slidedeck.co

Battery Ventures report-style deck, The State of AI Report, December 2025

Cover slide of the Battery Ventures 2025 report-style deck
Cover of Battery Ventures' 2025 report-style deck.

The deck opens with a cover titled The State of AI Report, dated December 2025, followed by a disclaimer slide and a page introducing the four Battery partners who wrote it. A market overview section follows, with charts on public-market value creation, cloud provider revenue growth and backlog, the shift from training to inference, the emerging infrastructure layer, and foundation model growth. Next comes a section on the application layer and open alternatives to closed models, then vertical integration and margin structure, followed by venture funding concentration in AI and public valuations of AI beneficiaries against SaaS 1.0 companies. Later sections cover IPO scale, a multi-trillion-dollar market estimate, and operational best practices for AI companies, including entry wedges, KPIs, go-to-market benchmarks, pricing, moats, product architecture, org design, capital structure, and partnerships. The closing section lists themes across runtime infrastructure, the AI software development lifecycle, agentic security, and vertical and workflow solutions, each with a grid of company logos.

Slides worth studying

  • Slide 7: A dual-axis combined bar chart with a clear headline and a single callout for the 4x demand overhang keeps the key number readable.
  • Slide 8: Two stacked pyramids with a connecting arrow and a colour key for training versus inference make the shift easy to grasp at a glance.
  • Slide 21: A side-by-side SaaS 1.0 and AI-native metric table with a short commentary column gives a clean before-and-after comparison.

Details

  • Company: Battery Ventures
  • Year: 2025
  • Type: Report-style deck
  • Industry: Venture capital
  • Pages: 35

The hosted copy of this deck is being prepared. Newsletter readers will find the download here.

Review score: 8 out of 10, Strong

Presentation quality, judged with the SlideDeck Co review heuristic. Ask, evidence and idea merit are not rated here.

Every content page carries a sentence title and a blue subtitle that states the finding, so a reader can skim the 35 pages by headlines alone: slide 7 (demand outpaces capacity, with the backlog versus capex bars and a 4x call-out), slide 14 (the inverted margin table with arrows for where value moves) and slide 21 (SaaS 1.0 and AI-native KPIs side by side with commentary between them). The three sections are announced by divider pages and the charts use one blue family with values on every bar. What is left to improve is the logo density: slide 8 crowds about 60 logos into two triangles that cannot be read, slide 28 fills the chart area with a logo cloud, and slides 31 to 35 are five consecutive market maps that work as reference but stop the argument.

Worth keeping: Keep the title-plus-finding pattern, the labelled charts on 5, 7, 10 and 15, the comparison tables on 14, 21 and 22 and the section dividers.

Most useful repair: Replace the logo triangles on slide 8 with counts and five named examples per tier, and move the five market maps to an appendix with one summary page in the body.

What would change the score: Readable logo pages and a closing page that states the three implications for founders would make this a 9.

Confidence high. 35 of 35 pages inspected on contact sheets, 4 at readable size. Reviewed 2026-11-11. Model-assisted review (claude-fable-5-1, in session).

Full text of the slides (35 slides)

Every slide's words and a description of its charts and images, so the deck can be searched and rebuilt. Also available as Markdown.

Slide 1

DECEMBER 2025 / Battery Ventures / THE STATE OF AI REPORT / This presentation includes proprietary information of Battery Ventures / Battery

Visuals: Full-bleed 3D rendered illustration (about 100% by 100% of the slide) of a receding row of tall transparent glass panels in perspective, running from lower-left to upper-right. A thin white bundle of fibres fans out from a small glossy blue cylindrical probe at the lower-left edge. Each panel holds white and light-blue components: round white discs with blue centres, dense blue dot and dash grids, small white capsule shapes and clusters of small blue pills. The panels stand on small white square bases on a pale grey floor. Soft, diffuse, even studio lighting with no hard shadows. Dominant colours are white, pale grey #E9EDED and bright blue #1E6FD9. The image is not cropped in a box and has a faint white haze over the upper-left.

Slide 2

Disclaimers

This disclaimer applies to this document and the verbal or written comments of any person presenting it. This document, taken together with such verbal or written comments, is referred to herein as the “presentation.” This presentation is being provided for informational purposes only. Nothing herein is or should be construed as investment, legal or tax advice, a recommendation of any kind or an offer to sell or a solicitation of an offer to buy any security or offer investment advisory services with regard to securities. This presentation does not purport to be complete on any topic addressed. The information in this presentation is provided to you as of December 18, 2025 unless otherwise noted and Battery Ventures does not intend to update the information after its distribution, even in the event the presentation becomes materially inaccurate. Certain information in this presentation has been obtained from third party sources and, although believed to be reliable, has not been independently verified and its accuracy or completeness cannot be guaranteed. Certain logos, tradenames, trademarks and copyrights included in the presentation are strictly for identification and informational purposes only. Such logos, trade names, trademarks and copyrights may be owned by companies or persons not affiliated with Battery Ventures and no claim is made that any such company or person has sponsored or endorsed the use of such logos, trade names, trademarks and copyrights in this presentation. This presentation includes various examples of companies in which Battery Ventures has invested. For a complete list of all companies in which Battery Ventures has invested, please visit: https://www.battery.com/list-of-all-companies/. Past performance is not evidence of future results and there can be no assurance that a particular Battery portfolio company will achieve comparable results to any other company.

The information contained herein is based solely on the opinions of Dharmesh Thakker, Danel Dayan, Jason Mendel, and Sudheendra Chilappagari and nothing should be construed as investment advice. This presentation and the anecdotal examples throughout are intended for an audience of entrepreneurs in their attempt to build cloud-focused businesses and not recommendations or endorsements of any particular business or an offering of investment advisory services.

