Mariana Minerals

1.1 Identity, Mission, and Operating Model

Mariana Minerals describes itself as a software-first, vertically integrated minerals company focused on supplying the metals needed for AI infrastructure, electrification, and defense systems. The company was founded in 2024, is headquartered in San Francisco, and frames its mission as rebuilding the U.S. critical-minerals supply chain by combining industrial execution with reinforcement-learning software and project orchestration tools. Unlike a software vendor selling mine-tech subscriptions, Mariana owns and operates assets itself: the business model is to acquire, build, restart, or expand mineral projects and improve their economics with MarianaOS. That distinction matters because Mariana is underwriting commodity price, permitting, construction, and operating risk directly rather than capturing high-margin SaaS revenue. The company says its near-term goal is to build ten projects in ten years and that Copper One and Lithium One already use the three MarianaOS modules — CapitalProjectOS, MineOS, and PlantOS — across buildout, mine orchestration, and refining control. Public materials consistently position Mariana closer to a project developer-operator with AI leverage than to a pure mining software startup.[CO001, CO002, CO003, CO004, CO005, CO006]

1.2 Leadership Bench, Talent Footprint, and Key-Person Dependence

The leadership bench supports the thesis that Mariana is trying to solve a hard operational problem rather than merely pitching a concept. Founder and CEO Turner Caldwell previously spent roughly nine years at Tesla, where multiple sources say he worked on battery minerals, recycling, factory design, and lithium-refinery construction. Mariana’s two other named co-founders are CTO Juan Lozano and CFO Baker Tilney; public materials give titles but still provide much less background detail on them than on Caldwell, which is a diligence gap. The broader leadership team on the company’s about page includes a COO, Chief People Officer, commercial/product leadership, plant-engineering, corporate-development, legal, product-development, analytical, and special-projects leads. Hiring data also points to a multi-site operating footprint spanning San Francisco HQ, Houston engineering, Moab operations at Copper One, and technical recruiting in Ann Arbor. That breadth helps the execution story, but Caldwell remains a clear key-person dependency because he anchors the fundraising narrative, the Tesla credibility transfer, and the core "factory mindset for mining" pitch. Public board composition and governance rights were not disclosed in the reviewed sources.[CO009, CO010, CO011, CO012, CO013, CO014]

1.3 Funding History, Investor Mix, and Capital Signaling

Mariana has moved unusually fast on fundraising for a company founded in 2024. Its July 2025 Series A announcement said the round was led by Andreessen Horowitz with continued support from Breakthrough Energy Ventures and Khosla Ventures, bringing total capital raised at that point to $85 million. Roughly thirteen months later, Mariana announced a $310 million Series B led by Khosla Ventures with participation from a16z, Breakthrough Energy Ventures, BHP Ventures, Mitsubishi Corporation, In-Q-Tel, and a longer list of new financial investors, bringing parent and project capital raised to approximately $400 million. Fortune, The Next Web, AI Weekly, and Discovery Alert all corroborate the $310 million round size and $1.5 billion valuation. The cap table matters strategically: venture firms validate software ambition, BHP Ventures brings mining-industry signal, Mitsubishi suggests industrial-commercial relevance, and In-Q-Tel adds a U.S. national-security angle aligned with the company’s domestic-supply-chain narrative. The tradeoff is that Mariana is now priced like a high-conviction hard-tech platform before public evidence on revenue, margins, or stable production throughput is available.[CO017, CO018, CO019, CO020, CO021, CO022]

1.4 Asset Base, Operating Milestones, and Current Scale

Mariana’s flagship assets are real projects rather than pilots-in-slideware. Copper One, announced in March 2026, is an operating copper mine and hydrometallurgical refinery in southeastern Utah that Mariana acquired in late 2025 from the prior Lisbon Valley team. Company and media sources agree the site includes mine, leach, solvent-extraction, and electrowinning infrastructure and that Mariana wants to integrate scrap feed alongside mined material. The public target is to scale combined output to 50,000 metric tons per year of refined copper, although sources differ on exact timing and current run-rate, so investors should treat the output target as aspirational rather than achieved. Lithium One, announced in October 2025 and backed by a Select Water Solutions partnership, is designed as a GWh-scale produced-water lithium facility in East Texas / western Louisiana with up to 3,000 metric tons per year of battery-grade lithium salts and commercial production targeted for the first half of 2027. Together the assets give Mariana a more concrete industrial base than many critical-minerals startups, but they also expose the company to restart, commissioning, and commodity-cycle risk immediately.[CO025, CO026, CO027, CO028, CO029, CO030]

