In the pre-AI software era, technology was defined by minimal capital expenditures and negative working capital. Customers paid annual contracts upfront and renewed on predictable cycles, creating a constant source of cash to fund growth. Low interest rates and near-zero marginal costs let companies scale even further with additional capital.

Now, developers of AI applications and services as well as frontier hardware technology must commit to compute capacity, integration work, hardware inventory, and specialized labor months before a project is accepted. A startup might prove itself with blue-chip customers, only to find momentum hampered by a lack of suitable growth capital. This can force companies to reduce their scope of ambition or seek relatively expensive equity.

Where working capital swallows cash, but project variables are known, there is a gap that can be funded at a lower level of risk (and thus a lower cost of capital). Yet lending has been limited to capped venture debt and asset-based financing against general equipment or receivables.^[1]

For companies at the leading edge, credit should be seen as a growth catalyst and a critical input rather than a component of post-growth enterprises. Moving from a breakout to durable business requires building trust with customers, equity investors, AND lenders in parallel. Over time, we believe these financing relationships can form a real competitive advantage [Balance Sheet Era].

DEFINING THE UNIT OF PRODUCTION

To make their business legible to a lender, companies must establish a specific unit of production assignable to a customer. This is often a contract or a deployment. The buyer’s requirements may vary, but each unit should have at least a defined scope of work, a start date or delivery date, and a set of acceptance criteria. The cash outflows from each unit can be split three ways: upfront costs (hardware, deployment labor), recurring operating costs, and variable costs that swing with utilization or supply. Separating these unlocks creative ways to pull cash inflows forward to match the real timing of expenses. Normalizing these parameters across units allows lenders to underwrite unit-level returns.

Critically, this shifts the underwriting from the startup’s own balance sheet to the creditworthiness of its customers, whether a single large buyer or a diversified network of them. This financing structure also decouples risk from the resale value of specialized hardware.^[2] As companies prove their capabilities to customers, their ability to raise non-dilutive growth capital is determined by:

1. Customer: customer’s credit attractiveness and the length and nature of the relationship

2. Operating history: demonstrated ability to deliver acceptable units over time

3. Contract terms: the specific provisions in each unit that make it more or less financeable, chiefly the tenor and payment structure (take-or-pay, minimum commitments, prepayments), the customer’s termination and cancellation rights, the acceptance criteria that define successful delivery, and the lender’s step-in rights and ability to claim the contractual cash flows in a downside scenario

Articulating these factors proactively helps lenders get comfortable with both the operating business and the paying customer. While venture capitalists are usually underwriting upside, creditors are principally concerned with capital preservation. Exceptional companies will translate their performance into credibility with both kinds of investors.

TRADING ECONOMIC POINTS FOR FINANCING

Suppose a breakout business is capital constrained and has the choice between the following options for one unit of production with the same delivery cost and acceptance criteria for a given creditworthy customer.

· Situation 0: $100 TCV, where the buyer pays at net-30 terms after delivery and has the option to terminate with one week of notice

· Situation 1: The same structure as Situation 0, but the company factors the invoice upon issuance

· Situation 2: $90 TCV, with a take-or-pay structure, a 5% prepayment, and lender financing at 15% loan-to-value at 15% cost of debt

· Situation 3: $90 TCV, with a take-or-pay structure, a 20% prepayment, and lender financing at 50% loan-to-value at 10% cost of debt

In this simplified example, Situation 0 earns the most overall, but risks the most company cash. In Situation 1, while invoice factoring pulls cash forward by 30 days and helps start relationships with lenders, pricing can be expensive for early-stage companies.^[3] In Situations 2 and 3, exchanging a discount on list price for upfront cash and optimal contract structure allows the unit to become financeable. With prepayment and debt covering an increasing share of the upfront costs, the difference can be used to begin the next unit of production earlier. At scale, contracting under favorable terms recycles equity faster, allowing for multiple units to be produced concurrently. This accelerated growth enables companies to capture benefits that drive unit economic improvements critical to unlocking a sustainable venture scale business [Future Fundamentals].

View the full analysis here.