2

Battery

Visuals: Wordmark logo: the word “Battery” in dark navy sans-serif lettering with a small blue triangular corner accent at its top right, at the bottom right corner. Thin vertical blue rule next to the page number at the bottom left. The file also contains about 40 tiny 2 by 2 pixel placeholder images scattered over the white card and the slide (for example at x 8.1 y 45.3 and x 58.4 y 26.6); they render as nothing visible. No photographs, charts, tables or diagrams.

Slide 3

The Battery team

Dharmesh Thakker

[email protected]

Danel Dayan

[email protected]

Jason Mendel

[email protected]

Sudhee Chilappagari

[email protected]

3

Battery

Visuals: Four circular head-and-shoulders portrait photographs, one per column, each about 12.5% of slide width. 1) Dharmesh Thakker: middle-aged man with short dark greying hair, smiling closed-mouth, wearing a purple sweater over a lilac collared shirt, outdoor background with blurred foliage and a brick wall at left, bright daylight. 2) Danel Dayan: young man with short light-brown hair and a light beard, smiling with teeth, grey crew-neck sweater, blurred pink and green autumn foliage behind, soft daylight. 3) Jason Mendel: young man with short dark swept-back hair, smiling broadly, navy blazer over light blue open-collar shirt, blurred pink foliage and a dark window frame at right, soft daylight. 4) Sudhee Chilappagari: young man with short dark hair, smiling, grey and white gingham check collared shirt, blurred blue-grey building facade behind, daylight. All four are framed head and shoulders, centred, cropped into circles, full colour, no overlay. Dashed vertical divider lines in mid grey, about 0.2% wide, running between columns at about x 26.9%, 50.0% and 73.0%, from about y 33% to y 70%. A small blue vertical rule near the page number. The Battery logo: dark navy wordmark 'Battery' in a geometric sans with a blue right-angled triangular corner accent at its top right.

Slide 4

Market Overview

4

This presentation includes proprietary information of Battery Ventures

Battery

Visuals: Photograph-style 3D render (full-bleed, not boxed, not cropped by any frame) of a tall column of translucent white and glass cubes rising from a white plinth on the right half of the slide; the cubes are sharp at the base and blur and fade into semi-transparent haze toward the top, with thin vertical glass-like strands running down from the top edge. A few cubes near the base and mid-column are tinted light blue (#7FB2E5 approx.). On the left a pale grey wall panel carries a vertical grid of thin fibre-like lines with small blue and red dots. Along the floor in the lower half, thin curved lines with scattered blue and red dots fan out toward the left and right edges. Lighting is soft and diffuse from above, giving a clean studio look. Dominant colours are white #FFFFFF, light grey #E9EDED, and pale blue accents. A white translucent haze overlays the top and left edges, fading the image toward white.

Slide 5

AI is driving public-market value creation / AI companies represent 50% of S&P 500 value, and since Nov. 2022 have added ~$18T in market cap, accounting for ~75% of all gains. / Market Cap Added Across S&P500 Companies Since ChatGPT Launch (Nov. 2022) / $20 / $15 / $10 / $5 / $0 / ($5) / Market Cap Added Since Nov 2022 (US$T) / Nov-22 / Feb-23 / May-23 / Aug-23 / Nov-23 / Feb-24 / May-24 / Aug-24 / Nov-24 / Feb-25 / May-25 / Aug-25 / Nov-25 / AI: +185% / +$18T / +$7T / Non-AI: +28% / Ex-AI Basket / AI Basket / Note: AI Basket represents a selected set of public companies across key segments of the AI value chain, including AI Power & Industrials (utilities, cooling, electrical systems), AI Hardware & Infrastructure (GPUs, CPUs, servers, / edge devices), AI Cloud (cloud providers and model platforms), AI Software (enterprise, analytics, security), and AI Components (memory, storage, analog, sensors, manufacturing tools). Ticker list reflects major S&P 500 AI-exposed / companies based on Battery Ventures analysis. Market data as of 12/15/2025. / Sources: CapIQ, Company filings. / 5 / Battery

Visuals: Line chart titled Market Cap Added Across S&P500 Companies Since ChatGPT Launch (Nov. 2022). Y-axis titled Market Cap Added Since Nov 2022 (US$T), ranging from ($5) to $20 with gridline labels $20, $15, $10, $5, $0 and ($5); x-axis runs monthly from Nov-22 to Nov-25 with labels every three months. Two series: AI Basket (dark navy line, ends at about +$18T, labelled AI: +185% and +$18T) and Ex-AI Basket (cyan line, ends at about +$7T, labelled Non-AI: +28% and +$7T). Values are not printed as a data table, so intermediate values are estimates: AI Basket about $0 in Nov-22, about $10 in Nov-23 to Feb-24 range, peak about $19 in Nov-25; Ex-AI Basket about $0 at start, dips to about -$2 in Oct-23, about $5 in late 2024, about $6 in Nov-25. Legend below the plot centred: cyan line Ex-AI Basket, navy line AI Basket. Highlighted by a vertical dashed double-headed navy arrow at the right end of the plot spanning between the two series endpoints, with the shaded grey area between the two lines filled light grey.

