Who Is Palantir (PLTR)?: Turning Disparate Data into an “Operationally Actionable” Form and Capturing On-the-Ground Implementation in the AI Era

* This report is based on data as of 2026-02-05.

PLTR’s business, explained like you’re in middle school

Palantir Technologies (PLTR), in one sentence, is a software company that connects scattered internal data and turns everything from frontline decision-making through execution (operations) into something that actually runs. It’s not just an analytics tool that spits out charts. The core value is taking “what we should do next” inside government agencies and large enterprises and translating that into executable operating workflows.

Think of it like this: every department is navigating with its own outdated map. Palantir helps merge those maps into one living, shared view of “where we are” and “where we need to go,” so teams can coordinate and act.

Who it serves (two customer pillars)

Pillar 1: Government (defense, public safety, and public-sector administration)

One major customer group is government agencies. In national defense and security, border/public safety/disaster response, and large-scale public administration—areas where errors are unacceptable, data handling must be rigorous, and operations are complex—PLTR’s design philosophy tends to fit particularly well. It has also been reported that U.S. government-related contracts have been a key growth driver recently.

Pillar 2: Commercial (especially large U.S. enterprises)

The other growing pillar is commercial customers across manufacturing, healthcare, energy, financials, and communications/infrastructure—industries that tend to be a strong fit given large data volumes and heavy regulation/control requirements. In recent years, many companies “want to use AI but don’t have their internal data in order,” which creates a backdrop where PLTR’s deal activity can more readily pick up.

What it sells (think of the product as three toolboxes)

1) The data foundation: turning scattered information into something “usable”

It ingests and harmonizes data—resolving mismatches like inconsistent naming conventions and IDs across departments—then translates it into screens and procedures that frontline teams can actually use. When this layer is strong, operations can run even before AI, and it also becomes the base layer for deploying AI.

2) The operations cockpit: making situational awareness → choice → execution continuous

The design is meant to go beyond visualization and include comparing options and carrying decisions through to execution (operations). For government, that looks like operational planning and response; for enterprises, it’s closer to production planning, inventory management, and optimizing sales activity.

3) AIP: a way to put AI into operations “safely” and “auditably”

What’s become increasingly important is AIP (Artificial Intelligence Platform). When enterprises and governments use AI, frontline requirements get strict—limits on data exfiltration, access-control administration, audit trails, the ability to verify incorrect outputs, embedding into approval workflows, and more. PLTR is positioning AI not as a “chat toy,” but as something you deploy and run as production operations inside a controlled environment.

As concrete examples, the company continues to publish guidance on (beta) functionality that enables chat-based analysis while tracing internal data structures and verifying intermediate steps, alongside ongoing expansion of developer-facing features and tools.

How it makes money (revenue model)

The core is enterprise and government software contracts. Deployments often require work to “connect data” and “embed it into operating procedures,” and support for that work can also contribute to revenue. Contracts can expand as usage broadens across departments, data sets, and workflows—but the upfront hurdle before go-live is not trivial.

There are also cited early signs of partner-led distribution, such as a telecom operator bundling PLTR software to deliver enterprise AI services (e.g., the partnership with Lumen).

Why customers pick it (the core value proposition)

• Fast at making data “usable” (not just collecting it, but getting it into a state frontline teams can work with)
• Doesn’t stop at analytics; it drives operations (built to cover the full path from decision-making to execution)
• Built for stringent environments (tends to resonate in government and regulated industries that require security, auditability, and access control)
• A design philosophy aimed at reducing AI incidents (messaging centered on traceability and verification of data and procedures)

Growth drivers: what’s providing tailwinds

• Pressure to “adopt AI” is intense, but many companies first run into a more basic problem: “our internal data is scattered and unusable,” making end-to-end data integration plus operational implementation a differentiator
• In periods when large U.S. government contracts are more likely to move, demand in defense, public safety, and public-sector operations tends to line up with PLTR’s strengths
• When land-and-expand (expanding after initial entry) works, contracts can compound more easily (spreading across departments, workflows, and decision layers)

Future pillars: initiatives that could matter even if revenue is still small

This section is less about “today’s revenue scale” and more about areas that could reshape future competitiveness and the profit model.