1.5 Adverse Signals, Contradictions, and Remaining Diligence Gaps

The strongest pushback on the Mariana story is not that the company lacks ambition; it is that public disclosure still lags the valuation. No reviewed source provided audited revenue, gross margin, cash burn, debt, project-level capex, board membership, or the detailed preference stack behind the financing rounds. Independent reporting reinforces that autonomy in mining and unconventional lithium extraction both face hard implementation risks. AI Weekly explicitly cautioned that Copper One production levels and unit economics remain unverified and that autonomous mining has a history of overpromising. The Dallas Fed separately noted that unconventional U.S. lithium projects often face low prices, billion-dollar capex needs, long payback periods, and commercialization slippage. Mariana also presents some metric drift across sources: Copper One is described as ramping toward 50,000 metric tons per year, but only some media specify a by-2030 frame; similarly, media describe Lithium One as Texas-based while company and Fastmarkets materials extend the feedstock and infrastructure footprint into western Louisiana and East Texas. These are manageable ambiguities, but they reinforce that the company should be diligenced as an execution-heavy industrial startup rather than a de-risked software platform.[CO034, CO035, CO036, CO037, CO038, CO039]

1.6 Exhibits

2.1 Market Boundary, Included Spend, and Status-Quo Alternatives

Mariana’s market is best defined as domestic supply of refined copper and battery-grade lithium chemicals rather than “AI for mining.” Copper One is aimed at selling refined copper cathode into U.S. industrial demand, while Lithium One is designed to sell battery-grade lithium salts produced from oil-and-gas wastewater. That means the relevant spend pools are metal and chemical procurement budgets, not enterprise software budgets. The status quo is not another startup dashboard; it is imported refined copper, imported lithium carbonate or hydroxide, conventional mining/refining capacity operated by large incumbents, and in some cases export of U.S. raw or scrap material for foreign processing. This boundary matters because Mariana’s economic proposition depends on delivering physical units with schedule, quality, and cost credibility. The company’s software can only monetize if it changes project economics or supply reliability enough to win real commodity offtake. In practice, Mariana is trying to insert itself into a market where buyers care about price, impurity profile, domestic-source resilience, and continuity of supply more than they care about software novelty.[CM001, CM002, CM003, CM004, CM005, CM006]

2.2 TAM / SAM / SOM Through Copper and Lithium Supply Lenses

The total addressable market is clearly large, but the useful diligence question is how much of it Mariana can plausibly access from its first two sites. On copper, USGS puts 2025 U.S. apparent primary refined copper and old-scrap consumption at 2.2 million metric tons, with 57% net import reliance and 1.7 million tons of refined imports. Using the 2025 U.S. producer cathode price of $4.90 per pound implies a domestic refined-copper demand pool above $20 billion annually. On lithium, USGS pegs global 2025 consumption at 263,000 metric tons lithium content, with an average battery-grade lithium carbonate price of about $9,000 per ton and U.S. net import reliance greater than 50%. Dallas Fed work shows 66 U.S. lithium projects, but only a small number are under construction and many require more than $1 billion of capital. Mariana’s immediate SOM is therefore much smaller: a 50,000 tpa copper target at Copper One and a 3,000 tpa lithium-salts target at Lithium One, with the latter representing a meaningful pilot-commercial wedge rather than a market-shaping scale position. The investment case depends on whether those first units become a replicable platform, not on the first-site volume alone.[CM008, CM009, CM010, CM011, CM012, CM013]

2.3 Buyer / User / Payer Segmentation and Adoption Path

The buyer map differs sharply between copper and lithium. Refined copper is ultimately consumed across construction, electrical/electronics, transport, and industrial machinery, but Mariana’s direct commercial interface is more likely to be traders, rod mills, brass mills, cable producers, and other industrial fabricators. On lithium, the direct commercial target is not the end EV buyer but cathode producers, battery-material traders, OEM procurement teams, and intermediaries that can absorb lithium carbonate into broader battery-material supply chains. Fastmarkets reports that Mariana intends to emphasize lithium carbonate first because it is more fungible across chemistries and that the company is negotiating long-term offtakes with OEMs and intermediaries. That is strategically sensible: a young producer needs flexible channels more than a single locked-in process path. The adoption path for both products still runs through qualification, reliability proof, logistics, and domestic-sourcing credibility. In other words, the market will not reward autonomy narratives on their own; it will reward on-spec tonnes delivered on time.[CM017, CM018, CM019, CM020, CM021, CM022]

2.4 Growth Drivers, Adoption Constraints, and Timing Frictions

Demand-side support for Mariana’s markets is real. IEA analysis shows copper demand rising as cleantech, grids, and broader electrification grow, while strategic-minerals policy is increasingly shaped by resilience concerns and Chinese refining concentration. Mariana’s own domestic-supply thesis therefore sits on a real macro foundation. But the supply side is brutal. IEA still expects copper and lithium deficits through 2035, yet Dallas Fed, Carnegie, CFR, EY, and sector reporting all underline the same constraint set: capital intensity, slow permitting, technical scale-up risk, upstream-to-midstream bottlenecks, and workforce shortages. Mining Technology’s 2025-to-2026 outlook adds that autonomy adoption is spreading but still early, with just over 4% of mining equipment autonomous or autonomous-ready and only a few thousand autonomous haul trucks installed globally. That makes Mariana directionally well-positioned if autonomy lowers cost and schedule, but it does not make the go-to-market easy. The company is competing in markets that are attractive precisely because they are difficult to enter.[CM024, CM025, CM026, CM027, CM028, CM029]

