Canonical Observation Case
Cross-Observer / Kosuke Protocol
OBSERVED · signal emerging
AI Capital Gravity
AI資本重力
AI does not only create productivity. It also creates capital gravity.
AIは生産性を生むだけではない。 資本を引き寄せる重力も生む。
Productivity story and capital-gravity story, held together.
- Macro
- Infrastructure
- Capital Markets
- Industrial Policy
Horizon: now · 1–3 years · 3–10 years
Scroll
- AI
- PRODUCTIVITY
Core Thesis
The AI boom may be raising the cost of everything that is not AI.
AIブームは、AI以外のものを高くしているかもしれない。
Final Question
Does AI lift the whole economy, or first pull capital, power, land, and talent toward itself?
AIは経済全体の生産性を押し上げるのか。それとも、その前に他産業の資本コストを押し上げるのか。
Starting Observation
各観測は Observed / Inferred / Speculative と source type で分離する。本文へ金額などの数値を固定しない。
- Observedmarket reporting
Hyperscaler and data-center investment has become a first-order capital-markets story, not only a technology story.
- Observedenergy / grid
Grid interconnection, electricity price, and siting constraints appear alongside model launches.
- Inferredstructural reading
If the same scarce inputs are bid up, non-AI capital projects can be delayed without any firm intending crowding-out.
- Speculativeopen question
Whether measured economy-wide productivity arrives before, with, or after this gravity remains unresolved.
No probability is attached.
Central Model
対立として断定しない。相互作用として読む。
Productivity force
- Model capability rises↓
- Task cost falls in some domains↓
- Firms report efficiency gains↓
- The story becomes: AI lifts the economy
PRODUCTIVITY
VS
CAPITAL GRAVITY
Capital gravity force
- Training and inference demand infrastructure↓
- Capital, power, land, and talent concentrate↓
- Non-AI projects compete for the same scarce inputs↓
- The story becomes: AI reprices everything else
Definition
AI Capital Gravity
AI Capital Gravity is the tendency of large-scale model training and inference infrastructure to attract capital, electricity, land, construction, and technical talent — and, in doing so, to reprice or delay investment outside that cluster.
大規模な学習・推論インフラが、資金・電力・土地・建設・技術人材を引き寄せ、その外側の投資を再価格化または遅延させる傾向。
単なる financial crowding out より広い概念。
- capital
- electricity
- land
- grid interconnection
- construction capacity
- technical talent
- policy attention
Market Signals
- Inferred
Cost of capital
If AI infrastructure is treated as a must-fund bid, real rates and hurdle rates for other projects can move with it.
The boom can reprice the rest of the book.
Which non-AI projects become unfinanceable first?
- Observed
Power and land
Data-center load is visible in utility planning and local land-use conflict.
Who loses access to cheap power and quiet land?
- Inferred
Talent gravity
Technical labor can be pulled into the same cluster that is already pulling capital.
What happens to industrial and public-sector technical capacity?
- Speculative
Allocation politics
Governments may have to choose between accelerating AI infrastructure and protecting other capital formation.
Is industrial policy now a capital-allocation dispute?
Crowding-Out Map
中央の AI Infrastructure Demand から資源へ、さらに非AI領域へ competition edge を伸ばす。因果証明ではなく観測上の接続。
Macro Scenarios
v1 は qualitative scenario のみ。probability は付けない。
Scenario A
Productivity Renaissance
Measured gains arrive fast enough to fund the infrastructure bill.
- Gravity exists but is paid down by surplus
- Non-AI investment recovers after a lag
Scenario B
Capital Scarcity
Infrastructure demand stays high while productivity remains localized.
- Hurdle rates rise outside the AI cluster
- Public and industrial projects wait
Scenario C
Bubble / Reversal
Capacity is built ahead of durable demand.
- Stranded power and land bets
- A hangover that still leaves grid constraints
Scenario D
Uneven Geography
Gains and costs concentrate in different places.
- Some regions get data centers and expensive housing
- Others get neither productivity nor infrastructure
Entangled Society
Finance chain
- AI infrastructure bid↓
- higher real rates / tighter credit screens↓
- non-AI capex delayed↓
- weaker real-economy investment
Energy chain
- inference and training load↓
- grid interconnection queue↓
- power price and siting conflict↓
- other electrification delayed
Two chains can run at once. Productivity and gravity are not a vote; they are coupled pressures.
Employment · Talent Gravity
Talent Gravity
The same cluster that attracts capital can attract the people who would otherwise staff factories, utilities, hospitals, and public systems.
- Frontier labs and infra vendors bid for talent↓
- Industrial and civic technical roles thin out↓
- The places that need electrification and maintenance wait
Industrial Policy
Allocation Politics
Once AI infrastructure is treated as national or corporate destiny, other capital formation becomes a political remainder.
Who is allowed to lose the bid for power, land, and engineers?