• Accelerating frontline “AI-ification” via AIP: pushing workflows forward in a way that reflects internal rules around permissions, logging, and accountability (an AI-agent-like usage pattern)
• Enterprise AI delivery via partners: early signs of building distribution through telecom, cloud, and infrastructure partners (potentially lowering implementation friction, while also raising questions around control and ownership of the customer relationship)
• Democratizing analytics via chat: making it easier for non-engineers to navigate internal data, potentially expanding users inside adopting enterprises and, in turn, expanding contracts

Long-term fundamentals: the “pattern” this company has grown in

PLTR’s revenue has grown over time, but the story also includes a clear transition from losses to profits—and the numbers tend to move in waves. The materials classify it, under Lynch’s six categories, as a “Cyclicals-leaning hybrid”. Here, “Cyclicals” is less about classic macro cycles like materials or autos and more about the idea that earnings and EPS can statistically show “waves” as results move from loss territory into profit territory.

Long-term revenue trend: high growth as the base case

Revenue increased from $595 million in FY2018 to $2.866 billion in FY2024, about 4.8x. The annualized growth rate is also shown as high: roughly 31.0% over 5 years and roughly 29.9% over 10 years.

Long-term profit trend: the move from losses to profits is driving the “pattern”

Net income was negative from FY2018 through FY2022, then turned profitable at +$210 million in FY2023 and +$462 million in FY2024. That inflection is the biggest driver of the “waveform,” and it also creates a setup where EPS growth can look extremely large once profitability is reached.

Cash flow: FCF expands after turning sustainably positive

Free cash flow (FCF) was negative from FY2018 through FY2020, then flipped to +$321 million in FY2021 and grew to +$1.141 billion in FY2024. The observed trajectory suggests cash generation has strengthened alongside revenue growth.

Margins and ROE: high gross margin; operating margin has moved into the black

• Gross margin has been high over the long term (generally in the 67%–80% range), with FY2024 at ~80.2%
• Operating margin was negative for many years, but turned positive at +5.4% in FY2023 and +10.8% in FY2024
• FCF margin also improved, reaching ~39.8% in FY2024

ROE is 9.24% in FY2024. Given its historical volatility driven by losses and capital-structure changes, it’s more natural to view this not as a “consistently high-ROE compounder,” but as a period where ROE is catching up after the shift into profitability.

“Uncomputable” growth-rate issues also matter

Annual EPS growth rates (5-year and 10-year) and annual FCF growth rates (5-year and 10-year) are treated as not computable as growth rates, in part due to the long loss-making period. Rather than filling the gap with conjecture, the right approach is to confirm the pattern via “level changes” (profitability achieved; FCF positive and rising since 2021).

Capital allocation: not a dividend story

Dividend yield, dividend per share, and payout ratio are described as difficult to verify due to insufficient data on a recent TTM basis. As a result, it’s not appropriate to build the thesis primarily around dividends; instead, the focus should be on capital allocation checks such as reinvestment and (if applicable) share repurchases and other mechanisms.

Short-term momentum (TTM / last 8 quarters): is the near-term “pattern” still intact

Whether the long-term “pattern” is holding in the short run matters directly for investment decisions. In the materials, short-term momentum is assessed as Accelerating.

Revenue and EPS: strong near-term growth

• EPS (TTM) YoY: +245.4%
• Revenue (TTM) YoY: +56.2%

This EPS spike fits a “recovery-phase growth profile,” where a long stretch of losses/low profitability followed by a move into profitability and margin improvement can make growth rates look unusually large. In other words, while the short-term figures broadly align with the long-term pattern (a Cyclicals-leaning waveform alongside growth), it’s also important not to treat them as steady-state growth.

FCF (TTM) is hard to judge: don’t force a conclusion with missing data

FCF (TTM) and its YoY change can’t be assessed because the most recent values are difficult to verify due to insufficient data, so last-year FCF momentum can’t be determined. Meanwhile, an 8-quarter (2-year) aggregation shows FCF trending higher on an annualized basis (CAGR +60.4%). It’s important not to mix up “a last-1-year (TTM) conclusion” with “the 8-quarter trend”.

Margins (near-term quality): quarterly improvement is showing up

Margins appear to be improving alongside revenue. As one quarterly example, operating margin is shown improving from 24Q4: +1.3% (Q) → 25Q4: +40.9% (Q). That supports the view that near-term growth isn’t just “revenue up without profits,” but a phase accompanied by improving profitability.

