OMNITECH CAPITAL Investment Thesis
Strategic Memo · April 2026

The Palantir Analogy: how a closed loop between data, decisions, and execution becomes the most defensible position in a category.

Once every decade or so, an underlying shift opens a category that didn't exist before, and one firm builds the institution that defines it. 2003 was that moment for Palantir. The intelligence community had data, software, and analysts. What it didn't have was a way to close the loop between data coming in, decisions being made, action being taken, and outcomes feeding back into the system. Palantir didn't sell software. It earned its way into the loop — by sending engineers to live alongside analysts, building what worked, and turning the patterns it learned into a platform. Twenty years later, that closed loop is a $330B+ public company. The pattern repeats. The same loop is what AI-native GTM needs in 2026, and Omnitech is the bet on closing it.
Prepared ForInvestors & Strategic Partners
AuthorOmnitech Capital Limited
ReferencePalantir Technologies · NASDAQ: PLTR
Read AsPattern Repeat — 2003 / 2026
01 · The Reference Case

What Palantir did, 2003 onward

Palantir did not start as a platform. It started by solving a single, urgent problem — and the platform emerged from the patterns it learned along the way. Twenty years of compounding on that approach produced a top-fifty US public company.

$330B+
Market Cap · April 2026
$2.9B
2024 Revenue · 29% YoY Growth
~50%
Operating Margin · Recently Profitable
600+
Commercial Customers · 2025
The moment

By 2003, US intelligence agencies were under enormous pressure to prevent another 9/11. The 9/11 Commission later concluded that the attacks had been preventable — the data needed to detect them existed inside government systems, just scattered across agencies, formats, and classification levels that didn't talk to each other. The agencies had data. They had software. What they didn't have was a way to bring it together fast enough to act on it. The technology to fix this existed; nobody had assembled it for the speed and security demands of national-security work.

The strategic act — start narrow, embed, then build

Palantir didn't pitch a platform. It pitched analysts a single capability: help you connect data fast enough to make a real decision. To deliver that, the founders sent their own engineers — Forward Deployed Engineers (FDEs) — to live inside intelligence agencies for months at a time. The FDEs sat with analysts, watched decisions happen in real time, learned what data was missing and what signals actually mattered, and modified the software based on what users needed. Every deployment was custom at first. The platform — Gotham — emerged from the patterns the FDEs saw across deployments. Palantir earned the platform by doing the work first.

The closed loop

What Palantir actually built was a loop, not a product. Data came in (intelligence feeds, sensor networks, classified systems). Analysts interpreted it with FDEs alongside them. Decisions were made. Action was executed (operations, responses, deployments). The outcomes generated new data (what worked, what didn't, what signals mattered) that fed back into the system, making the next decision faster and more accurate. This loop is the architecture. Each component — Gotham (the platform), Foundry (the commercial version), Apollo (the deployment layer), AIP (the LLM layer added in 2023), the FDE organization — exists to make the loop tighter and faster.

The proprietary data that emerged

Anyone could buy raw intelligence data. APIs and models were available to competitors. What Palantir alone had was what actually worked in the real world — outcome data generated by the loop running inside customers, year after year. That outcome data is what made the next deployment faster, the next analyst more effective, the next agency easier to onboard. The platform was useful. The proprietary outcome data was the moat. By the time Palantir crossed into commercial markets with Foundry, that data advantage was a decade old and still compounding.

The outcome

Twenty years of compounding on the same closed-loop model: $0 to $2.9B in 2024 revenue (29% YoY growth), $330B+ market cap as of April 2026, top-fifty US public company by market value, recently and durably profitable with operating margins approaching 50%. The company that nobody could classify for fifteen years — too software for defense, too services for software, too political for institutional investors — is now one of the highest-multiple stocks in the S&P 500. The platform was necessary. The Forward Deployed Engineers were necessary. The proprietary outcome data was necessary. The loop is what produced the returns.

02 · What the Playbook Teaches

Five moves that close the loop

Strip the intelligence-community details out of what Palantir did and what remains is a five-step pattern for closing the loop between data, decisions, and execution. The moves are sequential — each one earns the right to the next — and once the loop is running, the company that built it has a position no software vendor and no services firm can replicate. Below are those five moves, named generically, with the Palantir version of each as the proof.

1 · Start with a specific problem, not a platform vision

Palantir didn't pitch "a data platform." They pitched "help analysts connect data fast enough to make a real decision." The narrow wedge is what got them through the door. Companies that pitch broad platform visions to early customers fail; companies that solve a specific painful problem and let the platform emerge from the work succeed. The wedge has to be sharp enough to be undeniable, and small enough that the customer says yes.