Slide 6

Cloud providers are entering a new growth cycle driven by AI demand

Cloud provider growth is reaccelerating at $285B of run-rate revenue as new AI workloads come online

Macro-Driven Optimization

AI-Driven Reacceleration

Google Cloud

Microsoft Azure

aws

Y/Y Growth

$ in Billions

$300

$250

$200

$150

$100

$50

$0

45%

40%

35%

30%

25%

20%

15%

10%

5%

0%

$129 $140 $147 $156 $156 $166 $175 $188 $193 $207 $221 $234 $239 $263 $285

40% 35% 31% 25% 21% 19% 19% 21% 24% 25% 26% 25% 24% 27% 29%

$74 $79 $82 $86 $85 $89 $92 $97 $100 $105 $110 $115 $117 $123 $132

$32 $36 $38 $41 $40 $45 $49 $54 $54 $61 $66 $71 $72 $85 $92

$23 $25 $27 $29 $32 $34 $37 $38 $41 $45 $48 $49 $54 $61

Q1'22 Q2'22 Q3'22 Q4'22 Q1'23 Q2'23 Q3'23 Q4'23 Q1'24 Q2'24 Q3'24 Q4'24 Q1'25 Q2'25 Q3'25

Page number: 6

Note: Growth rates reflect FX-adjusted figures (i.e., not constant currency).

Sources: Company filings, Wall Street research

Battery

Visuals: Chart: stacked column chart titled '$ in Billions' (unit label at top left of the plot), with a combined line series 'Y/Y Growth' on a right-hand axis. Left value axis $0 to $300 in $50 steps; right axis 0% to 45% in 5% steps; x axis lists quarters Q1'22 to Q3'25 (15 quarters); no gridlines. Each column is stacked bottom to top: AWS (orange), Microsoft Azure (cyan), Google Cloud (green), with the total printed above each column. Values per quarter, AWS / Azure / Google Cloud, total: Q1'22 $74 / $32 / $23, total $129; Q2'22 $79 / $36 / $25, total $140; Q3'22 $82 / $38 / $27, total $147; Q4'22 $86 / $41 / $29, total $156 (the 25% growth box overlaps the total label); Q1'23 $85 / $40 / Google Cloud segment unlabelled (about $31 by subtraction), total $156; Q2'23 $89 / $45 / $32, total $166; Q3'23 $92 / $49 / $34, total $175; Q4'23 $97 / $54 / $37, total $188; Q1'24 $100 / $54 / $38, total $193; Q2'24 $105 / $61 / $41 (the $41 is partly overlapped by the 25% box), total $207; Q3'24 $110 / $66 / $45, total $221; Q4'24 $115 / $71 / $48, total $234; Q1'25 $117 / $72 / $49, total $239; Q2'25 $123 / $85 / $54, total $263; Q3'25 $132 / $92 / $61, total $285. Y/Y Growth line (navy #0B2E8C, about 3px stroke, plotted on the right axis) with each point labelled in a dark navy rectangle with white text: Q1'22 40%, Q2'22 35%, Q3'22 31%, Q4'22 25%, Q1'23 21%, Q2'23 19%, Q3'23 19%, Q4'23 21%, Q1'24 24%, Q2'24 25%, Q3'24 26%, Q4'24 25%, Q1'25 24%, Q2'25 27%, Q3'25 29%. Highlighted points are these navy growth labels; the line falls from 40% to a low of 19% at Q2'23 and Q3'23, then rises to 29%. Legend at top right: green square, Google Cloud logo (multicolour G cloud mark with grey 'Google Cloud' wordmark); cyan square, Microsoft Azure logo (blue A mark with cyan 'Microsoft Azure' wordmark); orange square, aws logo (black lowercase 'aws' with orange smile arrow); navy line sample labelled 'Y/Y Growth'. Logos: Google Cloud, Microsoft Azure, aws, and the Battery wordmark at bottom right. No photographs.

Slide 7

Explosive AI demand continues to outpace available infrastructure capacity

Despite investing $329B in CapEx in the last 12 months, cloud providers remain capacity-constrained with $1.2T of backlog, resulting in 4x demand overhang

Cloud Provider Revenue Backlog vs. CapEx

Beg. Revenue Backlog Net New Revenue Backlog Prior Year CapEx Incremental CapEx

$1,400 / $1,200 / $1,000 / $800 / $600 / $400 / $200 / $0

Cloud Provider Revenue Backlog & CapEx Spend ($B)

2021 / 2022 / 2023 / 2024 / LTM

$316 / $89 / $226 / $31 / $85 / $116

$425 / $109 / $316 / $14 / $116 / $130

$517 / $92 / $425 / $3 / $130 / $133

$666 / $148 / $517 / $89 / $133 / $222

$1,205 / $596 / $609 / $137 / $191 / $329

3.0x

Net New

Existing

4x Demand Overhang

7

Note: Revenue Backlog represents Remaining Purchase Obligations (RPO). Cloud providers represents AWS, GCP, Azure, and Oracle.

Sources: Company filings, Wall Street research.

Battery

Visuals: Chart: stacked and grouped vertical bar chart titled 'Cloud Provider Revenue Backlog vs. CapEx'. Y-axis titled 'Cloud Provider Revenue Backlog & CapEx Spend ($B)', range $0 to $1,400 in steps of $200, gridlines absent, labels in italic grey. X-axis categories 2021, 2022, 2023, 2024, LTM. Each year has a pair of stacked columns: left column = Beg. Revenue Backlog (dark blue, bottom) with Net New Revenue Backlog (bright cyan, top); right column = Prior Year CapEx (orange, bottom) with Incremental CapEx (pale peach, top). Values printed: 2021 left column $226 beg (about 226) and $89 net new, total $316; right column $85 prior year CapEx, $31 incremental, total $116; 2022 left $316 beg, $109 net new, total $425; right $116 prior year, $14 incremental, total $130; 2023 left $425 beg, $92 net new, total $517; right $130 prior year, $3 incremental, total $133; 2024 left $517 beg, $148 net new, total $666; right $133 prior year, $89 incremental, total $222; LTM left $609 beg, $596 net new, total $1,205; right $191 prior year, $137 incremental, total $329. Legend at top centre with four square swatches: dark blue, cyan, orange, pale peach. Highlights: a dashed dark-navy curved arrow from the 2021 left total to the 2021 right total labelled '3.0x'; on LTM a dashed curved arrow from the Net New segment (dot labelled 'Net New') down to the 'Existing' dot at the top of the dark blue beg segment, then to the $329 total labelled '4x Demand Overhang' in bold black.

