Both maps conclude that application-layer and model-serving vendors face severe margin compression due to the economics of inference and rapid model commoditization.

Did AI Value Capture Pool Almost Entirely in Compute and Cloud Infrastructure?, Is Open-Source AI Shifting the Primary Economic Surplus Directly to End Users? agree on this point.

Both maps document dramatic deflationary trends in per-unit compute and token serving costs over short operational horizons.

Did AI Value Capture Pool Almost Entirely in Compute and Cloud Infrastructure?, Is Open-Source AI Shifting the Primary Economic Surplus Directly to End Users? agree on this point.

The maps diverge on whether the primary economic surplus of AI concentrates upstream among semiconductor monopolies or downstream among end users.

Did AI Value Capture Pool Almost Entirely in Compute and Cloud Infrastructure?: For Q1 FY27, Nvidia reported net income of $58.3 billion on revenue of $81.6 billion, with GAAP gross margin rising to 74.9% from 60.5% YoY. Is Open-Source AI Shifting the Primary Economic Surplus Directly to End Users?: Brynjolfsson et al. concluded that generative AI consumer surplus substantially exceeds estimated producer revenues in the United States.

Families

Who Actually Benefits From AI?

2 investigations · 2 maps · 76 claims · Last updated 3d ago

LinkedIn X Facebook WhatsApp Email

Did AI Value Capture Pool Almost Entirely in Compute and Cloud Infrastructure? (Markdown)Is Open-Source AI Shifting the Primary Economic Surplus Directly to End Users? (Markdown)

Overview

Who Actually Benefits From AI?: 2 public investigations and 76 claims.

Map overviews

Sort

View

4
Did AI Value Capture Pool Almost Entirely in Compute and Cloud Infrastructure?
submitted 3 days ago by @djwide 75
40 claims
OverviewThis map analyzes whether the economic gains of the generative AI boom have pooled almost exclusively in hardware vendors, advanced foundries, and specialized cloud infrastructure. It builds its case using financial statements, foundry capacity constraints, and venture analyses showing widening revenue shortfalls for software applications. The overarching argument is that compute providers capture outsized margins w…
2
Is Open-Source AI Shifting the Primary Economic Surplus Directly to End Users?
submitted 3 days ago by @djwide 75
36 claims
OverviewThis map investigates whether open-weight architectures, algorithmic efficiency, and steep token deflation redirect the primary economic surplus directly to enterprises and end consumers. It synthesizes econometric welfare evaluations, enterprise survey data, and labor productivity benchmarks. It contends that rapid model commoditization suppresses producer rents and unlocks massive downstream utility for workers an…

Last generated Oct 22, 2026, 4:56 PM by a member

More information

Compared Oct 22, 2026, 4:56 PM

Maps

Open map

Comparing maps…

Themes across the family

A broad read of every map in this family: the themes that run through them and how each map treats them. Structured scans give a more targeted side-by-side.

Hardware Monopoly and Infrastructure Rents
2 of 2 maps
The maps evaluate whether physical compute providers and semiconductor foundries absorb the bulk of financial proceeds generated across the AI ecosystem.
M1Did AI Value Capture Pool Almost Entirely in Compute and Cloud Infrastructure?
Demonstrates that chipmakers and foundries achieved historic net margins and revenue growth driven by non-negotiable compute demand.
M2Is Open-Source AI Shifting the Primary Economic Surplus Directly to End Users?
Highlights that hardware monopolies face valuation shocks when algorithmic efficiency breakthroughs reduce compute requirements.
Model Commoditization and Margin Compression
2 of 2 maps
Both maps track how aggressive competition, rapid token price deflation, and infrastructure overhead hollow out margins for intermediate model developers and software vendors.
M1Did AI Value Capture Pool Almost Entirely in Compute and Cloud Infrastructure?
Shows that model developers and application startups suffer steep margin degradation and operating losses due to ongoing inference costs.
M2Is Open-Source AI Shifting the Primary Economic Surplus Directly to End Users?
Argues that open-weight model availability directly commoditizes the intelligence layer and prevents proprietary providers from defending profit moats.
Downstream Surplus and Productivity Capture
2 of 2 maps
The research family weighs whether the enduring economic dividend of AI accrues to end users and enterprises via operational productivity rather than vendor equity.
M2Is Open-Source AI Shifting the Primary Economic Surplus Directly to End Users?
Maintains that aggregate consumer surplus and worker productivity improvements substantially exceed total producer revenues.
M1Did AI Value Capture Pool Almost Entirely in Compute and Cloud Infrastructure?
Notes enterprise adoption and revenue run-rates while focusing on whether end-market value can scale fast enough to support infrastructure costs.
Inference Economics and Operational Realities
2 of 2 maps
The mechanics of model serving dictate operational feasibility, driving architecture choices between hosted APIs and private deployments.
M1Did AI Value Capture Pool Almost Entirely in Compute and Cloud Infrastructure?
Identifies growing inference spend as an existential drag that erodes traditional software gross margins as products scale.
M2Is Open-Source AI Shifting the Primary Economic Surplus Directly to End Users?
Analyzes how self-hosting operational overhead balances against volume-based cost savings and query routing techniques.