Consistency check vs. the long-term classification (pattern): broadly intact, but FCF can’t be verified

The materials conclude that the long-term “Cyclicals-leaning hybrid” framing has not materially broken down based on the last year’s EPS and revenue trends and is broadly consistent (maintained). Because FCF (TTM) is difficult to confirm, cash-based consistency checks are deferred.

Financial soundness (including bankruptcy risk): low leverage and plenty of liquidity

Enterprise software is often influenced less by the macro cycle and more by “customers’ investment decisions,” which makes financial flexibility important. Key points from the materials are as follows.

• Debt-to-capital ratio (FY): 0.048 (not a balance sheet that depends heavily on borrowing)
• Net Debt / EBITDA (FY): -14.59x (a level that typically indicates a net cash position)
• Cash ratio (FY): ~5.25, and on a quarterly basis a high 6.11 (25Q4)
• Current ratio (quarterly): most recent 7.11 (25Q4)

On interest coverage, the materials avoid a definitive near-term conclusion due to periods with insufficient data in the quarterly series, while noting that some FY periods can be confirmed as positive (though the most recent FY is difficult to verify). Overall, at least within the presented data, this is not a profile where bankruptcy risk is primarily about “getting squeezed by interest payments.” Instead, the size of the cash cushion stands out.

Where valuation stands today (historical vs. itself only): “where are we in the past range” across six metrics

Here, without peer comparisons, we summarize where the current level sits relative to PLTR’s own historical distributions (5-year and 10-year). The six metrics are PEG, PER, free cash flow yield, ROE, free cash flow margin, and Net Debt / EBITDA.

PEG: 0.980 (below the normal range over the past 5 and 10 years)

PEG is 0.980, below the normal range over the past 5 and 10 years (1.147–4.736). On a 5-year view, it sits on the low side (around the ~17th percentile), and over the last two years it’s described as trending downward (toward lower levels).

That said, PEG is highly sensitive to the denominator (the growth rate), and it can screen low when near-term EPS growth is unusually high. Accordingly, this section limits itself to the factual observation that “a relatively low PEG versus growth is being observed in the current phase.”

PER (TTM): 240.5x (lower in the historical distribution, but high in absolute terms)

Assuming a share price of $151.86, PER (TTM) is 240.5x. Because it is below the normal range over the past 5 and 10 years (281.5–426.7x), it screens on the “lower side” historically. At the same time, the absolute level is still high, which naturally reflects that PLTR has often traded in phases where PER looks extremely elevated.

Over the last two years, the direction is described as down from very high levels (e.g., a settling pattern from 457.7x → 426.7x → 281.5x).

Free cash flow yield (TTM): current position cannot be placed

Free cash flow yield (TTM) can’t be positioned versus the historical range (inside / breakout above / breakdown below) because the most recent value is difficult to verify due to insufficient data. Historical observations include a 5-year and 10-year median of 0.60% and a normal range of 0.38%–1.08%, but the current position can’t be determined from this section alone.

ROE (FY): 9.24% (above the 5-year range; within the 10-year range and above the median)

ROE is 9.24% on an FY basis. It’s above the 5-year normal range (-33.49%–6.68%), putting it on the high end in a 5-year context. Within the 10-year normal range (-21.08%–25.26%), it’s in-range but above the median (6.04%). The last two years are described as a continuation of improvement (an upward trend).

Free cash flow margin (TTM): current position cannot be placed (FY provides a separate fact of 39.83%)

FCF margin (TTM) can’t be positioned because the most recent value is difficult to verify due to insufficient data. Historical observations include a 5-year median of 20.83% and a normal range of 2.06%–33.03%. Separately from TTM, there is the fact that on an FY basis the most recent year’s FCF margin reached 39.83% (the difference between FY and TTM reflects different measurement periods).

Net Debt / EBITDA (FY): -14.59x (lower implies more cash; near the lower bound over the past 10 years)

Net Debt / EBITDA is an inverse indicator where a smaller (more negative) value generally implies a thicker cash position and something closer to net cash. The current FY value is -14.59x, which is within the 5-year normal range (-16.17–7.29x) but toward the low end, and near the lower bound of the 10-year normal range (-14.59–4.82x). Over the last two years, it’s described as staying negative, including phases where it moved toward smaller (more negative) values.