Slide 8

However, cloud provider revenue is shifting from training to inference

AI cloud revenue is still training-heavy, but the next wave of agentic applications will shift the center of gravity to inference

Low

Cloud Provider Revenue Contribution

High

Training

Inference

Training / (Today)

Revenue Driver / (Time Period)

Inference / (Future)

8

Note: 1. Denotes a past or current Battery company. For a full list of all Battery investments, please click here.

Battery

Visuals: Diagram with two large pyramids and two small pyramids. Left large pyramid (grey #D9D9D9, Training) is a trapezoid with a wide base at the bottom, labelled with company logos in six rows: row 1 OpenAI and the xAI mark; row 2 Reflection (with superscript 1) and ANTHROPIC; row 3 Gemini, ElevenLabs and cohere; row 4 Mistral AI, THINKING MACHINES and Suno and Safe Superintelligence Inc.; row 5 periodic labs, LUMA AI, Black Forest Labs and LLaMA by Meta. Above it, a small light-blue (#9AEAFE, Inference) triangle with a dashed orange (#FFB145) outline lists ChatGPT, CURSOR, replit, Claude Code, Lovable, perplexity, Copilot, glean and Cognition in rows. Right large pyramid (light blue #9AEAFE, Inference) has a dashed orange outline and contains about 70 logos in ten rows: ChatGPT, a blue logo, CURSOR, Claude Code, perplexity, Scribe; replit, LEGORA, Harvey, synthesia, OpenEvidence; eve, RILLA, SIERRA, Cognition, CodeRabbit, a magenta icon; exa, Rillet, ABRIDGE, DECAGON, clay, glean; Copilot, Parloa, a black icon, WRITER, Hebbia, Reevo, an orange icon; Abnormal, HeyGen, augment code, Lovable, Listen, rogo, aaru; Resolve.ai, Serval, ZAI, Traversal, CONSOLE, FACTORY, DayAI; WisdomAI, unify (1), attio, ROX, GONG (1), Attention, Actively, Nooks, Dropzone AI; prophet, outset.ai, Strella, dialogue, Deductive, Giga, tavily, Firecrawl; atomicwork (1), AVOCA, Basis, maxima, campfire, Jump (1), TENZAI (1), xbow, Lindy (1), an icon (1), levelpath (1). Small grey (#D9D9D9, Training) triangle at top-right lists Suno, LLaMA by Meta, Reflection (1), Mistral AI, THINKING MACHINES, Gemini, ElevenLabs, cohere, OpenAI, ANTHROPIC and the xAI mark. A thin orange (#FFB145) double-headed diagonal arrow runs from the top-left light-blue triangle down to the right large pyramid. A vertical dark-blue (#01308F) double-headed arrow on the left, with Low at top and High at bottom, and a horizontal dark-blue double-headed arrow along the bottom with its three labels. Legend top-right: grey square labelled Training and light-blue square labelled Inference.

Slide 9

A new infrastructure layer is taking shape

The next era of infrastructure expands beyond scaling storage and compute to scaling intelligence

Cloud Era Incumbents

AI-Native Challengers

Dev Tools and Collaboration

GitHub GitLab ATLASSIAN

CURSOR CLAUDE CODE Linear Cognition Codex CLI Lovable

Data and Analytics

snowflake elastic mongoDB

databricks1 supabase

ClickHouse1

Cloud Infrastructure

aws Google Cloud

Microsoft Azure

baseten together.ai fal

Fireworks AI CoreWeave

Networking and Silicon

AMD NVIDIA

ARISTA JUNIPER NETWORKS

nexthop.ai1 groq

SambaNova SYSTEMS Rivos

9

Note: 1. Denotes a past or current Battery company. For a full list of all Battery investments, please click here.

Battery

Visuals: Logo grid, no chart. Left column (Cloud Era Incumbents): Dev Tools row shows GitHub (black mark with wordmark), GitLab (orange fox mark with dark grey wordmark), ATLASSIAN (blue triangle mark with blue wordmark). Data and Analytics row shows snowflake (light blue asterisk mark with wordmark), elastic (multicolour circles mark with dark grey wordmark), mongoDB (green leaf with dark grey wordmark). Cloud Infrastructure row shows aws (black wordmark with orange arrow), Google Cloud (multicolour G mark with grey wordmark), Microsoft Azure (blue A mark with blue wordmark). Networking and Silicon row shows AMD (black wordmark with arrow mark), NVIDIA (green eye mark with black wordmark), ARISTA (dark navy wordmark), JUNIPER NETWORKS (dark grey wordmark with small caps subline). Right column (AI-Native Challengers): Dev Tools row shows CURSOR (black cube mark with black wordmark), CLAUDE CODE (red-orange pixelated wordmark), Linear (dark grey striped circle mark with wordmark), Cognition (black hexagon cluster mark with wordmark), Codex CLI (grey wordmark), Lovable (multicolour heart mark with bold black wordmark). Data row shows databricks (red-orange diamond mark with wordmark) with superscript 1, supabase (green bolt mark with bold dark wordmark), ClickHouse (black bar mark with wordmark) with superscript 1. Cloud Infrastructure row shows baseten (black mark with wordmark), together.ai (dark grey wordmark with blue dot), fal (black mark with bold wordmark), Fireworks AI (purple zigzag mark with dark wordmark), CoreWeave (blue CW mark with dark wordmark). Networking row shows nexthop.ai (teal wordmark) with superscript 1, groq (black wordmark with trademark mark), SambaNova SYSTEMS (orange striped mark with dark wordmark), Rivos (teal rounded square mark with dark wordmark). Footer: a thin blue vertical rule and the Battery wordmark in dark navy with a blue corner triangle at top right.