2.5 Market Judgment and Remaining Sizing Gaps

The market takeaway is that Mariana’s addressable opportunity is economically large but operationally narrow in the near term. Copper and lithium both benefit from structural demand, domestic-sourcing policy interest, and refining bottlenecks that should reward reliable U.S. production. At the same time, Mariana’s lithium SOM is tiny relative to the national buildout problem, and its copper target should be viewed as a site-level expansion plan rather than a solution to U.S. import dependence. The company does appear to be aiming at the right choke points — refined units, not just upstream resource claims — and that is important because midstream value capture is where many Western supply chains remain weak. The unresolved market questions are whether buyers will pay a domestic resilience premium, whether Mariana can qualify product quickly enough, and whether its software actually reduces cost and ramp time enough to offset the structural disadvantages of new U.S. supply projects.[CM035, CM036, CM037, CM038]

2.6 Exhibits

3.1 Landscape: direct peers, adjacents, incumbents, and substitutes

Mariana’s competitive set spans several classes that solve adjacent slices of the same supply-chain problem. The direct peer bucket includes AI-native or software-forward critical-mineral developers such as KoBold Metals and, on the lithium side, project-backed DLE companies such as Standard Lithium, Lilac Solutions, and EnergyX. Adjacent competitors include downstream platforms like Redwood Materials that do not mine ore or produced water directly but still compete for strategic relevance, customer attention, and industrial-policy capital by supplying domestic critical materials. The incumbent bucket includes established miners and refiners with existing offtake relationships, large balance sheets, and operational depth. The status quo substitute remains imported copper and lithium chemicals processed outside the United States. This matters because Mariana is trying to beat both a technology field and an industrial field at once: it must look more agile than incumbents while also looking more commercially real than technology-led peers.[CP001, CP002, CP003, CP004, CP005, CP006]

3.2 Direct competitor profiles and what each owns

KoBold Metals is the most obvious private comparison point because it also wraps a software-and-AI story around critical minerals, but the model is different. KoBold’s public materials emphasize exploration science, machine prediction, and very large-scale copper development at Mingomba, where KoBold says output is expected above 300,000 metric tonnes annually in the early 2030s. Redwood Materials sits farther downstream: it recycles batteries, produces lithium, nickel, cobalt, copper, and cathode active material, and deploys energy storage for data centers and the grid. EnergyX is a hybrid technology-plus-project company built around GET-Lit membranes and lithium projects such as Lonestar and Black Giant. Standard Lithium is a near-commercial U.S. lithium developer focused on Arkansas and East Texas, and Lilac is the most disclosed commercially mature DLE benchmark in this source set, with manufacturing, EPCM, and binding offtake milestones. Mariana’s distinctive combination is brownfield copper restart plus produced-water lithium plus mine/refinery autonomy. No single peer covers exactly that blend.[CP007, CP008, CP009, CP010, CP011, CP012]

3.3 Capability, packaging, and commercial proof comparison

Capability breadth is not the same as commercial readiness. Mariana’s public stack spans CapitalProjectOS, MineOS, and PlantOS and is applied to both copper and lithium assets, which suggests stronger cross-site operating ambition than many lithium-only DLE peers. But public commercial proof still lags some competitors. Lilac has disclosed a binding 10-year take-or-pay offtake agreement, Phase 1 capacity, EPCM partner, and U.S. manufacturing line. Standard Lithium has disclosed a binding offtake with Trafigura and a clearer pre-FID milestone map. Mariana, by contrast, has only publicly indicated ongoing OEM and intermediary negotiations for Lithium One. KoBold has larger funding and world-class asset scale but is less obviously focused on brownfield autonomy and hydrometallurgical process control. Redwood has unmatched downstream scale in domestic battery recycling and energy-storage deployment, but its business model is fundamentally different from Mariana’s mine-and-refinery ownership model. Across private peers, exact pricing is rarely public; the more relevant comparison is contract structure, disclosed customer proof, and stage-gate readiness.[CP016, CP017, CP018, CP019, CP020, CP021]

3.4 Moat durability, switching cost, and displacement risk

Mariana’s moat claim rests on three linked ideas: first, software-first execution can restart or build mineral assets faster; second, learning compounds across copper and lithium operations; third, domestic supply-chain urgency creates room for a new entrant. Those are plausible but still contestable. KoBold’s science and capital base could dominate the AI narrative in mining. Lilac and Standard Lithium have already disclosed stronger commercial-readiness signals in lithium. Redwood can capture domestic-supply-chain mindshare without taking greenfield mining risk. Incumbent miners and processors still own the deepest customer relationships, operating teams, and balance sheets. The practical switching cost for customers is also lower than software markets because copper cathode and lithium carbonate are specification-driven commodity-like products once qualified. That means Mariana’s software moat matters only if it lowers cost, improves reliability, or accelerates delivery faster than peers can imitate. The threat is not just one direct rival; it is a field of better-capitalized companies proving adjacent parts of the stack sooner.[CP024, CP025, CP026, CP027, CP028, CP029]