CRAWL, WALK, RUN – THE ARC OF FINANCEABILITY

In practice, a rising startup does not get to choose between these scenarios. It must earn its way to favorable contract terms and credit financing by proving execution. This is the basis of the crawl, walk, and run. In the ‘crawl’ phase, customer agreements often look like Situation 1. Cash is pulled forward with traditional receivables or equipment financing. Most of today’s frontier technology companies are here. As startups deliver for well-capitalized customers, they can ‘walk’ to economics like those of Situation 2. Credit can finance the subset of contracts with the highest counterparty quality and longest performance history. Founders should expect this to come at a fraction of contract value at fairly expensive terms. Lenders are likely to require supplier backstops and parent guarantees to limit losses. But by the ‘run’ phase, companies have an installed base of creditworthy customers and a growing backlog. Most of these contracts can be financed at an increasing percentage of contract value and a decreasing cost of capital. Parent guarantees could be lifted subject to tenor and performance. To graduate from crawl to walk to run, it is paramount for companies to communicate their operating performance well before needing credit. A relationship with history helps lenders to vouch to their investment committees with confidence.

THE ARC OF FINANCEABILITY, APPLIED

Today, there are a few examples of the arc at scale. Enterprise AI companies that bundle hardware deployments into software subscriptions have secured vendor financing and non-dilutive growth capital in lieu of turning to traditional lending or public equity markets.^[4] Neoclouds with years-long contracted backlogs have been able to raise debt at a lower cost of capital, even as fully loaded contribution margins have compressed significantly.^[5] Energy developers and advanced manufacturers have raised larger sums of venture debt and revolvers but have yet to finance contracted capacity or dedicated deployments.^[6]

Smaller startups are also beginning to move up the curve. Data collection startups supporting frontier AI labs are forming relationships with multi-asset managers who can transition more easily from equity underwriting to lending.^[7] In compute, financing for scaled GPU deployments has largely been confined to complex structures on the supplier’s side.^[8] However, with model serving platforms owning more capacity, enterprise adoption growing, and compute offtake markets maturing, we see opportunities for lending on the purchaser’s side. On the application side, we anticipate vertical AI startups with usage-based billing to move toward contract structures that resemble those of the underlying model API providers.^[9] As these business models crystallize across the stack, we expect more opportunities for non-dilutive capital to fund committed floors (given the need for contract terms that establish minimum economics and are financeable), while equity funds the remaining variable costs on top.

The Arc of Financeability presents an opportunity for accelerated scaling where founders retain greater ownership, equity investors continue to price uncertainty, and credit investors are repaid with contracted cash flows. We’re working with companies to design deployments around financeability from the outset. In parallel, we’re engaging with lenders on when they envision entering these markets, what kind of performance history they’re willing to underwrite, and which types of third-party credit support are most attractive. We believe collaboration between operators and investors will be crucial to transforming the frontier.

^[1] Stifel Bank’s terms in 2024 size venture debt at 25-30% of the prior equity fundraise, September 12, 2024. Astranis’s $155M delayed-draw credit facility concurrent with a $300M Series E, May 6, 2026. Hadrian’s $360M revolver concurrent with a $1.37B Series D, August 14, 2026. Heron Power’s $60M credit facility following a $140M Series B in February 2026, September 10, 2026. [1] [2] [3] [4]

^[2] The defaulted loans for certain Bitcoin miners serve as a historical failure mode for directly financing highly specialized hardware. In 2022 the resale price of mining rigs fell by 82-87% across every generation, and equipment loans that had been written at 8-12% over roughly two years against the rigs alone were impaired within eighteen months: Iris Energy defaulted on $105M of NYDIG loans held in two special purpose vehicles in November 2022, and Core Scientific handed 27,403 machines to NYDIG in February 2023 to extinguish $38.6M of a $77.5M loan because “the value of the machines is lower than the outstanding principal.” [1] [2] [3] [4]

^[3] Factors typically advance 70-90% of an invoice and release the balance on collection less a fee. US factors’ advance rates averaged about 85% in 2025. Factoring fees typically run 0.5-4% of the invoice and accrue weekly or monthly while the balance goes unpaid. [1] [2] [3]

^[4] Capchase’s contract financing for Genusys, March 27, 2025. General Catalyst’s growth investment in Motive from its Customer Value Fund, September 10, 2026. [1] [2]

^[5] CoreWeave has closed six GPU-backed facilities in three years. Its $2.3B DDTL 1.0 facility was repriced to SOFR+9.62% in May 2024 (~15%). It closed an $8.5B DDTL 4.0 facility priced at SOFR+2.25% (~6%) in March 2026 (non-recourse, A3/A- rated, making it the first investment-grade GPU-backed financing) and its first publicly syndicated DDTL 5.0 priced at SOFR+4.5% (~8%) in May 2026. Most recently it closed a $2.6B DDTL 5.5 in August 2026 at SOFR+5.5% (~9%). Over roughly the same period, its LTM margin after cost of revenue and technology & infrastructure expense fell from 24% at year-end 2024 to 8% in Q2 2026, while revenue backlog rose from $15B at year-end 2024 to $104B at quarter-end Q2 2026. [1] [2] [3] [4] [5]