Slide 10

Foundation models are powering the AI wave…

Foundation models are scaling at unprecedented speeds, reaching $10B of ARR 5x faster than platform incumbents

Foundation Model Growth

OpenAI

Anthropic

$120

$100

$80

$60

$40

$20

$0

$6

$1

$20

$9

$30

$26

$60

$40

$100

$70

OpenAI: 18.2x Growth

Anthropic: 70.0x Growth

2024A

2025E

2026E

2027E

2028E

Annual Recurring Revenue ($B)

Time to $10B ARR

OpenAI

Anthropic

Google

Azure

AWS

Meta

Salesforce

ServiceNow

Nvidia

2.5

3.5

8

9

10

10

19

20

25

Since ChatGPT launch

0

5

10

15

20

25

30

Years

10

Note: ARR represents annualized run-rate where ARR is not disclosed.

Sources: The Information, Epoch AI, Company filings.

Battery

Visuals: Two charts. (1) Left chart: clustered vertical bar chart titled 'Foundation Model Growth' in a light-blue header bar. Y-axis titled 'Annual Recurring Revenue ($B)', linear scale $0 to $120 in $20 steps, grey tick labels. X-axis categories 2024A, 2025E, 2026E, 2027E, 2028E. Series OpenAI (navy #01308F) with labelled values $6, $20, $30, $60, $100; series Anthropic (orange #FFB145) with labelled values $1, $9, $26, $40, $70. Legend top-right: navy square 'OpenAI', orange square 'Anthropic'. Two overlaid diagonal arrows rise from lower-left to upper-right: navy arrow labelled 'OpenAI: 18.2x Growth' and orange arrow labelled 'Anthropic: 70.0x Growth', both rotated to the slope of the arrow. No highlighted bars. (2) Right chart: horizontal bar chart titled 'Time to $10B ARR' in a light-blue header bar. Category axis (top to bottom): OpenAI 2.5 (navy), Anthropic 3.5 (orange), Google 8, Azure 9, AWS 10, Meta 10, Salesforce 19, ServiceNow 20, Nvidia 25 (all cyan #75F3FF except the first two). Value axis titled 'Years', linear 0 to 30 in steps of 5, grey italic tick labels. Values printed at bar ends in dark text. OpenAI and Anthropic rows are highlighted by a dotted orange (#FFB145) rounded rectangle enclosing both rows, with a thin dark arrow from the text 'Since ChatGPT launch' pointing left to the end of the OpenAI bar. Logo: Battery wordmark, dark teal-black text with a blue (#1C6FC4) triangular corner mark at top-right, in the bottom-right corner.

Slide 11

…And are enabling a new class of applications

Just as networking unleashed the internet and the cloud abstracted infrastructure, the agentic future is being built on top of foundation models

Infrastructure Era

Key Enablers

Applications Unleashed

Networking

(1990s-2000s)

PC & Internet

Cisco

Juniper Networks

intel

Google

amazon

Microsoft

Cloud (2010s)

aws

Microsoft Azure

Google Cloud

SaaS

salesforce

workday.

servicenow

Foundation Models

(2020s+)

AI Apps

OpenAI

ANTHROP\C

Gemini

SIERRA

GONG

glean

CURSOR

LEGORA

Note: 1. Denotes a past or current Battery company. For a full list of all Battery investments, please click here.

11

Battery

Visuals: Logo set, no photographs or charts. Row 1 (PC & Internet): left of divider Cisco (blue wordmark), Juniper Networks (black wordmark with small NETWORKS subline), Intel (blue lowercase wordmark with registered mark); right of divider Google (multicolour wordmark), amazon (black wordmark with orange smile arrow), Microsoft (four-colour squares and grey wordmark). Row 2 (SaaS): left of divider aws (black wordmark with orange smile arrow), Microsoft Azure (blue logo mark and wordmark), Google Cloud (multicolour cloud mark and grey wordmark); right of divider salesforce (white text on blue cloud), workday. (blue-grey wordmark with orange arc), servicenow (dark wordmark with green o). Row 3 (AI Apps): left of divider OpenAI (black knot mark and wordmark), ANTHROP\C (black bold wordmark), Gemini (blue-to-purple gradient wordmark); right of divider SIERRA (green atom-like mark and green wordmark), GONG with superscript 1 (purple star mark and purple wordmark), glean (blue lowercase wordmark), CURSOR (grey-black faceted cube mark and black wordmark), LEGORA (black serif wordmark). Icons: header icon top-right of Infrastructure Era bar is a line-drawn gear over city buildings in dark grey outline; Key Enablers icon is a line-drawn document with lines in dark grey; Applications Unleashed icon is a line-drawn phone with app grid and gear in dark grey. Row icons: Networking is a line-drawn globe inside a ring with eight nodes in dark grey; Cloud is a line-drawn cloud over a stack of database cylinders in dark grey; Foundation Models is a line-drawn open hand holding a small cube in dark grey. Dashed vertical dividers are light grey, 1 pt, one per row panel.

