NUAI — knowledge base

Overview

NUAI is an early-stage developer transitioning from natural-gas and helium activities into AI data-center and power infrastructure. Its strategy, branded “bring your own power” (BYOP), is to use behind-the-meter or islanded natural-gas generation to bypass congested utility-interconnection queues and accelerate time-to-power.

Its principal proposed development is Texas Critical Data Centers (TCDC), a claimed 438-acre West Texas campus between existing Vistra and Calpine power plants with access to three natural-gas transmission pipelines. The site, land rights, neighboring facilities, pipeline capacity, permits, and development rights remain unverified without filings and project documents. [[s:74@00:34:01]]

The described phasing has changed or remains internally inconsistent. In December 2025, management described approximately 450 MW of IT load in phase one and expansion beyond 1 GW in phase two. By June–July 2026, investors instead described a 207 MW Vistra-connected first phase and a separate 450 MW islanded-generation second phase. No disclosed engineering plan or company filing reconciles these configurations.

Ownership claims have also evolved. Management stated in December 2025 that all development within the 438-acre TCDC property was held through a 50-50 joint venture with Sharon AI, while NUAI’s subsequently announced second asset was wholly owned. By June 2026, speakers claimed NUAI had removed its former 50% partner and wholly owned TCDC; that later change remains unverified.

Management said TCDC was under exclusivity with one prospective tenant and was negotiating a long-term lease while engineering continued. The intended model is primarily triple-net data-center leasing, with a separate partner expected to own and finance the generation assets. Definitive tenant agreements, final investment approval, financing, permits, equipment orders, and construction remain outstanding. [[s:74@01:12:56]]

The bull case rests on scarce power, abundant Permian gas, reportedly suitable land and pipelines, modular construction, asset-level financing, and potential delivery before conventionally interconnected projects. The bear case is that almost every major source of project value—ownership, power rights, tenant commitment, gas capacity, permitting, financing, schedule, and attributable economics—still requires documentary confirmation.