2 · Embed inside the customer's workflow

The company that puts its own people inside customer operations — not as consultants billing hours, but as engineers and operators who own outcomes — learns things its competitors cannot. Palantir's FDEs sat with analysts during live missions. They saw what decisions actually looked like, what data was missing, what signals mattered, and how the work actually got done. That information cannot be acquired through customer interviews or product analytics. It can only be acquired by being there. Once it's acquired, it becomes the basis for every product decision that follows.

3 · Connect insight to action — don't stop at the dashboard

This is where most companies fail. Most stop at insights, dashboards, reports — the kind of output that a smart customer looks at and says "now what?" Palantir went further: identify the threat, coordinate the response, execute the operation. The jump from insight to action is what separates a tool from an institution. Customers pay differently for software that tells them what's happening versus software that closes the loop and produces the outcome.

4 · Generate proprietary data from the loop running

Anyone can buy raw data. Anyone can call APIs. Anyone can run models. What no one else has is what actually worked in the real world — the outcome data that only emerges when the loop is running inside live customers. That data compounds: every deployment makes the next one faster, every decision feeds back into the next. After a few years it becomes uncopyable. The buyers buying data; the institution generating it. That asymmetry is the moat.

5 · Earn the platform by doing the work first

Palantir spent years doing what looked like custom services. Each deployment was bespoke. From the outside, the company looked like a heavy-services consultancy. What was actually happening: every deployment was teaching them what the platform needed to be. Patterns emerged. Workflows standardized. Custom became reusable. Reusable became Foundry, Apollo, AIP. The platform was the destination, not the starting point — and that ordering is what made it impossible to copy.

03 · Structural Parallel

Palantir's playbook, mapped to today

Twelve dimensions of the Palantir build, mapped to the same pattern repeating in 2026. Read across — not as competition, but as the same loop being closed in two different categories, twenty-three years apart.

Dimension 2003 · The Palantir Build 2026 · The Same Loop, in AI-native GTM
Underlying Conditions Data, software, analysts all existed; the loop between them was broken. AI tools, data, models, operators all exist; the loop between them is broken.
The Wedge — Specific Problem "Help analysts connect data fast enough to make a real decision." "Help revenue teams decide who to target — and execute the pipeline."
Data Coming In Intelligence feeds, sensor networks, classified systems, agency databases. Buyer signals, intent data, CRM data, content engagement, vertical research.
Insight Layer Gotham + analysts; visual querying without code. GTMplus.ai + Decision-Maker Network research; intelligence on who to target and why.
Embedded People Forward Deployed Engineers — Palantir staff embedded inside agencies for months. GTMBench — Director-to-CxO operators placed inside customer companies in 96 hours.
Action Layer Operations executed by analysts, supported by Palantir software in real time. ENAI + IndustryGeniuses — campaigns and pipeline executed alongside customer teams.
Proprietary Data Generated Outcome data — what worked, what signals mattered, how decisions actually played out. Outcome data — what targeting worked, which campaigns produced pipeline, what messages converted in which verticals.
Feedback Into the System Outcome data fed back into Gotham, Foundry, AIP — making each next deployment faster. Outcome data fed back into GTMplus intelligence and ENAI workflows — making each next campaign sharper.
Authority Layer Karp's books and op-eds, AIPCon flagship event, partisan brand, recognizable identity. Decision-Maker Network (BoardroomAI, VerticalAI, GTMBench Review); GTM Summit London Q4 2026.
Pricing & Economics Multi-million-dollar contracts bundling software + Forward Deployed Engineers; sold as one. Subscription tiers + outcome-aligned operator placement; software and people sold as one.
Moat & USP Closed loop + proprietary outcome data + embedded people + brand institution. Closed loop + proprietary outcome data + embedded operators + Decision-Maker Network.
Proof Point $330B+ market cap; $2.9B 2024 revenue at 29% growth; recently profitable at ~50% margin. Five integrated operating units assembled and launching; loop architecture in place from day one.
Underlying Conditions
2003 · PalantirData, software, analysts all existed; the loop between them was broken.
2026 · AI-native GTMAI tools, data, models, operators all exist; the loop between them is broken.
The Wedge — Specific Problem
2003 · Palantir"Help analysts connect data fast enough to make a real decision."
2026 · AI-native GTM"Help revenue teams decide who to target — and execute the pipeline."
Data Coming In
2003 · PalantirIntelligence feeds, sensor networks, classified systems, agency databases.
2026 · AI-native GTMBuyer signals, intent data, CRM data, content engagement, vertical research.
Insight Layer
2003 · PalantirGotham + analysts; visual querying without code.
2026 · AI-native GTMGTMplus.ai + Decision-Maker Network research; intelligence on who to target and why.
Embedded People
2003 · PalantirForward Deployed Engineers — Palantir staff embedded inside agencies for months.
2026 · AI-native GTMGTMBench — Director-to-CxO operators placed inside customer companies in 96 hours.
Action Layer
2003 · PalantirOperations executed by analysts, supported by Palantir software in real time.
2026 · AI-native GTMENAI + IndustryGeniuses — campaigns and pipeline executed alongside customer teams.
Proprietary Data Generated
2003 · PalantirOutcome data — what worked, what signals mattered, how decisions actually played out.
2026 · AI-native GTMOutcome data — what targeting worked, which campaigns produced pipeline, what messages converted.
Feedback Into the System
2003 · PalantirOutcome data fed back into Gotham, Foundry, AIP — making each next deployment faster.
2026 · AI-native GTMOutcome data fed back into GTMplus intelligence and ENAI workflows — making each next campaign sharper.
Authority Layer
2003 · PalantirKarp's books and op-eds, AIPCon flagship event, partisan brand, recognizable identity.
2026 · AI-native GTMDecision-Maker Network (BoardroomAI, VerticalAI, GTMBench Review); GTM Summit London Q4 2026.
Pricing & Economics
2003 · PalantirMulti-million-dollar contracts bundling software + Forward Deployed Engineers; sold as one.
2026 · AI-native GTMSubscription tiers + outcome-aligned operator placement; software and people sold as one.
Moat & USP
2003 · PalantirClosed loop + proprietary outcome data + embedded people + brand institution.
2026 · AI-native GTMClosed loop + proprietary outcome data + embedded operators + Decision-Maker Network.
Proof Point
2003 · Palantir$330B+ market cap; $2.9B 2024 revenue at 29% growth; recently profitable at ~50% margin.
2026 · AI-native GTMFive integrated operating units assembled and launching; loop architecture in place from day one.
04 · The Same Shape, Today

