Software moves in quarters. Geology moves in decades. This mismatch determines the ceiling of AI scale.

1. Software Timelines: Months and Quarters

AI behaves like software: fast, iterative, compounding.

Characteristics

• Build → test → feedback → iterate
• Daily or weekly updates
• Quarterly capability jumps (Q1 to Q5 cycles)
• Mistakes are reversible
• A/B testing is trivial
• Optimization compounds quickly

This creates the illusion that all parts of the AI system move at software speed.
They do not.

2. Geological Timelines: 10–15 Years Minimum

AI’s physical foundation—minerals, metals, materials—lives on geological time.

Typical lifecycle

• Years 1–7: Environmental review
• Years 7–10: Permitting
• Years 10–12: Construction
• Years 12–15: Ramp-up
• Year 15+: Meaningful output

No iteration.
No fast feedback loops.
No A/B testing a copper mine.

Even if global demand doubles tomorrow, the supply timeline does not move.

(as per analysis by the Business Engineer on https://businessengineer.ai/p/this-week-in-business-ai-the-new)

3. Why This Mismatch Matters

AI capability depends on GPUs and datacenters.
GPUs and datacenters depend on minerals.
Minerals depend on geological time.

The constraints are absolute:

• Software speed cannot compress environmental review.
• Capital cannot compress ore formation.
• Optimization cannot compress mine development.

This is the core bottleneck shaping the next decade of AI.

4. The Strategic Consequence

The visible layer—models, apps, agents—moves fast.
The invisible layer—geology—moves slow.
The system is constrained by the slowest component.

Which means:

• AI scaling is physically bounded.
• Energy and minerals—not model architecture—determine long-run limits.
• Nations controlling minerals control AI trajectory.
• Software iteration does not overcome geological scarcity.

This mismatch is not just a detail. It is the determining factor of the AI supercycle, now unfolding on physical timelines rather than software ones.

(as per analysis by the Business Engineer on https://businessengineer.ai/p/this-week-in-business-ai-the-new)