Key facts & figures

Global data-center electricity consumption was approximately 413–415 TWh in 2024 and is projected by major energy-industry estimates to exceed 900 TWh by 2030. [[s:37@00:08:03]]
ERCOT supplies most Texas electricity demand and is largely isolated from the two other major North American interconnections, limiting imports during shortages and potentially increasing the strategic value of local generation. [[s:37@00:17:05]]
The claim that ERCOT is a roughly 105 GW system with approximately 200 GW of demand backlogged was misleading: installed capacity can exceed 100 GW, but served peak load is lower, while queue totals vary by date, category, duplication, and project viability. [[s:74@00:23:43]]
Management described TCDC as 438 acres between an approximately 1.1 GW Vistra plant and a 550 MW Calpine plant. The acreage, land control, adjacency, facility capacities, and associated rights are unverifiable. [[s:74@00:34:01]]
Updated project configuration: In December 2025, management targeted approximately 450 MW of IT load in phase one and more than 1 GW after phase two. In June–July 2026, investors described approximately 207 MW in phase one and 450 MW in phase two. The discrepancy remains unresolved.
The later 207 MW configuration was described as using an adjacent Vistra gas plant through a bidirectional grid connection, allegedly approved under ERCOT’s Batch Zero process and subject to Cipher or hyperscaler option rights. The approval, option ownership, transferability, and contractual mechanics remain unverified. [[s:37@00:22:12]]
A behind-the-meter arrangement does not universally require generation and load to share one substation or eliminate all transmission charges. Economics depend on physical configuration, tariffs, utility treatment, and regulation. [[s:37@00:19:58]]
Management described three pipelines serving the site and claimed access to 100 MMcf/d from ONEOK, 200 MMcf/d from Enterprise, and another 100 MMcf/d from Whistler WhiteWater. These volumes are unverifiable without interconnection, transportation, and capacity agreements. [[s:74@00:40:00]]
Approximately 80 MMcf/d of natural gas could plausibly support about 400 MW of generation at roughly 40% electrical efficiency. [[s:74@00:40:18]]
Phase-two generation was later described as approximately 450 MW of islanded capacity using modular natural-gas reciprocating engines. Equipment, balance-of-plant design, fuel access, permits, and financing remain unverified. [[s:37@00:24:06]]
Management proposed multiple smaller parallel generating units, approximately 20% spare capacity, and short-duration ride-through batteries to provide data-center-grade reliability. Actual reserve requirements depend on equipment, electrical architecture, maintenance, battery duration, black start, and tenant standards.
Claims that reciprocating engines require roughly 20–25% excess capacity versus about 50% for traditional turbines are misleadingly precise. Modularity can improve unit-level redundancy, but there is no universal reserve requirement. [[s:37@00:32:18]]
Reciprocating engines are generally less efficient and more carbon-intensive per MWh than combined-cycle gas plants, although smaller units can be installed incrementally and may offer faster replacement or maintenance isolation.
Management reportedly reserved generation equipment. The manufacturer, equipment type, quantity, deposits, delivery dates, cancellation rights, and whether any reservation became a binding order remain undisclosed.
Management targeted an approximately 18-month construction cycle and commercial operations in the second half of 2027. This schedule is unverifiable and contingent on definitive leases, financing, permits, equipment, engineering, construction, and commissioning. [[s:74@00:53:07]]
Modular prefabricated designs were claimed to reduce on-site construction labor by approximately 80%. The direction is plausible, but the project-specific percentage is unverifiable without design and vendor evidence. [[s:74@01:17:05]]
Management’s intended commercial model is a long-term triple-net lease under which the tenant bears specified operating expenses, while a partner owns and finances generation. Exact cost allocation will depend on lease, power-purchase, availability, escalation, and service-level terms.
Project construction is intended to be funded primarily through asset-level debt and project equity supported by tenant lease revenue, minimizing parent-company equity issuance. Financing has not been documented as committed.
Management claimed approximately $125 billion was raised for data-center project finance in 2025 and that 2026 capacity was effectively committed through the third quarter. The claim is unverifiable because no dataset or consistent definition of project finance was provided. [[s:74@01:03:54]]
Speakers later claimed a Macquarie credit facility exceeding $200 million. Its existence, borrower, collateral, draw conditions, covenants, availability, and connection to TCDC remain unverified. [[s:37@00:37:52]]
Waha Hub natural-gas prices have periodically traded below zero because Permian production exceeded takeaway capacity or infrastructure was constrained. [[s:37@00:24:34]]
The statement that notable negative Waha pricing began in 2016 was inaccurate; widely documented sustained negative-price episodes began in 2019, although earlier localized negative transactions may have occurred. [[s:74@00:26:57]]
Negative Waha prices do not mean free delivered fuel. Gathering, treatment, transportation, firm-capacity, interconnection, and contracting costs can remain substantial. [[s:37@00:25:02]]
The assertion that approximately 95% of global helium is associated with natural gas was directionally sound but misleadingly precise; nearly all commercial helium comes from helium-bearing gas streams, but the exact percentage depends on classification. [[s:74@00:29:49]]
The claim that gas-engine air permits can be obtained through a simple 90-day process was inaccurate. Timing depends on aggregate emissions, equipment, location, and state and federal requirements. [[s:37@00:33:56]]
Closed-loop cooling can sharply reduce water consumption but does not require literally no water after filling; maintenance and heat-rejection losses remain.
Large, concentrated GPU campuses favor tightly interconnected training clusters, while distributed facilities can better serve latency-sensitive inference. TCDC’s intended workload will determine its fiber, network, cooling, and reliability requirements. [[s:24@00:27:36]]
HBM supplied principally by SK Hynix, Micron, and Samsung has constrained AI-accelerator production alongside foundry and advanced-packaging bottlenecks. A completed campus may be underutilized if tenant GPU deliveries are delayed. [[s:24@00:29:30]]
The statement that ChatGPT-3 launched in November 2023 and triggered recent power demand was inaccurate: GPT-3 was released in 2020, ChatGPT launched in November 2022, and GPT-4 Turbo was announced in November 2023. [[s:74@00:22:20]]
Comparing 1 GW directly with Denver’s electricity consumption is misleading because power is a rate and city demand varies by season, boundary, and whether average or peak load is measured. [[s:74@00:22:58]]
Claims that CoreWeave exceeds 40 US data centers, 250,000 active GPUs, 3 GW of contracted power, and $60 billion of backlog were unverified and should not be used as validated NUAI comparables. [[s:24@00:09:16]]
Apollo acquired a controlling interest in Stream Data Centers, but the claim that Apollo paid approximately $50 billion was inaccurate and likely conflated the transaction with broader Apollo assets or investment capacity. [[s:37@00:45:10]]
Characterizing Macquarie’s Applied Digital arrangement as automatically funding all equity for only 15% of every project over roughly 70 years was misleading because it omitted staged commitments, conditions, preferred returns, and governance rights. [[s:37@01:05:39]]
Energy Dome’s CO₂ storage was discussed as a possible long-duration solution, but claims that it is definitively much cheaper than lithium-ion and broadly provides approximately 24-hour duration are misleading without project-specific evidence. [[s:37@00:52:39]]

