Explore

Slice the corpus on any axis.

Filter by firm, year, function, metric or slide category — then see the result as a narrative arc, a treemap of slide types, or a catalog of slides.

clear
51,416 slides across 1229 decks match.
Treemap Arc Catalog
Data Center Electricity Consumption by Region = USA Leads, per IEA
BondCap · industry_trends
Data Center Electricity Consumption by Region = USA Leads, per IEA
400 — Electricity consumption (TWh)
 AI Model Compute Costs High / Rising + Inference Costs Per Token Falling = Performance Converging + Developer Usage Rising
BondCap · context
AI Model Compute Costs High / Rising + Inference Costs Per Token Falling = Performance Converging + Developer Usage Rising
 AI Model Compute Costs High / Rising + Inference Costs Per Token Falling = Performance Converging + Developer Usage Rising
BondCap · problem_statement
AI Model Compute Costs High / Rising + Inference Costs Per Token Falling = Performance Converging + Developer Usage Rising
 AI Model Training Compute Costs = ~2,400x Growth Over Eight Years, per Epoch AI & Stanford
BondCap · market_landscape
AI Model Training Compute Costs = ~2,400x Growth Over Eight Years, per Epoch AI & Stanford
~2,400x — Training Cost (USD)
 AI Model Compute Costs High / Rising + Inference Costs Per Token Falling = Performance Converging + Developer Usage Rising
BondCap · diagnosis
AI Model Compute Costs High / Rising + Inference Costs Per Token Falling = Performance Converging + Developer Usage Rising
 AI Model Compute Costs High / Rising + Inference Costs Per Token Falling = Performance Converging + Developer Usage Rising
BondCap · context
AI Model Compute Costs High / Rising + Inference Costs Per Token Falling = Performance Converging + Developer Usage Rising
 AI Inference ‘Currency’ = Tokens
BondCap · context
AI Inference ‘Currency’ = Tokens
1MM — token count
 AI Inference Costs – NVIDIA GPUs = -105,000x Decline in Energy Required to Generate Token Over Ten Years
BondCap · data_table
AI Inference Costs – NVIDIA GPUs = -105,000x Decline in Energy Required to Generate Token Over Ten Years
105,000x — Energy consumption per token
 AI Inference Costs – Serving Models = 99.7% Lower Over Two Years, per Stanford HAI
BondCap · market_landscape
AI Inference Costs – Serving Models = 99.7% Lower Over Two Years, per Stanford HAI
99.7% — Inference price per million tokens
 AI Cost Efficiency Gains = Happening Faster vs. Prior Technologies
BondCap · industry_trends
AI Cost Efficiency Gains = Happening Faster vs. Prior Technologies
100% — Relative Cost
 Tech’s Perpetual A-Ha = Declining Costs + Improving Performance → Rising Adoption...
BondCap · industry_trends
Tech’s Perpetual A-Ha = Declining Costs + Improving Performance → Rising Adoption...
300M+ — USA Internet Users
 ...Tech’s Perpetual A-Ha = Prices Fall + Performance Rises
BondCap · industry_trends
...Tech’s Perpetual A-Ha = Prices Fall + Performance Rises
+360% — AI Model Training Compute
 AI Model Compute Costs High / Rising + Inference Costs Per Token Falling = Performance Converging + Developer Usage Rising
BondCap · framework_other
AI Model Compute Costs High / Rising + Inference Costs Per Token Falling = Performance Converging + Developer Usage Rising
 AI Model Performance = Converging Rapidly, per Stanford HAI
BondCap · industry_trends
AI Model Performance = Converging Rapidly, per Stanford HAI
1,385 — LMSYS Arena Score
 AI Model Compute Costs High / Rising + Inference Costs Per Token Falling = Performance Converging + Developer Usage Rising
BondCap · problem_statement
AI Model Compute Costs High / Rising + Inference Costs Per Token Falling = Performance Converging + Developer Usage Rising
 AI Model Compute Costs High / Rising + Inference Costs Per Token Falling = Performance Converging + Developer Usage Rising
BondCap · key_messages
AI Model Compute Costs High / Rising + Inference Costs Per Token Falling = Performance Converging + Developer Usage Rising
99.7% — cost-per-token
 The proliferation of foundation models is driving a flywheel of developer-led infrastructure growth, reducing friction and accelerating the time from idea to product.