3.5 Competitive judgment and remaining gaps

The competitive picture suggests Mariana is well positioned conceptually but not yet first in disclosed commercial proof. Its direct differentiation against KoBold is brownfield restart and refining control instead of pure discovery/development. Its differentiation against DLE peers is operating-model breadth across copper and lithium rather than lithium-only extraction. Its weakness is that those distinctions remain more narrative than bankable customer evidence in public materials. The most valuable unanswered questions are whether Copper One’s autonomy deployment produces measurable cost/throughput gains, whether Lithium One closes binding offtake and qualification milestones, and whether the company can keep raising capital without losing discipline to better-evidenced peers. For now Mariana looks competitively interesting, but not yet competitively settled.[CP032, CP033, CP034, CP035, CP036]

3.6 Exhibits

4.1 Revenue streams, monetization, and recognition logic

Mariana’s public materials indicate a physical-products revenue model rather than a software-licensing model. Copper One is described as a copper mining and refining asset intended to reach 50,000 metric tonnes per year of combined geologic and scrap-fed output, which implies revenue from copper cathode or other refined copper units sold into industrial channels. Lithium One is positioned as a produced-water lithium facility serving OEMs and intermediaries, with Fastmarkets reporting Mariana is negotiating long-term and spot-style offtake structures. The company markets MarianaOS as the execution layer across capital projects, mine operations, and plant management, but no public source indicates that Mariana sells MarianaOS as standalone software; it appears to monetize through asset ownership and operating performance. That distinction matters because revenue recognition and gross-profit structure should look like industrial commodity supply: realized price, delivered tonnage, recovery rate, and contract mix matter far more than seats or subscriptions. The public source set does not disclose current production revenue, realized copper price, lithium product mix, or any active revenue base. Investors therefore have to treat the model as economically plausible but not yet financially transparent.[CI001, CI002, CI003, CI004, CI005, CI006]

4.2 GTM motion and sales-efficiency proxies

Mariana’s go-to-market motion appears to be slow-cycle, qualification-heavy industrial selling rather than rapid-volume customer acquisition. Copper customers care about product specification, reliability, and delivery, while lithium buyers require qualification and often prefer long-term offtake structures. Fastmarkets says Mariana expects Lithium One to allocate roughly 70% of production under long-term offtakes with the rest sold on spot markets, which is consistent with an industrial contracting model. Public disclosures also point to use of intermediaries in addition to direct OEM relationships, suggesting Mariana may balance bankable long-dated contracts with channel flexibility. The company’s investor roster—Khosla, a16z, Breakthrough Energy Ventures, BHP Ventures, Mitsubishi, In-Q-Tel—likely helps credibility with strategic counterparties, but it does not substitute for signed customer proof. There is no public data on sales cycle length, customer-acquisition cost, payback, channel rebates, or renewal economics because the company is not selling software. The best public proxy for sales efficiency is milestone progression: Copper One restart speed, Lithium One construction progress, and the pace at which counterparties convert from dialogue to binding offtake.[CI008, CI009, CI010, CI011, CI012, CI013]

4.3 Cost structure, gross-margin drivers, and missing unit economics

The likely cost stack is legible even though the actual numbers are not. Copper One’s economics will be driven by mining productivity, heap leach and SX-EW recovery, refining throughput, energy, reagents, maintenance, labor, and logistics. Lithium One adds produced-water handling, extraction media or chemical inputs, water-treatment infrastructure, and conversion costs before any battery-grade product can ship. Brownfield and partner-based choices should reduce some upfront spending relative to greenfield analogs: Copper One leverages an acquired idle site and refining package, while Lithium One pairs with Select Water around produced-water infrastructure. But lower relative capex does not mean low absolute capex. Peer and partner benchmarks remain large: Select Water alone guided to $250-$290 million of 2026 net capex and reported $277.8 million of liquidity at June 30, 2026, underscoring the scale of infrastructure spending that produced-water systems can require. Publicly disclosed unit-economics fields for Mariana—cash cost per pound copper, lithium recovery rate, realized price, gross margin, working-capital cycle, and site-level utilization—are all unavailable, making underwriting impossible without management materials.[CI014, CI015, CI016, CI017, CI018, CI019]

