L1 Infrastructure
Layer 1 — AI Infrastructure
93.8from 85.8+8
Projected leaders
NVIDIA99
Microsoft99
Amazon Web Services92.7
Projected for May 2028
Industry
Companion view
The AI Futures Studio.
Set a future date and the forces that get you there. Watch the landscape — and the news — project forward.
The present-state market reference architecture is the starting point. Your levers feed a feature model — optionally synthesized by an LLM reasoning about second-order effects — which a deterministic statistical model uses to project positions, metrics, and winners forward with uncertainty bands. An AI narrator then reads the numbers back as a future-state briefing. The projection is reproducible; the narrative is grounded in it.
Baseline as of 2026-05 · Updated 2026-05-28
May
2028
Horizon
2
Years out
91.6
Market index +12.3
Heuristic
Feature source
Scenario read
Scenario shaped by Model capability high, Cost decline high, SaaS disruption high.
- Model capability rises relative to the base case.
- Cost decline rises relative to the base case.
- SaaS disruption rises relative to the base case.
Projected landscape
Where the layer cake lands.
L2 Models
Layer 2 — Model Portfolio
95.7from 87.4+8.3
Projected leaders
Google98
Projected for May 2028
L3 Data Mgmt
Layer 3 — Data Management
94.3from 86+8.3
Projected leaders
Snowflake92.5
Projected for May 2028
L4 AI Data
Layer 4 — AI-Ready Data
87.1from 71.8+15.2
Projected leaders
Vercel93.2
Weaviate89.6
Projected for May 2028
L5 Dev Tools
Layer 5 — Developer Tools
92.4from 78.8+13.6
Projected leaders
Projected for May 2028
L6 Agents
Layer 6 — Agent Tools
95.5from 74.4+21.1
Projected leaders
ServiceNow96
Hermes96
Projected for May 2028
L7 Commercial
Layer 7 — Commercial Tools
88.3from 78.3+10
Projected leaders
Harvey91.8
Projected for May 2028
Headline numbers
The news, as metrics.
Critical workflows AI-operated
24%from 7%
▲Cost per governed action (index, 100 = today)
66from 100
▼Agents per human (instrumented teams)
1.6xfrom 0.3x
▲SaaS seat pricing renegotiated
22%from 6%
▲Enterprise inference on open weights
32%from 23%
▲Enterprise data agent-ready
35%from 19%
▲Trajectories
Direction of travel, with uncertainty.
NVIDIA
96 → 99L1 Infrastructure
OpenAI
94 → 98L2 Models
Databricks
88 → 96L3 Data Mgmt
Pinecone
84 → 94L4 AI Data
Cursor
77 → 95L5 Dev Tools
Anthropic
87 → 96L6 Agents
Microsoft
92 → 95L7 Commercial
Movers
Who gains, who gives ground.
Projected gainers
Zeroclaw
agent tools
+30
66 → 96
Hermes
agent tools
+26
70 → 96
Sierra
agent tools
+24.7
69 → 93.7
Vercel
ai ready data
+22.2
71 → 93.2
Harvey
commercial tools
+19.8
72 → 91.8
Cursor
developer tools
+18
77 → 95
Projected decliners
Microsoft
commercial tools
+3
92 → 95
NVIDIA
ai infrastructure
+3
96 → 99
OpenAI
model portfolio
+4
94 → 98
Amazon Web Services
ai infrastructure
+4.7
88 → 92.7
Wiz
commercial tools
+6.6
80 → 86.6
Salesforce
commercial tools
+6.7
81 → 87.7
Transparency
The math behind the move.
Why the index moved
Signed contribution of each force to the aggregate projected index. This decomposition is what makes the projection auditable: the narrative is downstream of these numbers, not the other way around.
- Model capability+0.5
- Cost decline+0.5
- Capital+0.5
- Workforce+0.2
- Data governance+0.2
- Compute supply+0.1
- SaaS disruption+0
- Open weights+0
- Regulation+0
- Sovereignty+0
The future briefing
Generate the May 2028 briefing.
The dashboard above is the grounded projection. Generate a long-form narrative that reads the numbers in the voice of The State of AI & Data — written for the executive lens, grounded in the projected metrics, and shaped by the active levers.
How the projection works.
- Levers are the premise. A horizon up to five years out, ten forces, assumption shocks, and constraints. Presets set a coherent starting mix.
- Emergent features feed the model. A heuristic translates levers into features directly; the AI pass adds second-order reasoning — category growth multipliers, shocks, and vendor biases the linear model cannot see.
- A statistical model grounds the numbers. Damped-trend extrapolation over the real position history, force modifiers, probability-weighted shocks, and seeded Monte Carlo for p10/p90 bands. Same levers, same numbers, every time.
- The narrative is downstream. The briefing cites the projected metrics and movers; it never invents numbers that contradict the model.
Set the levers. Project the future. Read the news.
AI-readable exports.
Exports accept the same ?s= scenario parameter as the studio, so any projected future is reproducible and shareable as data.