IV. AI report card: Features vs. Fundamentals
AI companies are shipping features at full speed while the fundamentals that can drive profitability — adoption, trust, governance, cost and revenue — remain unresolved.
| Feature | Impact on bottom line |
|---|---|
| Multimodal inputs and outputs | |
| Text, image, voice, and mixed-media interaction improvements. | Improves demos and perceived capability, but shows no evidence of increased paid retention, pricing power or margin expansion. |
| Larger context windows | |
| Expanded in-session memory and longer prompts. | Reduces hallucinations and improves short-term trust, but increases compute cost and remains vulnerable without 100-percent loss-less memory across sessions. |
| UI / UX refinements and faster responses | |
| Interface polish, latency reduction and onboarding improvements. | Lowers friction but does not create durable usage, pricing leverage or long-term revenue. |
| Third-party integrations | |
| Connections to external apps, tools and workflows. | Expands surface area without generating revenue. At least one major player admits challenges. |
| “Memory-like” conveniences | |
| Lightweight recall of prior interactions or preferences. | Cosmetic persistence; no compounding value or iterative learning without 100% loss-less memory. |
| Wearable and health data integrations | |
| Ingestion of biometric or behavioral data from external devices. | Produces data without judgment; increases liability and compliance risk faster than revenue. |
| Safety tooling and content labels | |
| Filters, classifiers and policy enforcement layers. | Defensive cost center required to mitigate risk, not a revenue driver. |
| Image and video generation upgrades | |
| Higher-quality or faster media generation. | Commoditized parity features with limited differentiation or pricing power. |
| Coding assistants and copilots | |
| AI-assisted code generation and review. | Useful episodically, but weak long-term retention and high human-oversight cost. |
Fundamentals that can drive profitability
| Core functionality | Impact on bottom line |
|---|---|
| Sustained consumer adoption | |
| Habitual, repeat usage that persists beyond novelty. | Adoption rates are flat since Spring 2025. Potential revenue stream remains unrealized. |
| Enterprise-grade integration | |
| Reliable deployment into production enterprise workflows. | Remains unrealized; approximately 95 percent of enterprise deployments fail. |
| Capital efficiency / capex discipline | |
| Alignment between compute spend and revenue growth. | Compute spend, driven by ever-increasing features, continues to outpace revenue growth; margins structurally compressed. |
| Durable revenue model | |
| Pricing that scales profitably with usage. | Flat-rate subscriptions and token-based billing misaligned with actual underlying cost structure. |
| 100% loss-less memory across sessions | |
| Complete, persistent recall without context loss. | Required to eliminate hallucination risk and enable compounding value; absent today. |
| Trustable outputs | |
| Reliable refusal and truthfulness under uncertainty. | Partial trust is not bankable trust. |
| Governed restraint | |
| Ability to defer or decline action when appropriate. | Missing; exposes enterprises to costly errors. |
| Accountability and auditability | |
| Traceable decisions and reviewable outputs. | Required for regulated environments; largely unsolved. |
| Compute optimization | |
| Optimization at input and provisioning before execution to prevent unnecessary or low-value compute. | Without optimization at input and provisioning before execution, compute is consumed blindly. Costs scale with usage, not value, preventing margin expansion. Absent today. |
| User-level governance | |
| User-defined control over tools, behavior, preferences and operating boundaries. | Essential for adoption of any tool; remains unrealized. |
– Published Wednesday, January 7, 2026