^[6] Hadrian’s $360M revolving credit facility syndicated to eight firms, August 14, 2026. Form Energy’s $270M credit facility syndicated to eight firms, September 21, 2026. [1] [2]

^[7] Apollo’s recent equity investment in Mercor, September 16, 2026. [1]

^[8] In early 2026, Fireworks AI agreed to a four-year lease of dedicated capacity from GMI Cloud, a non-investment grade provider founded in 2023, on ~7,000 NVIDIA GB300 GPUs in a 16 MW facility in Taoyuan, Taiwan, with server cost of about $546M. Lease value was not disclosed. To fund the deal, GMI borrowed through a ring-fenced SPV: a NT$13.9B (about $433M) five-year term loan syndicated by CTBC Bank plus a NT$6.4B bridge and a NT$150M revolver, launched July 14, 2026, secured on the GPUs, the SPV’s equity, its cash accounts and the assigned contracts, with debt capped at 55% of the value of the NVIDIA agreement. NVIDIA agreed to lease any unused capacity for up to six years at a pre-agreed price if a customer stops paying, in exchange for a share of GMI’s revenue above that floor, reported at half, and a claim ahead of the banks. About a dozen banks offered roughly twice the amount sought by September 1. This is the first documented syndicated GPU-backed loan in APAC. [1] [2] [3] [4] [5]

^[9] Direct and managed service closed-source model APIs provided by the frontier labs and hyperscalers offer minimum commitments, provisioned throughput, and prepayment structures for customers to reserve capacity. [1] [2] [3]

Ben Thompson originally coined Aggregation Theory^[1] as a way of describing the consolidation dynamics present among businesses with zero marginal production and distribution costs. While Thompson was focused on the internet, AI will bring similar dynamics to markets that extend broadly across the service economy and, eventually, industrials. The result is a growing divide between scaled operators and a long tail of specialists, with the middle becoming increasingly unstable. These industries are on the cusp of structural change, and those who position capital strategically ahead of it will capture a disproportionate share of the value created.

The First Wave of Aggregation

Internet era aggregation shifted the markets from supply constrained to demand driven.

“[…] the best aggregators win by providing the best experience, which earns them the most consumers/users, which attracts the most suppliers, which enhances the user experience in a virtuous cycle.” — Ben Thompson, Aggregation Theory (2015)

The internet shifted power to platforms that owned the customer relationship through superior product experiences. By consolidating demand and enhancing the experience (in these cases by attracting suppliers), it created a virtuous cycle that pulled in more users and led to significant, and often winner-take-all, concentration.

The pattern played out repeatedly: Google captured ~84% of search, Meta ~42% of social ad spend^[2], Uber ~74% of US ride-hailing, and DoorDash ~65% of food delivery. What unites them is that once marginal costs approached zero, the customer relationship became the only defensible asset, with scale a reinforcing moat.

The Core Premise: Aggregation Offline

Industry concentration historically maps closely to the scalability of the underlying variable cost structure. Where the dominant costs are fixed (platform infrastructure, capital equipment), economies of scale drive concentration. Where the dominant costs are variable and labor-intensive, fragmentation persists, with each new customer requiring proportionally more people, limiting any single firm’s ability to pull far ahead.

Internet platforms, where marginal costs are effectively zero, are highly concentrated. Software is meaningful but less complete; distribution historically required enterprise sales forces rather than pure network effects. As you move into services and industrial markets, real constraints take over: labor availability, materials, capital equipment, and physical distribution. These have historically limited any single firm’s ability to scale far ahead of competitors, preserving fragmentation even in large markets. This explains why aggregation is extreme in digital, meaningful in software, and still limited across most services and manufacturing.

AI is now pushing aggregation offline. By compressing production and distribution costs, AI enables businesses to deliver better customer experiences (faster responses, higher quality, lower cost) without proportionally increasing headcount or operational complexity. This changes the nature of scale: growth becomes less tied to headcount, and quality improves with scale rather than degrading.

Those delivering AI-powered experiences will accrue resources that continually widen the gap versus competitors, pulling in more demand and reinforcing advantage in a compounding loop. Unlike Thompson’s original framework, this is not about attracting suppliers to a marketplace. These businesses deliver the product or service directly, but the same dynamics take hold: winners consolidate market share and the middle hollows out.