Cash flow tendencies (quality and direction): consistent with EPS, but watch “slower collection” friction

On an annual basis, PLTR turned net income positive in FY2023–FY2024, and FCF has been positive and rising since FY2021. So this is not a case where “cash is going out even though profits aren’t there.” Instead, the trajectory indicates that profitability and improving cash generation have advanced together.

At the same time, a key “quality of growth” issue is the deterioration in accounts receivable turnover (days). In annual data, days to collect increased from ~12 days in 2018 → ~73 days in 2024. If that trend continues, it could create friction where “profits look good, but cash conversion slows.” Without drawing a definitive conclusion, it’s faithful to the materials to flag this as a monitoring point.

Success story: why PLTR has been winning (the essence)

PLTR’s structural essence is taking scattered data, turning it into something that “runs as operations,” and supporting the full loop from decision-making through execution. It’s built to embed into “operations,” not just “analytics,” and that strength tends to show up in government, regulated industries, and large-enterprise operating environments.

In this domain, the more implementation progresses, the more data connectivity, operating procedures, and the design of permissions and accountability get embedded into the organization—making replacement costs (switching costs) likely to rise. If it becomes foundational, it can turn into value that’s hard to substitute.

The tradeoff is that the more value it can create, the heavier the implementation tends to be—and execution capability through “effective adoption” is continuously tested.

Story continuity: are recent strategies aligned with the winning path

The most notable narrative shift over the last 1–2 years is the move from “a data integration and analytics company” to “a company that brings AI into operations (an implementer of frontline AI-ification)”. Strengthening developer-facing capabilities around AIP and expanding mechanisms designed for agent operations reinforce that direction. This is consistent with the original winning path: running operations under constraints like control, auditability, and permissions.

At the same time, geographically, there are signs that U.S. concentration is increasing. A declining non-U.S. mix and slower adoption in Europe have been reported, suggesting the company may be leaning more toward “going deeper in the U.S. (government + large enterprises)” than “expanding globally.” That can support growth, but it also ties directly to the concentration risk and market-access constraints discussed below.

Quiet structural risks: fragilities that matter more as the story looks stronger

This section is not a definitive claim about what is happening today, but a set of fragilities that are structurally more likely to emerge.

• Skewed customer concentration (government / U.S.): Government business is a strength, but it’s influenced by budgets, administration policy, procurement priorities, contract terms, public opinion, and tighter oversight. If the non-U.S. mix keeps shrinking, geographic diversification becomes less effective
• Rapid shifts in the competitive environment: With the AI boom, boundaries between data platforms, analytics, and business applications are blurring, and mega-vendors may push “down into operations.” Partnerships with external platforms can be a tailwind, but they also expand customer choice
• Risk of losing differentiation: If decision-makers don’t understand the value of control design plus implementation capability and conclude “our existing cloud / data platform is enough,” partial deployments and stalled land-and-expand could become more common
• Dependence on a “digital supply chain” rather than a physical one: The core dependency is on cloud platforms, customer data platforms, and external model connectivity. AIP’s expansion toward ingesting multiple models can be read as a move to reduce reliance on any single model
• Deterioration in organizational culture: A common pattern is teams that are highly capable and mission-driven but carry heavy load and are prone to burnout. This can show up later as variability in implementation quality or reduced support capacity
• Profitability deterioration as a potential leading indicator: Margins could come under pressure from discounting, higher implementation support costs, and front-loaded investment to win large deals. In addition, worsening collection days (12 days → 73 days) could signal a “gap between profits and cash”
• Deterioration in interest-paying capacity: Borrowing dependence is currently low and unlikely to be the core risk, but if the company ramps large investments, equity compensation (dilution) and higher fixed costs could become burdens before interest expense does
• European adoption friction becoming structural: Data sovereignty, preference for domestic vendors, and political acceptability can constrain purchasing and may persist as a market-access issue

Competitive landscape: a multi-layered fight with a moving opponent

PLTR’s competition isn’t a clean head-to-head matchup with one peer. It competes across overlapping layers—data platforms, BI, business applications, MLOps/LLMOps, cloud integration platforms, and government SI/integration—while plugging into customers’ existing stacks. The key phrase in the materials is “compete while collaborating” (connecting/partnering with Databricks and Snowflake while aiming to establish a presence in higher-level use cases).