Both maps conclude that application-layer and model-serving vendors face severe margin compression due to the economics of inference and rapid model commoditization.
M1Did AI Value Capture Pool Almost Entirely in Compute and Cloud Infrastructure?
C0017 AI product gross marginsKnown correct
Average AI product gross margins dropped to 52% (compared to 70–80% for traditional SaaS), with inference costs accounting for roughly 23% of total revenue at scaling companies.
InferredLow significanceClaude
M2Is Open-Source AI Shifting the Primary Economic Surplus Directly to End Users?
C0005 Open-weight gateway token shareKnown correct
Vercel's July 2026 AI Gateway Production Index showed open-weight models executed 29% of June gateway tokens while representing under 4% of total customer spend.
InferredLow significanceClaude
Both maps document dramatic deflationary trends in per-unit compute and token serving costs over short operational horizons.
M1Did AI Value Capture Pool Almost Entirely in Compute and Cloud Infrastructure?
C0033 Nvidia H100 rental pricingKnown correct
Rental pricing on spot markets for Nvidia H100 GPUs dropped from roughly $8 per hour in 2023 to under $2–$3 per hour by late 2024 or early 2025.
Source-linkedLow significanceKimi
M2Is Open-Source AI Shifting the Primary Economic Surplus Directly to End Users?
C0007 Plummeting LLM inference costsKnown correct
Inference costs for GPT-3.5-level systems fell from $20.00 per million tokens in November 2022 to $0.07 in October 2024 (Gemini-1.5-Flash-8B), a deflation factor exceeding 280x.
Source-linkedLow significanceClaude, Kimi

Build on the map here to find more

The maps diverge on whether the primary economic surplus of AI concentrates upstream among semiconductor monopolies or downstream among end users.
M1Did AI Value Capture Pool Almost Entirely in Compute and Cloud Infrastructure?
C0004 Nvidia Q1 FY27 resultsKnown correct
For Q1 FY27, Nvidia reported net income of $58.3 billion on revenue of $81.6 billion, with GAAP gross margin rising to 74.9% from 60.5% YoY.
Source-linkedLow significanceClaude
M2Is Open-Source AI Shifting the Primary Economic Surplus Directly to End Users?
C0002 Consumer surplus exceeding producer revenuesKnown correct
Brynjolfsson et al. concluded that generative AI consumer surplus substantially exceeds estimated producer revenues in the United States.
InferredLow significanceClaude

Build on the map here to find more

Did AI Value Capture Pool Almost Entirely in Compute and Cloud Infrastructure? uniquely focuses on physical supply-chain bottlenecks and debt-financed hardware models such as GPU-backed lending.
M1Did AI Value Capture Pool Almost Entirely in Compute and Cloud Infrastructure?
C0035 CoreWeave GPU debt financingKnown correct
CoreWeave held approximately $8 billion in debt secured directly by its inventory of GPUs, following a debt financing model initiated with Magnetar Capital in 2023.
InferredLow significanceKimi
Is Open-Source AI Shifting the Primary Economic Surplus Directly to End Users? uniquely examines quantified worker-level productivity benchmarks across programming and customer service tasks.
M2Is Open-Source AI Shifting the Primary Economic Surplus Directly to End Users?
C0034 Developer speed with CopilotKnown correct
A 2022 controlled experiment reported by GitHub found software developers completed tasks up to 55% faster when using GitHub Copilot.
Source-linkedLow significanceKimi

Evidence conflict0

No evidence conflict.

Claims support0

No claims are supported by claims on another map.

Shared open questions0

No shared open questions.

Shared citations0

No citation appears on more than one map.

Discussion

Tag a map with @(M1), a claim on that map with @(M1/C0001), a generating prompt with @(M1/P…), or a person with @username. Type @ to search.

No comments yet.