Services: The First Mover

Cognitive services will shift first as software absorbs knowledge work and reduces dependence on individual expertise. These sectors (insurance brokerage, legal, IT services, accounting, wealth management, etc.) have light capital requirements and cost structures dominated by headcount. Across the cognitive services we track, average SG&A intensity runs ~41% of revenue at the leader level, with some sectors far higher (e.g. agencies at 75%, wealth management at 60%, and insurance brokerage at 55%). The average concentration among the three largest players (CR3) across these nine sectors is just 22%. These are fragmented markets with tens of thousands of firms competing on reputation and relationships rather than scale.^[4]

Physical labor will follow more slowly as the core work is not near-term automatable (we will save a discussion on robotics for another day); however, there is still material cost and labor in sales and operations. Across the physical services, average SG&A runs ~19% of revenue, lower than cognitive services but still meaningful. These costs sit in scheduling, quoting, billing, customer service, accounting, and project management, all functions that are directly compressible by AI. The most capable companies will automate these functions, delivering better service at lower cost and compounding their advantage as they capture share. When a homeowner wants their roof repaired all they want is low cost, high quality, and fast response time. Anything a company can do to compress sales cycles and reduce costs will win them the day.^[5]

A Note on Scale Enabled Network Effects

Historically, service industries lacked same side network effects. Incremental demand increased prices, strained labor supply, and often reduced quality. As services become AI mediated, incremental demand no longer detracts from quality. Instead, it enhances outcomes through data accumulation and the ability to reinvest more dollars into the product while constraining cost bloat in the system.

Manufacturing: Slower but Not Immune

Manufacturing is not monolithic and breaks into three distinct buckets with different concentration profiles. Capital-intensive manufacturing (e.g. motor vehicles, semiconductors, industrial gases, aerospace) is already relatively concentrated, with an average top three share (CR3) of ~49%. Here, massive capex requirements and process scale have historically propelled consolidation; TSMC alone holds 62% of semiconductor fabrication. Process manufacturing (chemicals, refining, paper, plastics) sits in the middle at an average top three share (CR3) of ~26%. The third bucket, discrete and job-shop manufacturing (machine shops, metal fabrication, printing, furniture, food processing), is deeply fragmented, with an average top three share (CR3) of just 10%. Machine shops alone have over 21,000 firms with no single player holding even 1% share. This fragmentation tracks the thesis: where production is labor-intensive and custom, no firm has been able to scale far enough ahead to consolidate.

However, all three buckets stand to benefit from the dynamics we have described. Even in capital-intensive sectors where concentration is already high, AI compresses SG&A, supply chain management, and engineering labor. The opportunity is largest in discrete manufacturing, where quoting, scheduling, CAD programming, and sales are all directly automatable. There is an aggressive push in some pockets of this sector among rollups and new platforms, with AI-transformed operations attempting to quote faster, program jobs more efficiently, and optimize scheduling to meaningfully compress costs while improving turnaround. The same dynamics reshaping services will take hold here: cost compression enables better customer experiences, and scale reinforces advantage.^[7]

The Future State: Consolidation

Aggregation may manifest differently across sectors, but the directional shift is consistent. As AI compresses the labor and operational constraints that historically limited scale, demand can concentrate around the best providers without triggering the quality degradation and capacity bottlenecks that previously kept markets fragmented.

The race to aggregate comes from multiple directions.

• Scale Incumbents with existing distribution have a natural advantage but face change management difficulties and business model friction (e.g. labor-hour billing in legal or engineering services).

• Mid-market Consolidators backed by VC and PE holdco structures, have been aggressive in the early days, acquiring traditional services firms, then layering AI into operations.^[9] These consolidators are racing to leapfrog incumbents, not just compete with them; it is not hard to imagine a fifth name joining the Big Four in accounting through this model. The funds backing these plays are tightly coupled with the leading AI labs and have access to the best engineering talent in the world, a combination that traditional incumbents cannot easily replicate.^[10]

• Vertically Integrated, AI-Native Providers will have a product and organizational structure advantage but are playing catch-up on distribution and data.

As aggregation expands, many industries move toward a bimodal equilibrium. A small number of scaled operators coexist with a long tail of niche players^[11]. The middle becomes increasingly unstable as neither cost leadership nor specialization is sustainable. This pattern mirrors earlier internet market outcomes but unfolds more slowly and unevenly in the physical world. As capabilities become software mediated, supply side defensibility weakens. Durable advantage shifts toward demand aggregation.