Key competitors (not just “similar products,” but competition for budget)

• Microsoft (Fabric / Azure AI and adjacent areas)
• Databricks
• Snowflake
• ServiceNow (workflow × AI)
• Booz Allen Hamilton / SAIC (government-focused SI and integrators)
• Anduril (closer to defense AI and autonomous systems; collaboration exists, but it can also become a contest for leadership within the same budget envelope)

A layer-based competitive map

• Data platforms: Snowflake, Databricks, Microsoft, etc. are strong. PLTR more often connects than replaces, aiming to capture higher-level value
• Analytics to AI implementation: Can compete with Databricks, Microsoft, and in-house builds + OSS. PLTR positions deployment, auditability, and control as differentiation axes
• Business workflows: Workflow-heavy players like ServiceNow are strong, and can compete with PLTR’s “decision-making to execution” positioning
• Government and defense: Outcomes are shaped not only by product, but by prime contract structures, institutions, and integration leadership, making competition with SI players more likely

Moat: what it is, and how durable it may be

In the materials’ framing, PLTR’s moat isn’t “exclusive ownership of data volume.” It’s an operational structure that can run under control (ontology-like objectification of business operations) plus accumulated implementation and deployment patterns (repeatability). As deployments deepen, the number of data connections, definitions of business objects, permission design, audit lines, and embedding into frontline procedures compound—raising switching costs.

That durability is sustained not by “product alone,” but by implementation and operational capability (talent, support organization, and partner network). If repeatability keeps shortening time-to-production, the moat can deepen. But if mega-platforms bake control, auditability, and operations into integrated stacks and standardize them as default features, the moat can erode.

Structural position in the AI era: tailwinds and headwinds intensify at the same time

Network effects: not an external SNS effect, but an internal “operational network effect” inside customers

Strong direct network effects appear limited, but the materials describe an operational network effect: within a given customer, as users, use cases, and departments expand, connection density rises and land-and-expand becomes easier.

Data advantage: not exclusive volume, but an edge in “structuring for operational use”

PLTR’s core is less about collecting data and more about integrating heterogeneous data into a form that can support decision-making and execution—then translating it into operations with auditability, permissions, and accountability. In other words, it has an advantage in structuring data so it can “run in the field.”

AI integration depth: high (deploy and run, not just experiment)

AIP isn’t limited to a chat UI; it’s a design philosophy embedded into workflows, permissions, auditability, and repeatability. The materials also note progress in providing functionality that validates and compares document processing and similar tasks in low-code form and pipelines them in a way that’s close to deployment—thickening the AI implementation layer.

Mission-criticality: high, but paired with political and institutional friction

While it tends to resonate in government, regulated industries, and large-enterprise operations, constraints beyond “product quality”—including politics, regulation, and acceptability—remain real market-access issues.

AI substitution risk: medium to low. The real issue is “absorption and commoditization”

Rather than being directly displaced by general-purpose AI advances, the core risk is framed as mega-platforms reaching the same conclusion (controlled AI operations are necessary) and then absorbing/commoditizing that capability through integrated stacks.

Position by structural layer: a middle layer aiming for “business OS-ification,” not an app

PLTR can be framed as a platform targeting a middle layer of operations (a business-OS-like common foundation) that connects data integration → decision-making → execution, rather than a standalone application. Developer-focused enhancements and partnerships with external infrastructure are positioned as moves to thicken this middle layer while reducing implementation friction and expanding real operational footprint.

Leadership and culture: a strength—and potentially a bottleneck

CEO vision: mission-critical × AI as production operations

CEO Alex Karp is described as consistently emphasizing—often in forceful terms—“software that can withstand national security and mission-critical frontline environments” and “implementing AI into organizations as production operations.” That vision fits PLTR’s business profile, which assumes control, auditability, permissions, and operational design.

More recently, he’s said to have highlighted U.S. demand more prominently and adopted a tone that raises concerns about slower AI adoption outside the U.S., including Europe. It’s more natural to view this as a shift in emphasis rather than a change in vision.