Slide 12

The walled garden of foundation models is being challenged by open alternatives

Open models, runtimes and inference engines are fueling the new generation of intelligent applications

Foundation Model Monthly Token Growth

Reasoning Tokens Explode

Tokens Processed (T)

1,400

1,200

1,000

800

600

400

200

0

10

480

1,300

May-24 Jun-24 Jul-24 Aug-24 Sep-24 Oct-24 Nov-24 Dec-24 Jan-25 Feb-25 Mar-25 Apr-25 May-25 Jun-25 Jul-25 Aug-25 Sep-25 Oct-25

Token Usage by Model Type

Closed Models

Open Models

2%

2024A

98%

30%

2025E

70%

Emerging Open Ecosystem

Open Models

Reflection 1

Qwen

deepseek

Gemma

Runtimes

TensorRT

SGL

vLLM

Modular

RAY

Spark

Inference

baseten

Fireworks AI

together.ai

Modal

fal

Note: 1. Denotes a past or current Battery company. For a full list of all Battery investments, please click here.

Sources: Company filings, OpenRouter State of AI.

12

Battery

Visuals: Chart 1: vertical bar chart titled Foundation Model Monthly Token Growth. Y-axis titled Tokens Processed (T), range 0 to 1,400 with gridline labels every 200 (0, 200, 400, 600, 800, 1,000, 1,200, 1,400). X-axis categories monthly May-24 through Oct-25 (18 months), labels rotated about 70 degrees in italic. One series, dark blue #0071CE bars. Labelled values: May-24 10, May-25 480, Oct-25 1,300. Unlabelled values are estimates: Jun-24 about 15, Jul-24 about 20, Aug-24 about 30, Sep-24 about 40, Oct-24 about 55, Nov-24 about 70, Dec-24 about 100, Jan-25 about 130, Feb-25 about 180, Mar-25 about 260, Apr-25 about 350, Jun-25 about 690, Jul-25 about 980, Aug-25 about 1,080, Sep-25 about 1,190. Highlighted region: the bars from May-25 to Oct-25 sit on a light cyan #C8FAFF band with a dark blue #0071CE label tab reading Reasoning Tokens Explode. A Gemini logo (blue-to-purple-pink gradient wordmark with a purple four-point sparkle above the i) floats in the empty upper-left of the plot area. Chart 2: donut chart titled 2024A at centre, legend top right: Closed Models (dark navy square) and Open Models (cyan square). Closed 98% (navy #01308F), Open 2% (cyan #03CBFD), with 2% label above the thin cyan sliver. Chart 3: donut chart titled 2025E at centre, Closed 70% (navy #01308F) and Open 30% (cyan #03CBFD); labels 30% and 70% in white inside the ring. Both donuts have a white inner hole and a thick ring. Arrow: a solid blue #0071CE circle with a white right-pointing chevron, sitting on the vertical divider. Logos in the ecosystem panel: Open Models row shows Reflection (dark green mark with superscript 1), Qwen (purple wordmark with icon), deepseek (blue whale icon with purple wordmark), Gemma (blue wordmark with flower icon). Runtimes row shows TensorRT (green circle logo), SGL (orange wordmark), vLLM (blue check and grey wordmark), Modular (dark grey wordmark), RAY (blue network icon and dark grey wordmark), Apache Spark (orange star and wordmark). Inference row shows baseten (black icon and wordmark), Fireworks AI (purple icon and grey wordmark), together.ai (dark grey wordmark), Modal (green outline icon and black wordmark), fal (black icon and wordmark). Battery logo: dark navy wordmark with blue corner triangle.

Slide 13

Early leaders are racing to vertically integrate…

Players across the stack are moving up and down the value chain to build stronger moats and defensibility

Cloud Providers

Foundation Models

AI Applications

Applications

Models

Infrastructure

Silicon

AI Search / Gemini App / Antigravity

CoPilot

Kiro / Amazon Q

ChatGPT

Claude Code

Cursor IDE / Cursor Agent / Bugbot

Gemini / Gemma

OpenAI

Nova

GPT

Claude

Composer-1

Claude/Gemini/OpenAI

GCP

Azure

AWS

Azure

AWS / GCP

Baseten / Fireworks / Together

TPUs

Maia 100

Trainium / Inferentia

NVIDIA / AMD

Trainium / Inferentia / TPUs

NVIDIA

Own the Infrastructure

Own the Models

Own the Apps

Strong/mature presence

Emerging/partial presence

Weak or absent

13

Sources: Company filings.

Battery

Visuals: A matrix diagram of 4 rows by 9 columns of rectangular cells plus 3 header bands, 6 logos and one row-label column. Row labels from top: Applications, Models, Infrastructure, Silicon. Column headers: Cloud Providers (Google, Microsoft, AWS), Foundation Models (OpenAI, Anthropic), AI Applications (Cursor). Cell colours encode presence: green #32DB77 for Strong/mature presence, orange #FFB145 for Emerging/partial presence, salmon red #FF9B9B for Weak or absent, and light grey about #D3D8D8 for the Claude/Gemini/OpenAI cell. Cells, top to bottom. Google column: AI Search / Gemini App / Antigravity (green), Gemini / Gemma (green), GCP (green), TPUs (green). Microsoft column: CoPilot (orange), OpenAI (orange), Azure (green), Maia 100 (red). AWS column: Kiro / Amazon Q (orange), Nova (red), AWS (green), Trainium / Inferentia (green). OpenAI column: ChatGPT (green), GPT (green), Azure (red), NVIDIA / AMD (red). Anthropic column: Claude Code (green), Claude (green), AWS / GCP (red), Trainium / Inferentia / TPUs (red). Cursor column: Cursor IDE / Cursor Agent / Bugbot (green), Composer-1 (green), Claude/Gemini/OpenAI (light grey), Baseten / Fireworks / Together (orange), NVIDIA (red). Logos: Google wordmark in multicolour (blue, red, yellow, blue, green, red); Microsoft four-square logo in red, green, blue and yellow with grey wordmark; aws wordmark in dark navy with orange smile arrow; OpenAI knot logo with black wordmark; ANTHROPIC black uppercase wordmark; CURSOR black cube icon with black uppercase wordmark; Battery wordmark in dark navy with a blue corner triangle at top right. Legend: green, orange and red square swatches followed by their labels. Three pills outlined in blue with the labels Own the Infrastructure, Own the Models and Own the Apps. About 40 tiny 2 by 2 pixel placed fragments exist with no visible content.