Thesis & bull case

AI infrastructure is creating secular demand for power-dense sites, making available power—not merely buildings—the critical development constraint: “if it's not going to be delivered by the grid, it has to be off grid.”
NUAI’s BYOP strategy could bypass utility queues by pairing a data-center load directly with dedicated natural-gas generation, potentially reducing time-to-power if fuel, equipment, permits, financing, and tenant requirements are secured.
A 438-acre site between two power plants and near three major pipelines could provide valuable land, gas, electrical, and infrastructure optionality if NUAI or TCDC possesses enforceable rights.
Multiple pipelines could improve fuel redundancy and supplier competition. Claimed capacity substantially exceeds the approximately 80 MMcf/d considered sufficient for 400 MW, although proximity does not establish firm delivery rights.
Staged development could limit initial capital requirements. Depending on the final configuration, TCDC may begin with either a 207 MW grid-related tranche or approximately 450 MW of dedicated generation before expanding beyond 1 GW.
The claimed 207 MW approved load could remove a major interconnection hurdle if the approval exists, remains valid, is transferable, and is controlled by TCDC.
A bidirectional connection could preserve optionality to import power, coordinate on-site generation, or provide grid services, subject to actual ERCOT, utility, Vistra, and tariff arrangements.
West Texas offers abundant natural gas and episodically negative Waha pricing. Dedicated generation could monetize constrained gas while avoiding transmission bottlenecks, though underwriting should use delivered and normalized fuel costs.
Smaller parallel generation units can be installed incrementally and isolate maintenance or failure at individual units. Ride-through batteries could bridge short interruptions while other units start or synchronize.
Reportedly reserved generation equipment could mitigate long lead times if reservations are binding, correctly specified, and supported by deposits and delivery commitments.
Modular prefabricated data halls could reduce on-site labor, shorten schedules, improve quality control, and create a repeatable design for later TCDC phases or additional NUAI sites.
A long-term triple-net lease with a creditworthy tenant could create predictable cash flow, contractual escalators, and financeable lease revenue while transferring specified operating costs to the tenant.
Keeping generation in a separately financed partner entity could reduce NUAI’s capital burden and isolate fuel and power-asset risks, provided power pricing and availability obligations remain commercially viable.
Asset-level debt and project equity could limit parent-company dilution by matching financing to contracted project cash flow rather than issuing NUAI common shares for the entire build.
The reported exclusive prospective-tenant process could validate demand and allow engineering to converge around a real customer’s density, redundancy, cooling, network, and delivery requirements.
A power-and-colocation model limits direct exposure to GPU obsolescence. NUAI could retain AI-infrastructure upside without purchasing accelerators or operating a neocloud.
Training customers may value several hundred contiguous megawatts for dense, low-latency clusters. If TCDC delivers suitable fiber and reliability, West Texas concentration could be an advantage.
Stream Data Centers was presented as an execution partner with development experience and hyperscaler relationships. Apollo’s ownership interest adds institutional signaling, although neither Stream’s obligations nor Apollo financing should be assumed without contracts.
Odessa/Midland stakeholders were described as supportive because the project could diversify the economy, increase construction and operating activity, and create demand for Permian gas.
Management’s Permian energy background may help with land, gas, pipeline, and local relationships, while infrastructure-development experience may support lease negotiation, project financing, and construction.
A binding lease with a creditworthy hyperscaler or similar tenant would be the most important rerating event because it could simultaneously validate customer demand, site design, power requirements, financing feasibility, and schedule.
If TCDC succeeds, modular BYOP campuses could become a repeatable development model rather than leaving NUAI dependent on one speculative asset.