Where the pattern points now

AI-native GTM in 2026 is in the same position institutional data work was in 2003. All the components exist — large language models, contact data, intent platforms, CRMs, sequencers, AI agents — but the loop between them is broken. Revenue teams cannot move from data to decision to action to outcome fast enough to operate at the speed competitors are. The market has tools. What it doesn't have is the institution that closes the loop. The differentiation is not in the tools themselves — it is in who knows what to point them at, and who can run the work alongside the customer until the loop starts compounding.

Omnitech is putting five components in place at once — instead of building them one at a time over two decades — each one matching a move from the Palantir playbook:

Move 1 · Start with a specific problem
"Decide who to target — and execute the pipeline"

The wedge is narrow and concrete. Not a platform pitch, not a vision deck — a single capability AI-native revenue teams will pay for from day one. The platform emerges from the work, the way Gotham emerged from Palantir's first deployments.

Move 2 · Embed inside the customer's workflow

GTMBench places senior operators (Director through CxO) inside customer companies in 96 hours. ENAI runs alongside them as the execution platform. Operators sit with revenue teams during live work — the Forward Deployed Engineer model applied to AI-native GTM. This is what produces the workflow knowledge that becomes the next product decision.

Move 3 · Connect insight to action

GTMplus generates the intelligence (who to target, when, with what message). ENAI executes the workflow. IndustryGeniuses runs the campaigns and produces pipeline. The jump from insight to action is built in — no dashboards-and-reports stop along the way. The customer pays for the outcome, not the analytics.

Move 4 · Generate proprietary data from the loop
Others buy data. Omnitech creates data.

Every customer engagement produces outcome data — what targeting worked, which campaigns generated pipeline, what messages converted in which vertical, which operator playbooks scaled. That data feeds back into GTMplus and ENAI, making the next campaign sharper. Competitors can buy contacts and call APIs. They cannot buy what worked in the real world. That asymmetry is the moat.

Move 5 · Earn the platform by doing the work
Five operating units, integrated from day one

Decision-Maker Network (BoardroomAI, VerticalAI, GTMBench Review) for authority. IndustryGeniuses for demand. ENAI for execution. GTMBench for operators. GTMplus for intelligence. Each unit makes the loop tighter. The platform is the destination — but unlike Palantir, Omnitech is launching with all five in place from day one, because the underlying components no longer require sequential build.

What's structurally different from 2003
Commercial-first, integrated from launch

Palantir spent eight years selling almost exclusively to government and defense before Foundry crossed into commercial. The structural choice was forced — that's where the urgent post-9/11 problem and the funding lived. Omnitech's market is commercial from day one: AI-native scale-ups, growth-stage SaaS, PE-backed companies, F500 enterprise revenue teams. The model that took Palantir twenty years to scale into commercial enterprises is the model Omnitech launches with.

Once every decade or so, an underlying shift opens a category, and one firm builds the institution that defines it. Palantir is one proof of that pattern — the data, the software, and the analysts already existed, but only the company that closed the loop between them produced the outcomes customers actually needed. The same setup is in front of us now, in revenue teams instead of intelligence agencies. The components all exist. The loop is broken. Omnitech is the bet on closing it. The pattern stands on its own.