Slide 14

…Which may improve the currently inverted margin structure

While silicon and cloud infrastructure command a majority of the margin today, value is beginning to expand up the stack

SaaS

Gross Margin

AI

Gross Margin

Future

Value Capture

Path to AI Margin Expansion Up the Stack

Application

80%+

0% - 30%

• AI margins reflect distribution and market-share capture

• Long-term pricing shifts toward value-based/outcome pricing

• Cost of intelligence falls via optimization, boosting gross margins

Model

Inference

N/A

30% - 60%

• Token cost deflation via smaller, specialized models and open models

• Smarter routing to cheapest effective models

• Improved efficiency techniques such as caching and speculative decoding

• Workflow stickiness creates pricing power beyond raw compute

Cloud

Infrastructure &

Platform

60%

60%

• Scale benefits are offset by intense competition

• High CapEx requirements limit margin expansion

Chips

45%

75%

• Supply catches up to demand, reducing pricing power

• Cloud Providers vertically integrate, pressuring incumbent margins

14

Sources: Battery Ventures estimates.

Battery

Visuals: A table with a header row and four body rows. Columns: (1) row labels, (2) SaaS Gross Margin, (3) AI Gross Margin, (4) Future Value Capture (arrow icons), (5) Path to AI Margin Expansion Up the Stack (bullet lists). Rows: Application: SaaS 80%+ (white oval outline around the value), AI 0% - 30%, Future Value Capture a solid green upward arrow, bullets: AI margins reflect distribution and market-share capture / Long-term pricing shifts toward value-based/outcome pricing / Cost of intelligence falls via optimization, boosting gross margins. Model Inference: SaaS N/A, AI 30% - 60%, Future Value Capture a solid green upward arrow, bullets: Token cost deflation via smaller, specialized models and open models / Smarter routing to cheapest effective models / Improved efficiency techniques such as caching and speculative decoding / Workflow stickiness creates pricing power beyond raw compute. Cloud Infrastructure & Platform: SaaS 60%, AI 60%, Future Value Capture a solid orange left-right double-headed arrow, bullets: Scale benefits are offset by intense competition / High CapEx requirements limit margin expansion. Chips: SaaS 45%, AI 75% (white oval outline around the value), Future Value Capture a solid dark red downward arrow, bullets: Supply catches up to demand, reducing pricing power / Cloud Providers vertically integrate, pressuring incumbent margins. Arrow icons are flat block arrows about 3.5% wide and 6% tall, centred in the Future Value Capture column. The white ovals are thin white outlines (about 0.3% of slide height) around 80%+ and 75%. Small Battery logo at bottom right.

Slide 15

VC funding is concentrated in AI deals

A small number of large-scale AI companies now absorb the majority of venture dollars despite accounting for only one-third of the deal volume

State of VC Funding: AI vs. Non-AI Companies

AI as a % of Venture Deals

17%

22%

25%

29%

29%

38%

50%

64%

AI represents 64% / of VC funding, despite / only accounting for / 32% of deals

19%

20%

22%

23%

24%

28%

33%

32%

AI Deal Funding %

AI Deal Count %

VC Funding Value ($B)

$360

$236

$216

$250

$166

$149

$156

$175

$103

$26

$35

$43

$257

$68

$64

$108

$161

$124

$121

$132

$168

$103

$107

$89

2018

2019

2020

2021

2022

2023

2024

2025 (YTD)

AI VC Deals

Ex-AI VC Deals

4 Companies = / 40% of funding

ANTHROP\C

OpenAI

1

databricks

Note: 1. Denotes a past or current Battery company. For a full list of all Battery investments, please click here.

Sources: Pitchbook.

15

Battery

Visuals: Two charts and a set of logos. Chart 1 is a two-series line chart titled 'AI as a % of Venture Deals' with no axis scale printed and no year labels; the funding series (light cyan #03CBFD, values 17%, 22%, 25%, 29%, 29%, 38%, 50%, 64%, labels printed above each point) and the deal count series (navy #01308F, values 19%, 20%, 22%, 23%, 24%, 28%, 33%, 32%, labels printed below each point) run across eight unlabelled points from 2018 to 2025. Legend below the chart: 'AI Deal Funding %' with a light cyan line and dot, 'AI Deal Count %' with a navy line and dot. The 2025 point (64% and 32%) is the highlighted end point, enclosed by a dotted orange rounded callout. Chart 2 is a stacked column chart titled 'VC Funding Value ($B)' with no y-axis scale printed; each column shows a total above it in bold, a navy lower segment labelled 'Ex-AI VC Deals' and a cyan upper segment labelled 'AI VC Deals'. Values by year (total / ex-AI / AI): 2018 $149 / $124 / $26; 2019 $156 / $121 / $35; 2020 $175 / $132 / $43; 2021 $360 / $257 / $103; 2022 $236 / $168 / $68; 2023 $166 / $103 / $64; 2024 $216 / $107 / $108; 2025 (YTD) $250 / $89 / $161. The 2025 column is the highlighted one, linked by an orange dotted connector with arrowhead to the side box. Grey baseline under the columns. Logos in the side box, top to bottom: ANTHROP\C wordmark, OpenAI wordmark with knot symbol, xAI black X mark, Databricks red-orange icon with wordmark 'databricks' and a superscript 1.