A Tiny Pirate Ship in a Hungry Ocean

We see our opportunity set across three categories: vertically integrated challengers rebuilding industries from scratch, enablers that serve incumbents and emerging aggregators, and platforms that build and serve the long tail.

Vertical Integration: Vertically integrated players start from scratch with small teams, tight product flywheels, and AI at the foundation. They must bootstrap distribution and aggregate data. Over time, some of these companies may acquire traditional firms to access their books of business, particularly in industries with long sales cycles, sticky customer relationships, or regulatory moats where organic customer acquisition is slow. But this is a fundamentally different strategy than building organically and will be difficult to execute simultaneously alongside organic growth.

Enablers: Technology companies serve incumbents and help drive industry wide change. If successful at driving value, these businesses risk creating a highly concentrated customer base as the industries they serve consolidate. To survive, their software/service will need to be inexorably tied to driving value across the full breadth of the business^[12] and may need to find ways to span multiple levels of the industry value chain (e.g. insurance brokers, carriers, MGAs, adjusters, etc.) rather than serving a single layer. As concentration increases among their customers, enterprise value becomes less dependent on broad customer count and more dependent on strategic indispensability to a smaller number of scaled buyers. The best outcome for some may be merging with the leading and most aggressive player in their vertical.

The Long Tail: Aggregation reshapes the long tail rather than eliminating it. Historically, small business software has required distribution advantages that are difficult to build. AI-driven discovery and autonomous agents compresses the cost of reaching and serving fragmented operators, which create durable opportunities to enable or serve this market segment even as the top of the market concentrates. We feel least confident in our ability to project what happens at this part of the market, but individuals are creative and artisanal products and offerings should survive.

Compute, Capex, and Raw Materials

The logical endpoint of these dynamics is a world where business complexity reduces to three inputs: compute, capex, and raw materials^[13]. Every layer of labor, coordination, and distribution that AI can absorb moves an industry closer to that endpoint, and closer to the concentration dynamics that have already played out on the internet. We see the transformation of cognitive services measured in years, physical services in a decade, and manufacturing in a generation. But the direction is consistent, and the companies and investors who position against that arc earliest, before aggregation is obvious and before markets fully reprice, will capture a disproportionate share of the value created.

[1] Ben Thompson, “Aggregation Theory,” Stratechery, July 21, 2015.

[2] Facebook + Instagram

[3] Charts are directional thanks to AI. Actual data via The World of Aggregation Theory

[4] Financial Advisory and Wealth Management: Morgan Stanley’s wealth division generates ~$820K in revenue per employee, yet carries a 60% SG&A load, the vast majority of which is advisory labor, compliance, and client servicing. The industry average is ~$450K per employee[4]. Much of this work (portfolio construction, financial planning, compliance documentation, client reporting) is highly codifiable. An AI-augmented firm could deliver comparable or better advisory output at a fraction of the headcount, compressing the cost structure and reinvesting the savings into client acquisition and product quality.

[5] Landscaping: a $130B market with 630,000 firms and a CR3 of just 3.8%, the most fragmented physical services market we track. The industry average revenue per employee is ~$80K, with SG&A at roughly 18% of revenue at the leader level[5]. On a typical job, labor and materials dominate, but route planning, quoting, scheduling, invoicing, and customer management represent a meaningful share of overhead. An AI-native operator that automates these functions can serve more customers per crew, respond faster, and quote more competitively, creating a compounding advantage in a market where the largest player, BrightView, holds less than 2% share.

[6] Charts are directional thanks to AI. Actual data via The World of Aggregation Theory

[7] Machine Shops: comprise a $55B market with over 21,000 firms and a CR3 of just 1.2%, the least concentrated market in our entire dataset. The industry average revenue per employee is ~$140K with SG&A at roughly 15% of revenue. These are small-run, custom fabrication businesses where quoting, scheduling, CAD programming, and sales coordination consume a disproportionate share of overhead relative to the actual machining work.

[8] Charts are directional thanks to AI. Actual data via The World of Aggregation Theory

[9] Notable examples include Thrive Holdings (Shield Technology Partners in IT services, Crete in accounting), General Catalyst’s Creation fund (Eudia in legal, Accrual in accounting, Titan in IT services, Long Lake in property management and other services), and early experiments from Khosla Ventures, 8VC, and others.

[10] Holdco structures are long-term oriented, enabling sustained investment in AI transformation while the market conveniently provides cost-of-capital arbitrage versus traditional PE fund models.