Risks & bear case

Development-stage risk: NUAI has not yet demonstrated that it can finance, build, commission, and lease a hyperscale data-center campus.
Configuration uncertainty: Management’s December 2025 description of a 450 MW first phase conflicts with later descriptions of a 207 MW first phase followed by 450 MW. Investors cannot reliably value capacity until the company publishes a definitive configuration.
Ownership risk: TCDC was described as a 50-50 Sharon AI joint venture in December 2025, while later speakers claimed NUAI removed the partner and wholly owns the project. The transaction, consideration, retained rights, and present economics remain unverified.
Land and site risk: The 438 acres, land-control structure, boundaries, neighboring plants, easements, zoning, and development rights require documentary confirmation.
Power-rights risk: The claimed 207 MW Batch Zero approval, Vistra connection, bidirectional rights, and Cipher or hyperscaler option may not be owned, transferable, exercisable, or sufficient for the proposed load.
Tenant risk: Exclusivity is not a lease. The prospective tenant may terminate negotiations, change requirements, demand concessions, or select a competing project.
Concentration risk: Exclusivity with one tenant can improve coordination but also gives that counterparty leverage over pricing, milestones, design changes, termination rights, and required equity.
Contract-structure risk: Headline lease value can obscure free rent, tenant-improvement obligations, construction milestones, availability guarantees, service credits, termination rights, pass-through limits, and parent guarantees.
Final-investment-decision risk: Engineering and lease negotiations can continue without producing a board-approved, funded project.
Financing risk: Asset-level debt generally requires a bankable lease, permits, fixed-price or capped construction arrangements, sponsor equity, and substantial conditions precedent.
Dilution risk: If lease-backed project finance is unavailable or delayed, NUAI may need common equity, convertibles, warrants, preferred securities, or project-level equity carrying senior economic rights.
Generation-partner risk: The intended partner must finance, construct, fuel, operate, and guarantee the power assets. Failure to secure that partner could shift significant capital and operating obligations back to NUAI.
Macquarie-facility risk: The reported facility exceeding $200 million remains unverified, and headline capacity may differ from cash available after collateral, covenants, milestones, and borrowing-base limits.
Schedule risk: An 18-month build and second-half-2027 commercial operation target may be aggressive given outstanding leases, financing, permits, equipment, gas interconnections, electrical infrastructure, data halls, fiber, testing, and commissioning.
Equipment risk: A reservation is not necessarily a binding order. Generation units, transformers, switchgear, batteries, cooling systems, and network equipment have separate lead times and integration risks.
Fuel-capacity risk: Claimed pipeline volumes do not establish firm transportation, pressure, gas quality, meter-station capacity, interconnection rights, or delivery during extreme weather.
Fuel-price risk: Negative Waha pricing is episodic. New pipeline capacity, production changes, maintenance, and basis normalization could materially increase delivered fuel costs.
Reliability risk: A fully islanded campus must address spinning reserve, black start, maintenance outages, frequency control, synchronization, battery duration, and multi-contingency events. A simple 20% spare-capacity assumption may not meet tenant requirements.
Environmental risk: Large reciprocating-engine fleets can produce substantial CO₂, NOx, noise, methane-related exposure, and local air-quality impacts that conflict with customer sustainability objectives.
Permitting risk: There is no guaranteed 90-day pathway for several hundred megawatts of gas generation. Aggregate emissions could trigger more complex state or federal review.
Cooling and water risk: Closed-loop systems reduce but do not eliminate water use. West Texas heat raises peak cooling loads, parasitic power consumption, and performance risk.
Fiber risk: A West Texas site requires sufficient carrier diversity, route diversity, bandwidth, latency, and construction commitments for hyperscale workloads.
Construction risk: An 80% on-site-labor reduction is unverified and does not eliminate factory capacity, transport, foundations, utility work, site integration, skilled commissioning, or quality-control risks.
Semiconductor risk: Buildings and power do not guarantee occupancy if tenants cannot obtain GPUs, HBM, advanced packaging, or network equipment. [[s:24@00:29:56]]
GPU-obsolescence risk: If NUAI moves beyond leasing into compute operations, accelerator rental prices and residual values can decline rapidly as newer hardware arrives. [[s:24@00:31:01]]
Workload-fit risk: Training, inference, enterprise colocation, and disaster recovery require different network, latency, cooling, security, and reliability designs.
Partner-signaling risk: References to Stream, Apollo, Macquarie, Vistra, Calpine, Cipher, pipeline operators, or a confidential tenant do not establish contractual commitments.
Comparable-financing risk: Applied Digital or other developer transactions may include preferred returns, staged funding, governance rights, and project dilution that make headline capital commitments poor NUAI valuation anchors.
Competitive risk: Hyperscalers, utilities, established data-center developers, neoclouds, and former Bitcoin miners are competing for the same customers, equipment, contractors, financing, gas, and power.
Valuation risk: Applying a per-MW multiple to conceptual capacity can ignore ownership percentages, unbuilt phases, financing seniority, lease economics, customer credit, development costs, and fully diluted shares.
Transition risk: The helium-to-AI pivot leaves NUAI with limited data-center operating history and possible legacy assets, liabilities, or spending obligations.
Promotional-claim risk: Several management and speaker statements—including AI chronology, ERCOT queue size, Waha history, construction labor savings, and financing-market capacity—were inaccurate, misleading, or unverifiable, increasing the need for primary documents.

