This view keeps each lane tied to the dominant AI supplier weakness so leadership can reduce the right dependency first.
| Lane | Tier | Dimension | Risk headline | Concentration signal | Blocking issue | Concentration | Switching cost | Contract exposure | Fallback readiness | Compliance dependency | Platform coupling |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Foundation model control plane | LOCKED_IN | MODEL_LOCK_IN | Model strategy is moving faster than vendor diversification. | Critical routes still assume one provider's latency, tool semantics, and release cadence. | The fallback model path is not strong enough to absorb a commercial or reliability shock. | 86 | 82 | 71 | 39 | 63 | 84 |
| Cloud AI bundle and hosting estate | CONCENTRATED | INTEGRATION_DEPTH | Platform leverage is improving while switching leverage is getting worse. | The same provider keeps winning adjacent layers because procurement and technical handoffs reward the bundle. | The team lacks one clean isolation boundary for moving AI workloads without pulling infra and identity along with them. | 79 | 76 | 69 | 48 | 57 | 81 |
| Regulated disclosure and policy operations | WATCHLIST | COMPLIANCE_DEPENDENCY | Proof portability is weaker than the public trust story suggests. | Policy exports, audit evidence, and disclosure packets still map best to one provider's control framework. | The estate lacks one portable control vocabulary that survives outside the current trust vendor contract. | 68 | 64 | 58 | 56 | 77 | 52 |
| Data pipelines and model fine-tuning estate | CONCENTRATED | DATA_EGRESS | Portability remains weaker than procurement leverage. | The migration packet still requires too much format cleanup and too much model-specific adaptation work. | There is no one durable export contract for prompts, evals, and supporting data flows. | 74 | 83 | 54 | 44 | 49 | 73 |
| Revenue-facing AI experience estate | WATCHLIST | PRICING_POWER | Margin quality is improving while supplier leverage is slipping. | Growth workflows keep adding higher-value use cases faster than multi-vendor negotiating leverage is improving. | The commercial team still lacks one clear downgrade or alternate-provider story for key AI-enhanced experiences. | 71 | 68 | 79 | 51 | 43 | 61 |
| Board and investor intelligence estate | WATCHLIST | MULTI_VENDOR_GAP | Narrative diversification is ahead of operational diversification. | Too many leadership surfaces can describe supplier options but cannot yet demonstrate clean substitution under pressure. | There is no one recurring drill packet that proves a concentrated lane can survive provider disruption or repricing. | 65 | 59 | 52 | 46 | 41 | 57 |