[11] It is perfectly reasonable to believe that consolidation is the only thing that happens and the long tail of small businesses and individual proprietors are unable to compete. However, the vectors of relationships, regional fragmentation / regulation, and craft/customization provide angles for sustainability.

[12] AI Power Software

[13] The preceding analysis specifically excludes robotics, which will further accelerate these dynamics by compressing physical labor costs, the last major constraint on aggregation in services and manufacturing. We did not talk about it, we know.

Venture investing in operationally and/or capital-intensive businesses requires a different foundation than traditional software or consumer models. At Logos, we use a simple, two-pillar framework for scalable disruption called Future Fundamentals. These pillars are: Quantifiable Value and Unit Economics. Together, they inform our evaluation of opportunities and our work with portfolio companies.

Quantifiable Value

Disruption starts with the delivery of superior value. We view Value = Quality / Cost, measured against incumbents and alternatives.

• Cost – Is the solution outright cheaper, or does it lower the total cost of ownership (upfront investment, ongoing spend, working capital, etc.)?

• Quality – Domain and product specific attributes such as throughput, efficiency, speed, yield, durability, uptime, precision, repeatability, resolution, ease of use, and/or safety. Identifying the most important product quality element(s) is an essential part of the analysis.

We group quality-over-cost advantages into three bands:

SpaceX’s drop in launch cost, Illumina’s reduction in genome sequencing cost, and NVIDIA’s GPU order-of-magnitude better performance per dollar than CPUs all reset industry baselines and ignited markets. Step-change gains, like Tesla’s three-fold reduction in EV sticker price or Astranis’ five-fold cheaper satellites, drive market capture. Incremental gains such as Wolfspeed’s modest EV range bump, Redwood Materials’ estimated 30 percent materials cost reduction, and Desktop Metal’s savings on niche parts have struggled to achieve material market penetration.

To achieve venture-scale growth, a business must deliver value that clearly outweighs the frictions of adoption. While value compared to alternatives is critical, the solution must also target a critical point in the customer’s business, either a primary economic lever or an operational bottleneck – a principle we call process centrality. When strong value is applied at critical choke points, it creates the urgency required for rapid adoption and scale.

Unit Economics

The excitement around capital-intensive businesses has obscured the fact that only those with exceptional capital efficiency can generate venture returns. When each dollar deployed drives disproportionate incremental growth, promising technologies become enduring businesses.

To assess capital efficiency, we look at the fully loaded unit economics of a solution which includes:

• Defining the unit of production (e.g. device, job, customer, site, geography, etc.)

• Determining the revenue per unit

• Attributing all costs to delivering a unit

  • Direct variable costs

  • Amortized fixed asset costs

  • Allocated fixed operational overhead

Strong unit economics drive attractive returns to equity, both in unlevered and levered contexts. They need not exist at inception, but must be reasonably attainable as the business grows. Ideally, fully loaded unit economics should be predictable and demonstrate a path towards 50%+ margins. This enables rapid (<2 years) capital recycling and supports non-linear scaling. Predictable economics also unlocks the potential for financing that further accelerates the cycle times of equity capital. Margins above 30% can still be viable, but often require more equity and lead to greater early-investor dilution.

Understanding both the path and the destination of an economic model is essential. The journey from early unit costs to at-scale economics often follows Wright’s Law, with costs falling and quality improving as cumulative production grows.[1]

Three questions frame scalability: Where is the company on its development curve? How far and how quickly can it move along the curve? Which levers, such as innovation or process advantage, will drive that progress?

Other Considerations

Quantifiable Value and Unit Economics do not live in isolation. Greater value creation accelerates adoption, shortens the sales cycle, lowers CAC, and increases the share of value captured.

Even when a company appears disruptive, there are additional dynamics that separate mediocre businesses from great ones.

Understanding the True Benchmark

A common mistake is benchmarking solutions against the replacement cost of existing systems. The correct comparison is often the marginal cost of the installed base, not the replacement cost. Ignoring this can create the illusion of cost advantage when the solution is uncompetitive. Asset replacement cycles, the useful life of existing systems, and the switching costs or residual value will define the adoption window and inform how the product should be priced relative to incumbent options.

Commodities and the Supply Curve

When a company produces a commodity, long-term defensibility and margin potential rest on being the lowest-cost producer. Competing purely on price typically requires durable cost advantages driven by scale, supply chain efficiency, process innovation, or access to advantaged resources. A strong cost position relative to peers allows a company to stay profitable even in down cycles. On the wrong side of the supply curve, the market sets the price, not the company.