- grid queues
- municipal siting fights
- industrial strategy documents
- public-infrastructure delays
This is an observational target, not a completed policy claim.
AI Productivity Signals
- task-level cost decline in software and knowledge work
- reported headcount leverage in some firms
- new product classes that did not exist five years ago
AI Capital Gravity Signals
- data-center capex as a market headline
- utility load-growth revisions
- land and interconnection as scarce assets
- talent concentration in a few metros and firms
Balance Indicator
PRODUCTIVITY ————————— CAPITAL GRAVITY
Productivity dominant
Balanced
Gravity dominant
Unclear
Unclear — both forces are visible; neither is settled.
The indicator is qualitative. It is not a score of the world economy.
Capital Gravity Index — Experimental
Level: experimental
Inputs are reserved. No numeric index is published as a fact about the economy.
score: n/a · inputs reserved for future observation
Key Visual
PRODUCTIVITY
CAPITALGRAVITY
THE AIECONOMY
スクロールで比重が変化する。勝敗の判定ではない。
Cross-Observer
Market Signals ↔ Entangled Society ↔ Employment ↔ Energy / Infrastructure ↔ Industrial Policy ↔ Kosuke Protocol
Market Signals
Observed- Observation
- AI is sold as productivity. Capital markets also treat it as an infrastructure bid.
- Structural Shift
- A technology story becomes a capital-allocation story.
- Question
- What must firms now buy besides the model?
↓
Entangled Society
Inferred- Observation
- Rates, power, land, and talent are coupled to the same demand cluster.
- Structural Shift
- Second-order costs appear outside the firms that capture the gains.
- Question
- Which public goods wait?
↓
Employment
Inferred- Observation
- Talent gravity can thin the technical staff of non-AI sectors.
- Structural Shift
- Labor substitution and labor concentration can happen together.
- Question
- Where do industrial skills go?
Cross-case
Two observations of the same machine
The Company Without Headcount
- Organizational capability detaches from headcount
- Human-light firms still need compute, power, and capital
AI Capital Gravity
- Those remaining inputs are scarce and bid-up
- Human-light can still be capital-heavy
A firm can have few people and still pull a large share of society’s capital.
人は少なくても、社会の資本を大きく引き寄せうる。
The two pages are complementary, not duplicates.
→ The Company Without HeadcountObservation Patterns
Dual force
Hold productivity and gravity as interacting pressures, not a binary.
Input competition
Watch the scarce inputs, not only the model card.
Opportunity Layer
What a buyer may need to see
Targets
- CFO / treasury
- infrastructure investors
- utilities and grid operators
- industrial strategy / municipalities
Evaluate
- interest-rate exposure via AI infra bid
- power and land competition
- talent drain from core operations
- delayed public or industrial capex
BtoB Proposal Example
A non-AI industrial firm whose cost of capital and power is rising
The board is briefed on AI productivity. It is not briefed on whether the boom is repricing its own investment plan.
AI Capital Exposure Assessment
- Map: Interest, power, land, talent, and delayed projects.
- Scenario: Qualitative renaissance / scarcity / bubble / uneven.
- Hold: What the firm will not fund while the bid continues.
Kosuke Protocol
Observe
The discomfort is not only job loss. It is that other projects quietly wait.
違和感は失業だけではない。他の計画が静かに待たされることだ。
↓
Sample
Who feels the bid: utilities, municipalities, manufacturers, hospitals, SMEs.
↓
Question
Is this a productivity boom, a capital-allocation boom, or both?
↓
Meaning
A society that can train models but cannot fund a hospital expansion has already chosen.
モデルは訓練できても病院を増設できないなら、それはすでに選択である。
↓
HOLD
Do not convert this into a market-size forecast. Keep it as an exposure map.
Headline Insight
Technology does not only create value.It changes what a society can still afford to fund.
技術は価値を生むだけではない。社会が何に投資できるか、その配分そのものを変える。
Research Questions
- 1. Which non-AI capex is most sensitive to a rise in real rates driven by infra bids?
- 2. Where do grid queues and housing costs move together with data-center siting?
- 3. Does talent gravity show up in industrial vacancy before it shows up in wages?
Observatory Links
- Entangled Society / 絡み合う社会
Simulate second-order movement of rates, power, and public goods.
- The Company Without Headcount
Organizational leverage; this page is the capital cost of that leverage.
Do Not Overclaim
- Do not treat crowding-out as proven for a named country or year.
- Do not publish a numeric gravity index as if it were a statistic.
- Do not attach probabilities to the four scenarios.
- observed
- inferred
- speculative
- hold
Entangled Society
SIMULATE CONSEQUENCES
このシグナルが他の社会領域へどう波及するかを見る
SIMULATE CONSEQUENCES →AI does not only create productivity.It also creates capital gravity.
AIは生産性を生むだけではない。資本を引き寄せる重力も生む。