Timeline of developments

2025-12-03: Management presented BYOP as NUAI’s core strategy and described TCDC as a 50-50 joint venture with Sharon AI covering a claimed 438-acre West Texas property. It targeted approximately 450 MW of IT load in phase one, expansion beyond 1 GW, an 18-month build, second-half-2027 operations, an exclusive prospective tenant, triple-net leasing, separately financed generation, modular construction, and asset-level financing intended to limit dilution; leases, financing, final approval, and execution remained outstanding. [[s:74]]
2026-04-14: Broader AI-infrastructure discussion characterized the buildout as an early-stage supercycle in which hyperscalers, model developers, neoclouds, and former Bitcoin miners compete for power, land, GPUs, memory, and construction capacity. It reinforced the potential value of power-ready sites while highlighting financing, hardware supply, contract structure, utilization, and execution as decisive constraints. [[s:24]]
2026-06-18: Investors described a changed or differently defined TCDC configuration: approximately 207 MW connected to Vistra in phase one, a proposed 450 MW islanded gas-engine phase two, and possible expansion toward 1 GW. They also claimed NUAI had removed its former 50% partner, reserved engines, engaged Stream, and obtained a Macquarie facility above $200 million; these claims remained unverified. [[s:37]]
2026-07-05: Investors again framed NUAI as an AI data-center and power-infrastructure transition centered on TCDC, emphasizing Waha economics, modular construction, Stream/Apollo/Macquarie involvement, project-level financing, management credibility, and Odessa/Midland support while identifying dilution, permitting, option ownership, and delivery timing as unresolved.
2026-07-21: No NUAI-specific developments discussed.
2026-08-04: No NUAI-specific developments discussed. [[s:115]]

