Add edge compute offload — inference tokens served on-device bypass D… - #93
Merged
Merged
Conversation
…C supply chain The single biggest counter-thesis to the "Energy Wall" forecast: model distillation (Llama-4-Small, Gemini Nano, Phi) running on Apple Neural Engine / Snapdragon NPU means those tokens never hit the grid. Implementation: - New `edgeOffload` parameter per demand block: fraction of inference tokens served on-device rather than in datacenter GPUs (per segment: consumer, enterprise, agentic) - In computeRequiredGpus: datacenter tokens = total × (1 - edgeShare), offloaded tokens bypass GPU/transformer/cooling/grid pipeline entirely - Adoption curve ramps across time blocks: Consumer: 2% → 5% → 12% → 22% → 35% → 50% → 60% → 65% Enterprise: 0% → 1% → 3% → 8% → 15% → 25% → 35% → 40% Agentic: 0% → 0% → 1% → 2% → 5% → 10% → 15% → 20% - Edge offload table added to Assumptions UI (user-editable per block) - Results track offloaded tokens and per-segment shares for charting All values are user-tunable — drag the consumer edge share to 80% to model an aggressive "Apple Intelligence wins" scenario, or 0% to model "cloud stays dominant." https://claude.ai/code/session_01PHK9qmguUSTyvykAoho467
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
…C supply chain
The single biggest counter-thesis to the "Energy Wall" forecast: model distillation (Llama-4-Small, Gemini Nano, Phi) running on Apple Neural Engine / Snapdragon NPU means those tokens never hit the grid.
Implementation:
edgeOffloadparameter per demand block: fraction of inference tokens served on-device rather than in datacenter GPUs (per segment: consumer, enterprise, agentic)Consumer: 2% → 5% → 12% → 22% → 35% → 50% → 60% → 65%
Enterprise: 0% → 1% → 3% → 8% → 15% → 25% → 35% → 40%
Agentic: 0% → 0% → 1% → 2% → 5% → 10% → 15% → 20%
All values are user-tunable — drag the consumer edge share to 80% to model an aggressive "Apple Intelligence wins" scenario, or 0% to model "cloud stays dominant."
https://claude.ai/code/session_01PHK9qmguUSTyvykAoho467