4.4 Capital adequacy, financing dependency, and next-round risk

Mariana’s public financing story is strong at the headline level and weak at the operating-detail level. The company announced a $310 million Series B in August 2026 at a $1.5 billion post-money valuation, bringing total capital raised to about $400 million. Those funds are described as supporting Copper One, Lithium One, and continued MarianaOS buildout. That is a credible balance-sheet starting point for an industrial startup, especially with strategic investors on the cap table. It is not obviously enough to fully build, ramp, and derisk two mineral assets plus a software-and-automation organization if commercial timelines slip. Public analogs reinforce the point: Standard Lithium’s 2026 disclosures show customer offtake, permitting, engineering, and financing milestones remain critical even for a single-commodity DLE developer; Select Water’s filings and quarterly update show that water-infrastructure businesses can consume hundreds of millions of dollars in annual capex. Mariana has disclosed no cash-on-hand figure, no monthly burn, no debt, no project-finance package, and no explicit next-round trigger. The likely conclusion is that Mariana will need either asset-level debt, strategic project finance, more equity, or some combination before both flagship assets are fully scaled.[CI022, CI023, CI024, CI025, CI026, CI027]

4.5 Financial verdict and diligence blockers

The financial case for Mariana is directionally attractive but not yet underwritable from public evidence alone. Positively, the company appears to be monetizing real industrial outputs in markets with strategic demand, it has secured unusually strong investors for a private mining startup, and it is choosing operating models that may compress development timelines relative to greenfield peers. Negatively, nearly every underwriting-critical field is missing: current revenue, shipment volume, realized pricing, gross margin, EBITDA, burn, cash balance, debt covenants, project-finance assumptions, and contract backlog. The most important diligence blockers are whether Copper One can actually deliver a low-cost 50,000 tpa profile, whether Lithium One closes binding customer contracts and reaches commercial production in 2027, and whether the capital stack can absorb inevitable schedule volatility. Until those items are evidenced, Mariana should be treated as a promising but financially opaque industrial buildout, not as a de-risked operating company.[CI029, CI030, CI031, CI032, CI033, CI034]

4.6 Exhibits

5.1 What Mariana actually delivers

Mariana does not appear to sell generic mining software into a broad external customer base. Instead, its public materials frame the company as an owner-operator of critical-minerals assets whose proprietary operating system improves how projects are built, mines are run, and refining plants are controlled. The named software layers are CapitalProjectOS for capital delivery, MineOS for mine operations, and PlantOS for refinery management. These are paired with physical assets: Copper One in Utah and Lithium One in Texas. In practical workflow terms, Mariana is selling future copper and lithium output produced from a tighter human-machine operating loop, not simply code. This distinction is important because the product’s utility must be judged by safety, throughput, recovery, cost, and deployment speed—not by seats, APIs, or UI polish alone. Public sources suggest the company sees software and industrial execution as inseparable.[CE001, CE002, CE003, CE004, CE005, CE006]

5.2 Module, asset, and workflow map

The module map visible from public sources is richer than the homepage alone suggests. CapitalProjectOS is described as a single source of truth across planning, engineering, procurement, cost control, scheduling, and construction management. MineOS monitors and orchestrates decisions across geology, fleet, haulage, maintenance, and workforce coordination. PlantOS is oriented toward refining performance, including control of heap bioleaching, solvent extraction, electrowinning, and lithium process steps. Copper One functions as the main public proof site for combined software-and-operations claims, while Lithium One extends the stack into produced-water lithium extraction and process development. The careers surface reinforces the same picture: Mariana is staffing software development, process development, mining operations, plant-and-process engineering, and site roles in parallel. This is not a thin software overlay on someone else’s mine; it is a full-stack industrial operating model.[CE007, CE008, CE009, CE010, CE011, CE012]

5.3 Architecture, autonomy, and critical dependencies

The public architecture is directional rather than fully specified, but it is detailed enough to infer the main layers. Mariana says MarianaOS uses simulation, predictive maintenance, and reinforcement learning to improve mine and refinery workflows. The Series B announcement adds a more explicit control-loop description for PlantOS and says lab-scale experiments and kinetic models connect to a world model that exchanges control signals and sensor feedback with the plant. At Copper One, TechCrunch says Mariana tapped Pronto to help automate haul trucks and drill rigs. Pronto’s own materials describe a turnkey autonomy stack using computer vision, GNSS, rugged camera/GPS hardware, machine-learning object detection, a retrofit drive-by-wire kit, and site-specific safety rules with little infrastructure modification. This indicates Mariana’s architecture depends not only on internal software but also on specialized autonomy partners, sensors, skilled site personnel, and robust process-control engineering. The dependency chain is wide: failure in safety systems, partner integration, plant instrumentation, or data quality can all weaken the stack.[CE014, CE015, CE016, CE017, CE018, CE019]

5.4 Deployment maturity, roadmap, and validation limits

Mariana’s technology maturity is uneven but legible. Copper One is the most mature operating proof point because it is an active site and the company has publicly described specific autonomy and control ambitions there. Lithium One looks more like a scaling and process-development program approaching commercial production in 2027. The company’s hiring profile strongly implies that substantial core systems remain under active construction: multiple open roles target machine learning, data engineering, full-stack software, mechatronics, pilot-plant operations, metallurgy, drilling engineering, and process controls. That breadth is a good signal for seriousness, but it also means the technology stack is still being assembled and industrialized. Public proof of outcome remains limited. Mariana claims meaningful uplift in mining and refining operations, but no source provides audited KPI deltas for throughput, recovery, downtime, or safety. Product maturity should therefore be treated as partially deployed and internally promising rather than independently validated.[CE022, CE023, CE024, CE025, CE026, CE027]