Open questions

What entity currently owns TCDC, and what percentage of fully diluted economics is attributable to NUAI?
Was the December 2025 Sharon AI 50-50 joint venture terminated, acquired, or restructured, and what consideration or retained rights apply?
How does the original 450 MW phase-one design relate to the later 207 MW phase one and 450 MW phase two?
Does the announced wholly owned second asset refer to a separate campus, and what land, power, gas, customer, or development rights does it possess?
Does NUAI own, lease, or option the claimed 438 acres, and what surveys, deeds, easements, and development restrictions apply?
Are the neighboring Vistra and Calpine facilities correctly identified, and what contractual relationship does TCDC have with either operator?
Is the 207 MW load approved, reserved, or energized, and what ERCOT, utility, Vistra, or Batch Zero records establish its status?
Who owns the Cipher or hyperscaler-related option, when does it expire, what does exercise cost, and is it assignable?
What imports, exports, or self-generation does the proposed bidirectional connection allow, and which transmission or ancillary charges remain?
Has the prospective tenant signed only exclusivity, or also an LOI, capacity reservation, development agreement, lease, guarantee, or binding offtake?
What are the tenant’s credit quality, required capacity, workload, lease term, escalators, milestones, termination rights, guarantees, and service-level requirements?
What conditions must be satisfied before the tenant agreement becomes definitive?
Is NUAI targeting powered land, powered shell, turnkey colocation, joint-venture compute, or any direct neocloud operations?
Is the generation fleet fully islanded, behind the meter but grid-connected, or capable of switching between both modes?
Who will own, finance, operate, maintain, and dispatch the power-generation assets?
How will power prices, fuel costs, capital recovery, availability penalties, and emissions obligations be allocated between the generation owner, TCDC, and tenant?
Which generation units have been reserved, from what manufacturer, in what quantity, and under what deposits, delivery dates, cancellation rights, and warranties?
What are the fleet’s heat rate, emissions, installed cost, maintenance cycle, black-start capability, and required reserve margin?
What battery power, energy duration, control system, and ride-through events are contemplated?
Do ONEOK, Enterprise, and Whistler WhiteWater have sufficient available firm capacity, pressure, gas quality, and interconnection rights at the site?
Are the claimed 100/200/100 MMcf/d volumes contracted, or merely estimates based on nearby pipelines?
What are the contracted fuel price, transportation cost, basis exposure, curtailment rights, and extreme-weather protections?
What exact development, engineering, construction, operating, and financing obligations does Stream have?
Is Apollo involved beyond its controlling interest in Stream?
What is the borrower, size, collateral, maturity, availability, and current draw status of the reported Macquarie facility?
How much sponsor equity must NUAI contribute before project debt funds, and what project-level dilution or preferred returns are expected?
Which air, water, noise, construction, and local approvals have been obtained?
What is the realistic permitting and construction critical path for second-half-2027 operations?
What cooling system will be used, and what are peak power consumption and annual water requirements?
Which fiber carriers, routes, latency profiles, and redundancy commitments are available?
Can the modular design substantiate the claimed 80% reduction in on-site labor?
If NUAI supplies compute, who purchases GPUs and bears delivery, utilization, obsolescence, and residual-value risk?
Does management intend to retain TCDC, sell it after development, or recapitalize it with infrastructure investors?
What commitments are required to scale beyond 1 GW?
What helium and legacy natural-gas assets, liabilities, obligations, and expenditures remain?

Notable predictions to track

TCDC will secure a definitive lease with the prospective tenant that was under exclusivity in December 2025. [[s:74@01:12:56]]
TCDC will complete construction in approximately 18 months and achieve commercial operations in the second half of 2027. [[s:74@00:53:07]]
Management will reconcile the December 2025 450 MW phase-one design with the later 207 MW/450 MW phased configuration.
NUAI will document whether TCDC remains a 50-50 Sharon AI joint venture or became wholly owned.
Management was said to be targeting an October 2026 milestone linked to the Macquarie credit facility; this remains unverified.
The claimed 207 MW first phase will retain valid and transferable power rights after resolution or exercise of the Vistra/Cipher option.
NUAI will document the reported Macquarie credit facility exceeding $200 million and demonstrate that funds are drawable for project development. [[s:37@00:37:52]]
A separate partner will finance and own the generation assets, allowing NUAI to avoid funding the full power plant at the parent level.
Construction will be financed mainly with lease-backed asset-level debt and project equity, limiting NUAI common-equity issuance.
Reserved generation equipment will become binding orders with delivery dates compatible with the targeted operating schedule.
TCDC will secure firm gas-interconnection and transportation rights sufficient for approximately 450 MW of initial generation.
The claimed three-pipeline configuration will provide genuine redundancy during maintenance, curtailment, and extreme-weather events.
Modular prefabrication will materially shorten construction and reduce on-site labor, though the claimed approximately 80% reduction remains to be demonstrated. [[s:74@01:17:05]]
TCDC will demonstrate hyperscaler-grade reliability using parallel generation, approximately 20% spare capacity, and ride-through batteries.
Stream will become a documented development or operating partner, with defined obligations and economics.
TCDC will scale beyond 1 GW without disproportionate parent-company dilution.
A creditworthy long-term tenant will enable refinancing and predictable escalated lease cash flow rather than reliance on speculative compute revenue.
Waha gas economics will provide a durable delivered-power advantage rather than a temporary benefit from episodic negative benchmark pricing. [[s:37@00:24:34]]
TCDC will achieve materially faster time-to-power than conventionally interconnected ERCOT projects without compromising reliability or environmental compliance.
NUAI will use TCDC as a repeatable template for additional modular BYOP data-center projects.