If durable cost leadership is unattainable, the only viable alternative is defensible differentiation. This can come from product features that customers care about, such as superior reliability, lower maintenance needs, faster delivery times, or better integration with existing systems. At Logos, this is a critical part of our assessment of energy investments.

Vertical Integration and the Jobs to be Done

A common failure is to mistake a novel technology for a business model. A breakthrough creates outsized value only when applied to a meaningful profit pool, the point in the value chain where the majority of margin is captured. Branded products, critical components, manufacturers with limited supply, or specialized capabilities often extract outsized margins. We assess:

- Where profit pools live in an industry and how they might shift with the introduction of disruptive technology

- What jobs need to be done to deliver and capture value, and decide where the business must integrate or outsource

The goal is to maximize profitability. Sometimes that means taking on more operational responsibility to ensure performance, utilization, or customer ownership. Other times it means resisting the urge to over-build. This is an ROI question where the incremental investment and effort must be compared to the incremental economics captured by integrating another job/part of the value chain.

Future Fundamentals

Future Fundamentals is not a rigid formula but an operating lens. At Logos, we rely on it twice: first, to assess opportunities during diligence; and again, to engage the founders we back. The framework surfaces the economic and technical levers that determine whether a company can scale in capital‑intensive, real‑world sectors. Hardwiring these fundamentals early helps founders avoid costly course corrections as growth accelerates.

Logos

[1] Nagy, B., Farmer, J. D., Bui, Q. M., & Trancik, J. E. (2013). Statistical Basis for Predicting Technological Progress. PLoS ONE, 8(2), e52669. https://doi.org/10.1371/journal.pone.0052669

h/t Roxanne Tully Baine and Claire Goldsmith

Highlights

1. Instant Context & Memory: Years of nested folders, version histories, and document metadata give an AI assistant deep context on user and corporate workflows. Rather than building a file store from scratch, OpenAI could plug into Dropbox’s opt‑in data pipelines and get immediate access to personal and enterprise documents—fuel for personalized models that remember past conversations, drafts, and analyses. People have loved the memory capabilities recently launched by OpenAI. Imagine those but with the last 15 years of your saved data and work.

2. Built‑In Distribution Engine: 700 million registered users (18 million paid and 575K businesses)

3. Exabyte‑Scale Backbone: Dropbox’s Magic Pocket system operates multi‑exabyte storage across proprietary data centers (handling the bulk of user data) and a global edge network. Acquiring this would spare OpenAI years of capital spending and ops build‑out, letting it focus on product.

4. Ecosystem Neutrality: As a platform‑agnostic hub—integrated with Office, Google Workspace, Slack, Zoom, and hundreds of third‑party apps—Dropbox lets OpenAI’s tools plug seamlessly into diverse corporate stacks, avoiding lock‑in and maximizing reach.

5. Defensive: Preempt other players from securing this strategic asset.

Considerations

• User Trust & Churn Risk

  • Risk: Fears over “OpenAI owning my documents.”

  • Mitigant: Historically, Dropbox users have stayed through acquisitions when new capabilities outweighed concerns. If OpenAI enforces strict opt‑in AI usage, honors existing privacy commitments, and demonstrates clear productivity gains, churn should be minimal—and those who do leave are a small fraction of the newly AI‑empowered base.

• Enterprise Compliance & SLAs

  • Risk: Large corporations worry about data governance.

  • Mitigant: OpenAI would inherit Dropbox’s ISO/SOC certifications, HIPAA/GDPR compliance, and existing SLAs. A clear pledge to maintain these standards—and to operate Dropbox as a standalone trust‑neutral subsidiary—smooths the path.

• Financial & Execution Load

  • Risk: A $8–10 billion price tag and running a global storage business.

  • Mitigant: Dropbox is net‑debt‑zero and generates $700–800 million of free cash flow annually. Its infrastructure is already lean—OpenAI’s acquisition would be immediately accretive, funding further R&D and share repurchases.

Key Data Points (Updated - May 2025)

• Scale & Distribution: 700 million registered users, 18 paid seats, 575K businesses

• Data & Infrastructure

  • Multi‑exabyte storage (Magic Pocket)

  • Hybrid cloud: 90% on proprietary data centers, edge network for global performance

• Financial & Ownership

  • Enterprise Value: ~$1.4 B; EV/Revenue: ~4.1×; EV/FCF: ~12.1× (CF from Ops less Capex)

  • Voting Control: Drew Houston holds ~77% voting power via dual‑class shares

  • Ownership Split: 75.7% institutional, 6.2% insiders, 18.1% retail

Trusted data scale, enterprise‑grade storage infrastructure, and a global distribution engine—all primed for immediate AI integration. Acquiring Dropbox could be the most direct path from powerful models to transformative, memory‑enabled AI assistants.