5.5 Trust, safety, privacy, and quality controls

Trust and quality controls are the least mature public surface. Pronto emphasizes safety systems suitable for mine operators, including access controls, activity notifications, monitoring, utilization metrics, and all-stop interactions with manually operated equipment. Mariana’s privacy policy confirms the company has at least a formal web privacy regime, but it does not substitute for operational safety, cybersecurity, or industrial-quality certifications. No public source in the reviewed set shows named ISO certifications, mine-safety performance dashboards, incident disclosures, status-page reliability data, or independent assurance around autonomy. This does not prove the controls are absent; it means buyers and investors cannot verify them from public evidence. Given that Mariana’s wedge depends on software controlling physical industrial processes, the diligence bar on safety and control quality should be high. The correct conclusion is that trust posture is directionally serious but still under-disclosed.[CE029, CE030, CE031, CE032, CE033, CE034]

5.6 Exhibits

6.1 Who the customers, users, and payers probably are

Mariana’s customer map is atypical because the company sits between industrial software and physical commodity supply. The external payers that matter economically are likely copper buyers, battery and cathode supply-chain participants, OEMs, and intermediaries able to take lithium product under long-term or spot contracts. The day-to-day users of MarianaOS are internal operators, engineers, and process teams at Copper One and Lithium One rather than outside enterprise-software users. In the current public record, investors remain the near-term economic payers for the buildout while end customers are still mostly future contracted buyers. This means the segmentation problem has to separate buyer, user, payer, and proof source. Mariana also has a strategic-counterparty layer that is not the same as end demand but still matters to commercialization: Select Water provides infrastructure validation for Lithium One, and Pronto provides deployment validation for autonomy at Copper One. Public evidence therefore supports a tiered segmentation: future commodity buyers, current operating users, enabling commercial partners, and strategic capital providers who may open doors but do not themselves prove recurring product demand.[CU001, CU002, CU003, CU004, CU005, CU006]

6.2 Adoption trajectory and what counts as proof today

Because Mariana has not published customer counts or active revenue, the best public adoption markers are milestone-based rather than volume-based. Copper One shows real-world operating deployment: Mariana publicly describes an operating mine and refinery where MineOS, PlantOS, and Pronto-enabled autonomy are being pushed into production settings. Lithium One shows infrastructure commitment: Mariana and Select Water broke ground on a commercial produced-water lithium extraction facility, which is stronger than a slideware announcement even if buyer contracts remain undisclosed. Fastmarkets adds the clearest direct buyer signal by reporting Mariana is in discussions with OEMs and intermediaries and expects a majority of output to sit under long-term offtake structures. None of these facts amount to repeat-purchase, NRR, or account-scale evidence. They do, however, show the company is moving from pure story to counterparties willing to allocate equipment, infrastructure, and negotiation effort around the projects. Public adoption should therefore be called early and strategically meaningful, but not yet broad or contractually mature.[CU008, CU009, CU010, CU011, CU012, CU013]

6.3 Named proof: what counterparties actually confirm

The best named proof in the public record comes from counterparties rather than from end customers. Select Water’s own investor-relations site confirms the Lithium One groundbreaking and frames the facility as a commercial produced-water lithium extraction project in Texas. TechCrunch and Pronto confirm that Mariana is using Pronto to automate haul trucks and drill rigs at Copper One, while Pronto’s own materials describe the capability stack and mine-operator orientation of the system. These are meaningful proof points because they show external organizations are committing to Mariana’s operating model. But they are not the same as disclosed product buyers or long-term offtake customers. The strongest missing proof is a named copper off-taker or named lithium buyer. Strategic investors such as Khosla, a16z, Mitsubishi, and BHP Ventures provide credibility and potential channel adjacency, yet investor presence cannot be counted as customer adoption. The right evidence-quality ranking is therefore: named operating or infrastructure counterparty proof is real; named end-buyer proof is still mostly absent.[CU015, CU016, CU017, CU018, CU019, CU020]

6.4 Retention, repeat usage, expansion, and concentration risk

There is no public retention dataset for Mariana. No NRR, GRR, renewal, churn, cohort, or satisfaction metric is disclosed, which is unsurprising for a still-ramping industrial platform. The most useful proxy for retention today is continued partner commitment and milestone progression: Select Water’s broader 2026 disclosures show it continues to invest in water infrastructure and won new contract work, while Pronto continues marketing mining autonomy as an active operating business. These facts do not prove Mariana-specific renewals, but they do reduce the risk that the counterparties are ephemeral. Expansion logic is plausible: if Copper One proves low-cost, autonomy-enabled production, Mariana can add more buyers or more sites; if Lithium One reaches commercial production with a long-term offtake base, buyer expansion could follow. Concentration risk remains high. Initial commercial success will likely depend on a small number of large industrial counterparties whose procurement cycles are long and whose bargaining power is significant. The most realistic customer outcome over the next phase is not a diversified customer count but a narrow set of high-value contracts.[CU022, CU023, CU024, CU025, CU026, CU027]