BondCap · key_messages
The proliferation of foundation models is driving a flywheel of developer-led infrastructure growth, reducing friction and accelerating the time from idea to product.
 AI Tool Adoption by Developers = 63% vs. 44% Y/Y
BondCap · industry_trends
AI Tool Adoption by Developers = 63% vs. 44% Y/Y
63% — Share of Developers
 AI Developer Repositories – GitHub = ~175% Increase Over Sixteen Months
BondCap · industry_trends
AI Developer Repositories – GitHub = ~175% Increase Over Sixteen Months
175% — Cumulative Number of AI Repositories
 AI Developer Ecosystem – Google = +50x Monthly Tokens Processed Y/Y
BondCap · case_study
AI Developer Ecosystem – Google = +50x Monthly Tokens Processed Y/Y
480T — Monthly Tokens Processed
 AI Developer Ecosystem – Microsoft Azure AI Foundry = +5x Quarterly Tokens Processed Y/Y
BondCap · market_landscape
AI Developer Ecosystem – Microsoft Azure AI Foundry = +5x Quarterly Tokens Processed Y/Y
100T — Quarterly Tokens Processed
 AI Developer Use Cases = Broad & Varied
BondCap · other
AI Developer Use Cases = Broad & Varied
 AI Model Compute Costs High / Rising + Inference Costs Per Token Falling = Performance Converging + Developer Usage Rising ...(Likely) Long Way to Profitability
BondCap · problem_statement
AI Model Compute Costs High / Rising + Inference Costs Per Token Falling = Performance Converging + Developer Usage Rising ...(Likely) Long Way to Profitability
 Al Usage + Cost + Loss Growth = Unprecedented
BondCap · context
Al Usage + Cost + Loss Growth = Unprecedented
 Technology Disruption Pattern Recognition = Hundreds of Years of Consistent Signals
BondCap · context
Technology Disruption Pattern Recognition = Hundreds of Years of Consistent Signals
 AI-Related Monetization = Very Robust Ramps
BondCap · key_messages
AI-Related Monetization = Very Robust Ramps
 To understand the evolution of AI hardware strategy, we’ll look at how control over chip design is shifting from traditional vendors to the platforms that rely on them.
BondCap · context
To understand the evolution of AI hardware strategy, we’ll look at how control over chip design is shifting from traditional vendors to the platforms that rely on them.
 Custom chips also reflect a broader effort to manage the economics of AI infrastructure.
BondCap · context
Custom chips also reflect a broader effort to manage the economics of AI infrastructure.
 AI Monetization = Chips
BondCap · section_divider
AI Monetization = Chips
 AI Monetization...Chips = NVIDIA Quarterly Revenue +78% to $39B Y/Y...
BondCap · financial_analysis
AI Monetization...Chips = NVIDIA Quarterly Revenue +78% to $39B Y/Y...
