock
NASDAQ:APLDNASDAQ
APLD
Riot Platforms, Inc. - Common Stock
NASDAQ:RIOTNASDAQ
RIOT
BitMine Immersion Technologies, Inc. Common Stock
AMEX:BMNRAMEX
BMNR
Cipher Mining Inc. - Common Stock
NASDAQ:CIFRNASDAQ
CIFR
Core Scientific, Inc. - Common Stock
NASDAQ:CORZNASDAQ
CORZ
GDS Holdings Limited - American Depositary Shares
NASDAQ:GDSNASDAQ
GDS
MARA Holdings, Inc. - Common Stock
NASDAQ:MARANASDAQ
MARA
Bitdeer Technologies Group - Ordinary Shares
NASDAQ:BTDRNASDAQ
BTDR
CleanSpark, Inc. - Common Stock
NASDAQ:CLSKNASDAQ
CLSK
VNET Group, Inc. - American Depositary Shares
NASDAQ:VNETNASDAQ
VNET
HIVE Digital Technologies Ltd - Common Shares
NASDAQ:HIVENASDAQ
HIVE
American Bitcoin Corp. - Class A Common Stock
NASDAQ:ABTCNASDAQ
ABTC
WhiteFiber, Inc. - Ordinary Shares
NASDAQ:WYFINASDAQ
WYFI
Bit Digital, Inc. - Ordinary Share
NASDAQ:BTBTNASDAQ
BTBT
SuperX AI Technology Limited - Ordinary Shares
NASDAQ:SUPXNASDAQ
SUPX
TSS, Inc. - Common Stock
NASDAQ:TSSINASDAQ
TSSI
Soluna Holdings, Inc. - Common Stock
NASDAQ:SLNHNASDAQ
SLNH
Hyperscale Data, Inc. Common Stock
AMEX:GPUSAMEX
GPUS
The9 Limited - American Depository Shares
NASDAQ:NCTYNASDAQ
NCTY
Argo Blockchain plc - American Depositary Shares
NASDAQ:ARBKNASDAQ
ARBK
Greenidge Generation Holdings Inc. - Class A Common Stock
NASDAQ:GREENASDAQ
GREE
BTC Digital Ltd. - Ordinary Shares
NASDAQ:BTCTNASDAQ
BTCT
Bit Origin Limited - Class A Ordinary Shares
NASDAQ:BTOGNASDAQ
BTOG
Get AI analysis for AI Compute & Data Infrastructure companies
Management credibility, business model strength, growth catalysts, and risk assessment with exact page citations.
Get started free