6.5 Customer verdict and diligence blockers

The public record suggests Mariana is past the point of being a purely conceptual customer story, but it is not yet at the point of proven customer durability. There is credible operating proof through Copper One, credible infrastructure proof through Select Water, and credible buyer-intent signals via Lithium One offtake discussions. What is missing are the fields most investors care about for commercial confidence: named buyers, signed offtake contracts, repeat purchase behavior, concentration disclosure, and satisfaction or retention evidence. Mariana’s customer risk is therefore not lack of market need but lack of disclosed commercial closure. The most important diligence asks are simple: name the buyers, show the contract terms, quantify qualification progress, and separate operating partners from actual customers. Until then Mariana should be treated as pre-broad-adoption with strategically meaningful but still incomplete customer proof.[CU029, CU030, CU031, CU032, CU033, CU034]

6.6 Exhibits

7.1 Regulatory and legal risk

Mariana’s regulatory exposure spans mine operations, refining, water handling, and any industrial autonomy deployment layered on top. Lithium One’s produced-water model sits in a water-and-oilfield regulatory context where EPA Class II injection-well rules, state-level water and disposal regimes, and environmental permitting all matter. Copper One and any future mine expansion sit within federal and state mining-law frameworks and mine-safety oversight. These exposures are manageable in principle, but they are non-trivial because Mariana is trying to accelerate timelines in sectors where permits, environmental controls, and operating rules can create delays or cost inflation. The public record does not surface litigation or enforcement against Mariana, which is good, but it also does not provide a permit matrix, incident history, or an environmental and safety compliance dashboard. Mariana’s posted privacy policy helps only at the web-governance layer; it does not answer the more important industrial legal questions around site safety, environmental obligations, or operational assurance.[CR001, CR002, CR003, CR004, CR005, CR006]

7.2 Operational, technical, and quality risk

Operationally, Mariana is doing several difficult things at once. Copper One requires continuous mine-and-refinery coordination, feedstock variability management, and a credible path from restart narrative to materially higher output. Lithium One requires scaling a produced-water lithium process into dependable commercial output in a market where even focused peers still face engineering and financing gates. Mariana’s software and autonomy claims increase upside but also create failure modes: bad instrumentation, model drift, weak human override design, or poor partner integration can harm throughput, quality, or safety. Public sources describe simulation, predictive maintenance, reinforcement learning, and control loops, which is specific enough to take seriously but not specific enough to verify independently. The core operational risk is therefore not that the company lacks a plan; it is that the plan touches many tightly coupled systems with limited public KPI proof.[CR009, CR010, CR011, CR012, CR013, CR014]

7.3 Partner and dependency risk

Mariana’s model depends on more counterparties than a simple “own the asset” story implies. Select Water is strategically important to Lithium One’s infrastructure logic, and Pronto is important to the autonomy deployment story at Copper One. Strategic investors may help with future financing or commercial introductions, but they can also create expectation risk if milestones slip. End-customer dependence is likely to be concentrated when it eventually becomes visible: a small number of anchor offtakes could dominate revenue, making counterparties powerful. The good news is that the current partner set looks credible and industrially relevant. The bad news is that concentrated, high-value partners can become chokepoints. Public evidence does not disclose substitution options, termination terms, or fallback plans if a key counterparty changes strategic direction. That makes dependency risk material even before any obvious counterparty stress appears.[CR017, CR018, CR019, CR020, CR021, CR022]

7.4 Financial and model risk

Financial risk remains high because public visibility into burn, cash, debt, and project-finance strategy is low while capital needs are plainly large. The company has raised meaningful capital, but it is advancing two industrial projects plus a software-and-automation build. That combination is expensive even when milestones are met; it becomes much more expensive if schedule slips or qualification takes longer than planned. Commodity exposure adds another layer. Copper and lithium markets may be strategically attractive, but they are still volatile and cyclic. Mariana’s story also relies on autonomy and software improving economics, yet public evidence does not show the size or durability of that uplift. Financial-model risk therefore includes both classic capex risk and a subtler proof-risk: if the software benefit is smaller than hoped, Mariana could end up looking like a capital-hungry miner with a thinner moat than the current valuation assumes.[CR025, CR026, CR027, CR028, CR029, CR030]

7.5 People risk, mitigations, and thesis-break triggers

People risk is visible in the hiring pattern. Mariana is recruiting across machine learning, process controls, mechatronics, metallurgy, drilling automation, pilot-plant operations, and site execution. That breadth supports the ambition of the model, but it also shows the company still depends on assembling and retaining a rare multidisciplinary team. Mitigants exist: strategic capital access, brownfield copper positioning, credible partners, and a mission aligned with industrial policy. Still, the most important monitors are straightforward. If Lithium One fails to translate dialogue into binding commercial contracts, if Copper One cannot show measurable operating improvement from the software stack, if regulatory or safety issues delay execution, or if additional financing arrives on punitive terms, the thesis weakens quickly. Mariana’s risk posture is therefore investable only for backers willing to monitor milestones closely and revise views as evidence arrives.[CR033, CR034, CR035, CR036, CR037, CR038]