The incentive for humans creating and sharing original content will decline as it becomes more challenging to breakthrough the noise and guaranteed to be scraped by AI models for further training. A doom-loop emerges: AI-generated content proliferates, authentic human content recedes, and the open internet gradually becomes less authentic and useful—a phenomenon sometimes referred to as the “dead internet.”

In response, users may migrate towards trusted, smaller networks characterized by stronger identity verification and selective or gated access. We’ve seen early evidence of this trend across social media (to private messaging)1, forums (to invite-only communities), and ecommerce (to group-buying, live-streaming, and DTC storefronts). While these shifts point to a broader movement, large segments of the internet remain unchanged and vulnerable to degradation from unchecked AI content.

Degradation is the most benign interpretation of what is currently taking place. The internet as a mechanism for predatory behavior is growing rapidly as crime organizations professionalize. AI supercharges this challenge. Sue-Lin Wong (The Economist) has done excellent reporting on internet fraud. Please read or listen to her Scam Inc series and share it with anyone for whom you care.

Informal solutions, such as industry Slack groups or private job boards, currently provide interim relief. However, there is a significant opportunity to build businesses around trusted networks, leveraging strong identity verification and targeted, closed-network dynamics. Companies able to offer compelling services within these networks could quickly establish leadership positions, subsequently expanding their offerings into workflow automation, AI-assisted coordination, and vertical-specific software solutions. We believe this is a second-order way to create vertical AI-powered software businesses.

There is enormous potential in companies solving this trust and truth gap and are starting to invest behind this hypothesis. As AI-generated noise increases, trusted networks will become essential infrastructure for preserving discovery-driven value creation.

This fails to capture the full story. While operational labor replacements will be a temporary value creation mechanism, AI point solutions will eventually compete against similar software offerings and have limited standalone moats. Competition (and subsequent margin compression) is accelerated by AI, as code generation enables rapid software development and replication.

Companies will need to deliver both lower costs and increased product quality to capture and maintain material take rates. AI that replaces labor can and will drive cost savings, but only those businesses that offer transcendent and differentiated product quality will build enduring moats. To achieve these outcomes, rapid feature/product expansion (beyond initial labor savings features) and the creation + usage of previously unusable/unavailable data will be required. With this data, the correct positioning in the value chain, and the features enabled by foundation models, companies will be able to drive incremental volume, revenue, and/or product margin, making their products an inextricable part of a customer’s business.

Therefore, we believe the largest opportunities exist in companies that focus on specific verticals where unique data can be combined with industry insights to deliver meaningfully better customer experiences. Even the best AI models generally do not contain the industry expertise or context necessary for the best outcomes. Vertical businesses can connect to legacy systems to extract and leverage existing data, drive downstream operations, and, over time, expand feature and product offerings that span the full value chain.

An example (that conveniently talks the Logos book) is Mia, which provides AI-powered automotive retail software. At first glance, Mia is a straightforward replacement of labor with its AI-powered customer service for automotive dealers. Alone, this first product solves a material labor and cost challenge for dealerships. Mia’s AI-powered customer service provides 24/7 coverage and connectivity to additional dealer systems that enable automated and seamless downstream operations including scheduling, test-drives, inventory management, service optimization, and financing administration. However, Mia sees these customer interactions as a wedge to capture customer data and build a machine deliver outbound customer communications that deliver high ROI customer retention and marketing interactions. Dealerships spend $500K+ annually1 on marketing across channels to refill the sales funnel due to low consumer retention rates (i.e. repeat purchase) of ~30%2. Successful delivery of these services can make Mia an invaluable part of the full dealer value chain and can transform both the processes and economics of consumer acquisition and retention.

This perspective is not in keeping with the historical outcomes of software. Total software market capitalization in the United States is $6T+3, with the top fifteen vertical software businesses only comprising ~$300B (<5%)4. IT budgets in most industries are <10% of revenue5, with vertical SaaS spend (where applicable) often <1% of customer revenue. As software replaces headcount, we expect to see material expansion of IT budgets driven by vertical software offerings.

The shift toward vertical AI businesses—those that leverage unique industry data to create differentiated, deeply embedded products—will redefine traditional IT budget constraints and significantly expand the total addressable market for software. Identifying these businesses early, as they transform operational economics, is central to our investing approach.