EN JA
CONFIDENTIAL ─ FOR NIINAMI / JOI
BOSWELL
Business Plan
Japan's Palantir ─ Vertical AI × FDE / CTO Embed Model
Date
2026.06.04
By
Tatsuya Ishibe
Contents
01Who is Palantir?
02What's Remarkable, How They Earn
03The Market We Should Target
04BOSWELL Business Model
CHAPTER01 / WHO IS PALANTIR
Who is Palantir?
Founded 2003 ─ A software company built for U.S. intelligence in the wake of 9/11
Founding Story (2003)
Founded immediately after the 9/11 attacks as counter-terrorism investigation software for U.S. intelligence. Replaced the practice of soldiers in Afghanistan and Iraq hand-mapping insurgent networks on paper — with software.
Source: Palantir S-1 (SEC, 2020/8/25) Business Overview
Identity Declaration
"We are not in the business of collecting, mining, or selling data. We build software platforms that enable our customers to integrate their own data."
= They don't sell data. They sell an OS that operates on the organization's own data — the polar opposite of Google / Meta.
2008 ─ GOVERNMENT
Gotham
Finds the "needle in the haystack" across vast signal volume and connects it through to operator-level execution. Born in intelligence/defense, now spans across government.
U.S. DoD / CIA / FBI / UK NHS / U.S. Army
2016 ─ COMMERCIAL
Foundry
Unifies a company's entire data and operates as the central OS of the organization. Deployed across all commercial sectors — manufacturing, energy, finance, healthcare.
"Foundry is becoming a central operating system for entire industries."
125
Customers (H1 2020)
36
Industries served
150+
Countries
$742.6M
2019 Revenue (+25% YoY)
53 / 47
Commercial / Government mix
"We have
chosen sides."
Declared in their S-1 filing that they stand with the Western liberal democracies. They do not do business with the Chinese Communist Party. They went public not as a mere SaaS vendor but as a company with a geopolitical stance — this is the root of trust with their Government customers.
BOSWELL Business Plan
02 / 05
CHAPTER02 / REVENUE ENGINE
What's Remarkable, and How They Earn
The journey of a single customer = "enter at a loss → embed into the business → 89% high margin" ─ the 3-stage model Palantir explains in its own S-1
A
ACQUIRE
1 pilot team
FDEs embed on-site under NDA
Ship "something working" to production in weeks
Enter free / low-priced
E
EXPAND
1 dept → company-wide
Workflows & decisions all route
through BOSWELL
Switching cost maxed out
S
SCALE
Cross-industry + SaaS
$24.8M / customer × 6.6-yr retention
Top 20 = 67% of revenue
SaaS long-tail
Contribution Margin Trajectory ─ The J-curve Source: Palantir S-1 (SEC, 2020) — same-cohort disclosed figures
±0% +50% +89% -50% Loss phase — pilot investment up-front High-margin phase — SaaS long-tail Deep loss ▲43% Same cohort '19 +35% ─ Break-even +55% Scale '19 +68% H1 '20 +89% Top 25%
The Core Mechanism
FDE Model
"Our forward deployed engineers have travelled to bases in Afghanistan and factories in the industrial Midwest." Where consultants "pitch and leave," their engineers embed on-site and build production. Deploy time 45 days → 14 days, government revenue +76% YoY (H1 2020).
BOSWELL Business Plan
03 / 05
CHAPTER03 / TARGET MARKET
The Market We Should Target
Japan's LDP AI White Paper 2.0 (April 2026) classifies into 3 categories the verticals where Japan should win in vertical AI
¥3T
Japan domestic vertical AI market (2030 forecast)
The white paper explicitly names three categories where Japan should leverage its strength in vertical AI.
LDP Digital Society Promotion HQ
AI White Paper 2.0 / April 2026 / p.9
"(In AI middleware) an oligopoly by a small number of foreign firms is forming. From the standpoint of securing Japan's autonomy, developing domestic alternatives is an urgent priority."─ AI White Paper 2.0 p.9 (= a statement explicitly aimed at Palantir Foundry)
01 ─ INDUSTRIAL COMPETITIVENESS

Tech & Industry Domain

Manufacturing Finance Drug Discovery Science & Tech
Japan's "field knowledge" and "tacit knowledge" are the single largest source of differentiation. "We must asset-ize field knowledge and build an AI nation that can win inside industry." (white paper p.4)

Our target: traditional enterprises that are extremely behind on AI investment. The more a company has outsourced everything to SIers, has failed to build in-house, and lets tacit floor knowledge die before reaching the boardroom — the bigger the marginal value of FDE-embedded BOSWELL.
▶ The domain Palantir Foundry conquered
02 ─ SOCIAL FUNCTION CONTINUITY

Population-Decline Response

Logistics Healthcare Eldercare Education Construction
Problem-frontier nation = opportunity-frontier nation. "Precisely because Japan is a problem-frontier nation, it has both the need and the potential to lead the world in the social implementation of agentic AI." (white paper p.2)

DOU is already pioneering AI-instructor deployment in education. The same scheme (embed on-site → accumulate knowledge → roll out horizontally) is directly transferable to healthcare, eldercare, and construction.
▶ A massive market unique to Japan (U.S. Palantir does not have this)
03 ─ PUBLIC SERVICES

Government & National Security

Defense Disaster Response Public Admin
High security = where FDE bites hardest. "The transition to AI- and data-centric warfare." (white paper p.16). Fast-path procurement for startup entry is explicitly stated.
▶ Palantir Gotham's heartland = our main battlefield
OUR FOCUS
GOV & DEFENSE
The asset stack of Joi and Niinami (government access, international networks, geopolitical reach) maps directly onto 03 Government & National Security. Just as Palantir conquered the world on a single axis — "military" — we too commit to one theme. Not everything, just this. That is the essence of the Palantir model.
BOSWELL Business Plan
04 / 05
CHAPTER04 / OUR MODEL
BOSWELL Business Model
Reproduce Palantir's 3-stage model with "Japan's field × Vibe Coding"
A

Acquire ─ Embed CTO / FDE to win the field

Embed on-site under NDA with Vibe Coding. Surface "waste, irregularity, overload." Ship the first working solution to production in 3–6 months. Show decision-makers something working at 1/10th the speed of an SIer.

E

Expand ─ Become the org's context OS

Department-level win → company-wide rollout. Accumulate field knowledge, decisions, and workflows inside BOSWELL. As Palantir proved, once a system becomes operational infrastructure, no one rips it out (switching cost maxed out).

S

Scale ─ Cross-industry rollout + SaaS licensing

Take one flagship customer's success story to its industry peers. Brand as "the BOSWELL of [X] industry." Target the same economic structure as Palantir's Scale contribution margin of 68% (top 25% at 89%).

Revenue Model ─ 3-Layer Structure
01
Talent fee (FDE embed)
¥0.6–1.0M / mo
02
Initial build (one-off)
¥2–5M
03
Monthly maintenance (long-tail)
¥0.3–0.8M / mo
LTV per customer ─ 5-year basis
~ ¥36M
10 customers = ¥360M / 30 customers = ¥1.08B / SaaS layer extra

⚡ Our only weapon ─ VIBE CODING

Business systems that take SIers 6 months — done in 1 week. Decision-makers see something working in the room and call GO / NO-GO on the spot. This is the only reason we beat both consultants and SIers. We implement the same principle as Palantir's 14-day deploy — in a Japan-spec, AI-coding-era form.

BOSWELL Business Plan
05 / 05