7.6 Exhibits

8.1 Investment thesis and anti-thesis

The bull thesis is coherent. Mariana sits in a strategically important market, owns or controls real asset proof points, and has a differentiated software-plus-operations story that is more concrete than many industrial-AI startups. Copper One gives it a public operating site, Lithium One gives it a high-upside domestic lithium wedge, and the investor base suggests the company has attracted serious conviction. The anti-thesis is equally clear: almost every underwriting-critical field remains private. The public record does not show revenue, margins, signed end-buyer contracts, burn, or an auditable capital plan. That means the current valuation rests more on a forward option value than on de-risked operating performance. Mariana may deserve a premium to a purely conceptual software company or a pre-asset slideware startup, but it does not yet deserve to be treated like a proven industrial operator. The key valuation question is therefore not whether the story is interesting; it is whether $1.5B already discounts too much of the future proof curve.[CV001, CV002, CV003, CV004, CV005, CV006]

8.2 Recommendation, confidence, and valuation stance

Recommendation: Research More / Wait at current price. Confidence: Medium. Risk rating: High. Valuation stance: Full to rich on disclosed evidence. This is a price-sensitive call, not a dismissal of the company. Mariana could absolutely grow into or beyond the current mark if Copper One produces measurable operating gains, if Lithium One signs bankable contracts and hits the 2027 timeline, and if financing remains available on reasonable terms. But public evidence today does not justify high-conviction underwriting at $1.5B. Existing insiders or strategic investors may rationally hold because they likely have access to private operating data and because the optionality is real. A new investor without privileged diligence should treat the current price as demanding relative to disclosed proof. The most likely near-term improvement in the call would come from signed offtakes, site-level KPI disclosure, or a financing structure that shows scale-up is fundable without punitive dilution.[CV009, CV010, CV011, CV012, CV013, CV014]

8.3 Financing context, comparables, and entry discipline

The cleanest public price anchor is the August 2026 Series B itself: $310 million raised at a $1.5 billion post-money valuation and roughly $400 million total capital raised. That is a serious but not absurd valuation for a private company trying to own both software optionality and critical-minerals assets. The problem is that the comp set cuts both ways. Mariana is already worth more than public DLE names like Standard Lithium and Lithium Americas on a market-cap basis, despite those companies offering more public disclosures and, in some cases, clearer offtake and permitting milestones. On the other hand, Mariana is far smaller, earlier, and less proven than major listed material producers or royalty companies such as Albemarle, Freeport-McMoRan, Royal Gold, Franco-Nevada, and Wheaton Precious Metals. Entry discipline therefore has to ask what exactly the premium is paying for. The answer is a blend of domestic supply-chain relevance, software-option value, and multi-asset upside—not current financial proof. That makes the mark understandable, but also fragile if milestones slip.[CV016, CV017, CV018, CV019, CV020, CV021]

8.4 Bull, base, and bear scenarios

The scenario range is unusually wide because Mariana is simultaneously an operating story, a financing story, and a technology-premium story. In the bull case, Copper One demonstrates real software-led economics, Lithium One secures anchor customers and remains on track for 2027 production, and additional capital arrives on reasonable terms; that can support a valuation well above the current round. In the base case, proof arrives slowly, the company remains interesting, and the current mark proves roughly fair but not obviously cheap. In the bear case, customer proof slips, autonomy benefits remain narrative-heavy, or financing becomes more punitive; that would imply a lower fair value than the Series B. Because public data on economics is thin, these scenarios are best interpreted as discipline tools rather than precise price targets. They clarify what evidence must arrive to justify upside and what failures would compress valuation quickly.[CV024, CV025, CV026, CV027, CV028, CV029]

8.5 Exit readiness, thesis-breaks, and final diligence asks

Mariana is not public-market ready on current public evidence. It may be institutionally financeable in private markets because strategic investors can underwrite longer-duration industrial options, but an IPO-quality record would require much more operating disclosure. The most likely future positive exit paths are: a later-stage private round at improved proof, a strategic transaction once one or both assets are materially de-risked, or an eventual public listing after customer, capex, and operating metrics are clearer. The most important diligence asks are direct and familiar from the earlier chapters: current cash and burn, signed or near-signed buyers, site-level unit economics, permit status, partner terms, and measured Copper One / Lithium One performance deltas. The thesis-break triggers are also clear: no credible customer closure around the 2027 lithium target, no measurable software-driven operating uplift, regulatory or safety disruptions, or punitive next-step financing. Until those are resolved, the valuation call should stay evidence-sensitive and cautious.[CV031, CV032, CV033, CV034, CV035, CV036]

8.6 Exhibits