$39B — Quarterly Revenue
 ...AI Monetization...Chips = NVIDIA Revenue +28x Over Ten Years...Big Six CapEx + R&D +6x
BondCap · financial_analysis
...AI Monetization...Chips = NVIDIA Revenue +28x Over Ten Years...Big Six CapEx + R&D +6x
28x — Revenue growth
 AI Monetization...Chips = Google TPU Sales* +116% to $8.9B Y/Y, per Morgan Stanley
BondCap · market_sizing
AI Monetization...Chips = Google TPU Sales* +116% to $8.9B Y/Y, per Morgan Stanley
$8.9B — Google TPU Sales
 AI Monetization...Chips = Amazon AWS Trainium* Sales +216% to $3.6B Y/Y, per Morgan Stanley
BondCap · market_sizing
AI Monetization...Chips = Amazon AWS Trainium* Sales +216% to $3.6B Y/Y, per Morgan Stanley
$3.6B — Estimated Annual Sales
 AI Monetization = Compute Services
BondCap · section_divider
AI Monetization = Compute Services
 AI Monetization...Cloud Computing = CoreWeave Revenue +730% to $1.9B Y/Y
BondCap · case_study
AI Monetization...Cloud Computing = CoreWeave Revenue +730% to $1.9B Y/Y
$1.9B — Revenue
 AI Monetization...AI Infrastructure = Oracle Revenue +50x to $948MM Over Two Years
BondCap · case_study
AI Monetization...AI Infrastructure = Oracle Revenue +50x to $948MM Over Two Years
$948MM — Revenue
 AI Monetization...Infrastructure Connectivity = Astera Labs Revenue +242% to $396MM Y/Y
BondCap · financial_analysis
AI Monetization...Infrastructure Connectivity = Astera Labs Revenue +242% to $396MM Y/Y
$396MM — Revenue
 AI Monetization...Data Collection + Supercomputing = Tesla AI Training Capacity +8.5x
BondCap · case_study
AI Monetization...Data Collection + Supercomputing = Tesla AI Training Capacity +8.5x
8.5x — H100-Equivalent GPUs
 AI Monetization = Data Layer
BondCap · transition
AI Monetization = Data Layer
 AI Monetization...Data Labeling & Evaluation = Scale AI Revenue +160% to $870MM Y/Y
BondCap · case_study
AI Monetization...Data Labeling & Evaluation = Scale AI Revenue +160% to $870MM Y/Y
$870MM — Revenue
 AI Monetization...Data Storage / Management / Processing = VAST Data Lifetime Sales From 0 to $2B in Just Over Six Years
BondCap · case_study
AI Monetization...Data Storage / Management / Processing = VAST Data Lifetime Sales From 0 to $2B in Just Over Six Years
$2B — Cumulative Lifetime Sales
 AI-Related Cost Ramps Relative to Revenue = Can Be Head-Turning
BondCap · section_divider
AI-Related Cost Ramps Relative to Revenue = Can Be Head-Turning
 AI Monetization – OpenAI = Revenue vs. Compute Expense, per The Information
BondCap · financial_analysis
AI Monetization – OpenAI = Revenue vs. Compute Expense, per The Information
 AI Monetization – Microsoft / Amazon / Alphabet / Meta = CapEx Up…Free Cash Flow Margins Down
BondCap · financial_analysis
AI Monetization – Microsoft / Amazon / Alphabet / Meta = CapEx Up…Free Cash Flow Margins Down
+58% — CapEx
 So...We Have... High Revenue Growth + High Cash Burn + High Valuations + High Investment Levels = Good News for Consumers... Others TBD...
BondCap · key_messages
So...We Have... High Revenue Growth + High Cash Burn + High Valuations + High Investment Levels = Good News for Consumers... Others TBD...
 Select Private AI Model Companies – 5/13/25 = ~$11B+ Annualized Revenue vs. ~$95B Raised…
BondCap · data_table
Select Private AI Model Companies – 5/13/25 = ~$11B+ Annualized Revenue vs. ~$95B Raised…
$11B — Annualized Revenue
 ...Select Private AI Model Companies – 5/13/25 = High Valuation-to-Revenue Multiples
BondCap · financial_analysis
...Select Private AI Model Companies – 5/13/25 = High Valuation-to-Revenue Multiples
75x — Revenue Multiple
 Valuation-to-Revenue Multiple – OpenAI = Looks Expensive...
BondCap · valuation
Valuation-to-Revenue Multiple – OpenAI = Looks Expensive...
6.9x — Enterprise Value / Next 12 Months Revenue Multiple
Next page →