Persona, values, and communication: conviction-led, comfortable with conflict, focused on practical value

• Conviction-driven communication that tends not to avoid conflict (e.g., framing around inter-state competition)
• Pragmatism centered on “whether it works in the field” (whether implementation creates value; whether value exceeds what is paid)
• At times favors high-density execution with small teams over labor-intensive scaling (which can create high load)

Cultural reflection: high autonomy and high intensity can improve implementation repeatability

Karp’s mission orientation and implementation-first mindset are described as pulling the organization toward “winning in the field” and “implementing and running,” making a high-autonomy, high-intensity culture more likely. Higher autonomy and smaller teams can speed decisions, but in certain phases they can also make cross-team alignment and prioritization harder.

To scale with small, high-density teams, it becomes difficult to operate without productizing implementation and deployment and improving repeatability. The materials present a causal link: that pressure aligns with developer-focused strengthening of AIP and templating of deployment patterns.

General patterns in employee reviews: strengths and side effects

• Often cited strengths: strong talent, hard problems, mission orientation, high autonomy
• Often cited weaknesses: high intensity and burnout, moments where people cite weak training/evaluation/career transparency, values friction on ethical issues

Fit with long-term investors: separate the positives from the frictions

On the positive side, mission orientation and implementation-first execution can become more valuable as the product becomes more foundational. On the friction side, political and ethical issues can affect reputation and hiring, and the side effects of a high-intensity culture can spill into variability in implementation quality. From a governance perspective, the materials argue the key isn’t the aggressiveness of statements by itself, but monitoring how statements practically affect hiring, customer acquisition, and regulatory response.

The investor “map”: organizing causality with a KPI tree

PLTR is a name where the story can get flashy, but for long-term investing, anchoring on “causal variables” helps cut through the noise. In the materials, enterprise value causality is organized via a KPI tree.

Ultimate outcomes

• Profit growth (including sustained profitability)
• Higher cash generation (financial flexibility to self-fund)
• Improved capital efficiency (ROE, etc.)
• Improved profitability (operating profit compounding on a base of high gross margin)
• Business durability (accumulating “hard-to-stop usage” in mission-critical domains)

Intermediate KPIs (value drivers): the “core variables” for this name

• Revenue growth: as contracts compound, absolute profit and cash dollars tend to rise
• Land-and-expand within customers: as departments, use cases, and users expand from “points to surface area,” it tends to drive contract expansion and durability
• Shorter lead time to production: heavy implementations are more prone to bottlenecks; the more lead time can be reduced, the easier repeatable scaling becomes
• Improving profitability: as the cost structure—including implementation and operational support—matures, more profit tends to drop through
• Quality of cash conversion: the less collections lag, the more investment capacity and stability improve (collection days are a monitoring point)
• Repeatability of implementation and operations: reducing dependence on people and process and creating implementation templates can become competitiveness itself
• Fit with control, auditability, and access management: meeting adoption requirements in government and regulated industries and making mission-critical embedding easier

Constraints (friction): what tends to slow growth

• Heavy implementation and rollout (requires design, operations, and talent; depends on customer readiness)
• Phases where ROI is hard to explain (time before outcomes show up in operational KPIs)
• Geographic and institutional friction (especially outside the U.S., including Europe)
• Many competitive layers and a shifting main battlefield (opponents aren’t fixed)
• Friction where customers decide “the existing stack is sufficient” (difficulty communicating differentiation)
• Constraints in implementation talent and support capacity (burnout, etc. can spill into implementation quality)
• Cash conversion friction (pressure toward slower collections)
• Quality of partner-led distribution (lower implementation friction vs. diluted control)

Two-minute Drill: the backbone for evaluating this name for long-term investing

Palantir is less “a company with amazing AI” and more a company selling an implementation foundation that connects data and operations while incorporating the constraints that show up when enterprises and governments run AI in production (permissions, auditability, accountability, and procedures). The long-term strength is that as implementation deepens, operations get embedded into the organization and switching costs tend to rise.

The weakness comes from the same root. Implementation and rollout are heavy and often depend on people and process. Competitors aren’t fixed, and mega-platforms have incentives to commoditize capabilities through integration. And the more concentrated the business becomes in the U.S. government and U.S. market, the more political, institutional, and procurement friction can matter.

Accordingly, what long-term investors should focus on isn’t narrative momentum. It’s whether land-and-expand (operational footprint) keeps widening, whether time-to-production is shrinking in a repeatable way, whether profitability quality and cash conversion (collection terms) are holding up, and whether the “reason to add another vendor” (the need for control, auditability, and operations) persists even as competitors consolidate.