Slide 16

The public market is rewarding early AI beneficiaries / Incumbents are building AI products, buying AI startups and hiring AI talent to capture value in the public market / Median Market Cap ($B): AI Beneficiaries vs SaaS 1.0 / AI Beneficiaries / SaaS 1.0 / $80 / $60 / $40 / $20 / $0 / $74 / $52 / $40 / $16 / $6 / $5 / $5 / $4 / 2022 / 2023 / 2024 / 2025 / EV/NTM Revenue: AI Beneficiaries vs SaaS 1.0 / 20.0x / 15.0x / 10.0x / 5.0x / 0.0x / 18.6x / Regardless of growth, public SaaS 1.0 companies / without an AI story are capped at ~6.0x EV/NTM revenue / 6.2x / 6.0x / 4.6x / 3.7x / AI Beneficiaries / SaaS 1.0 (>25% Growth) / SaaS 1.0 (15%-25% Growth) / SaaS 1.0 (Total Bucket) / SaaS 1.0 (<15% Growth) / 16 / Note: "AI Beneficiaries" classification is based on Battery Ventures internal analysis and includes: ADBE, APP, CRM, CRWD, CYBR, DDOG, DOCS, FIG, FROG, MDB, NET, NOW, PLTR, RBRK, SHOP, SNOW, and ZS. Market data as of 12/15/25. / Sources: CapIQ, Company filings. / Battery

Visuals: Two charts. Left: a dot-and-dumbbell (dumbbell) chart titled 'Median Market Cap ($B): AI Beneficiaries vs SaaS 1.0'. Y axis ticks $0, $20, $40, $60, $80 (range $0 to $80, linear, italic grey labels); X axis years 2022, 2023, 2024, 2025 with horizontal baseline. Series 'AI Beneficiaries' in orange-yellow dots (#FFB145) with labelled values: 2022 $16, 2023 $40, 2024 $52, 2025 $74. Series 'SaaS 1.0' in blue dots (#0071CE) with labelled values: 2022 $4, 2023 $6, 2024 $5, 2025 $5. Each year is joined by a vertical dashed cyan (#29C5F6-like) connector line between the two dots. Legend at top centre: orange dot 'AI Beneficiaries', blue dot 'SaaS 1.0'. No highlighted points. Right: a vertical bar chart titled 'EV/NTM Revenue: AI Beneficiaries vs SaaS 1.0'. Y axis ticks 0.0x, 5.0x, 10.0x, 15.0x, 20.0x (range 0.0x to 20.0x). Five bars with labelled values: AI Beneficiaries 18.6x (orange #FFB145); SaaS 1.0 (>25% Growth) 6.2x; SaaS 1.0 (15%-25% Growth) 6.0x; SaaS 1.0 (Total Bucket) 4.6x; SaaS 1.0 (<15% Growth) 3.7x (all four blue #0071CE). Bars are solid, flat, no gridlines. A callout pill sits above the four blue bars, overlapping the space between the 10.0x and 5.0x gridline heights, light blue fill with dashed blue outline stating 'Regardless of growth, public SaaS 1.0 companies without an AI story are capped at ~6.0x EV/NTM revenue'. Bottom-right: Battery wordmark logo in dark navy with a blue triangular corner mark.

Slide 17

IPO candidates have reached unprecedented scale and growth

AI-native companies are approaching IPO scale, not only replacing legacy software but also capturing spend that once went to services and human labor

Median NTM Revenue at IPO ($M)

$1,000

$900

$800

$700

$600

$500

$400

$300

$200

$100

$0

$381

$436

$766

$878

2018-2019

2020-2021

2022-2023

2024-2025

Emerging AI Companies Nearing IPO Scale

$1B+ ARR

$200M+ ARR

$100M+ ARR

1

1

1

1

1

1

1

1

1

1

Note: Data revenue at IPO date exclusively represents software and infrastructure companies. 1. Denotes a past or current Battery company. For a full list of all Battery investments, please click here.

Sources: CapIQ, ARR.Club.

17

Battery

Visuals: Bar chart titled 'Median NTM Revenue at IPO ($M)' with a vertical value axis from $0 to $1,000 (tick labels $0, $100, $200 to $1,000 in $100 steps, no gridlines, no axis title) and a category axis labelled 2018-2019, 2020-2021, 2022-2023, 2024-2025. Four bars, each with a blue (#1E6FD9) outline and white fill: 2018-2019 printed value $381 (bar top about $381); 2020-2021 printed value $436; 2022-2023 printed value $766 (only logo is klaviyo); 2024-2025 printed value $878 (tallest bar). Bars contain company logos stacked vertically. No legend, no highlighted points beyond the blue outlines. Bar 2018-2019 logos from top: Dropbox, elastic, Zoom, DATADOG (superscript 1), zscaler, Cloudflare, CrowdStrike. Bar 2020-2021 logos from top: snowflake, monday.com, HashiCorp, sprinklr (superscript 1), GitLab, PROCORE, CONFLUENT (superscript 1), JFrog (superscript 1). Bar 2022-2023: klaviyo. Bar 2024-2025 logos from top: rubrik, Figma, CoreWeave, netskope, NAVAN, SailPoint, ServiceTitan (superscript 1). Right side: three white cards, each a vertical list of logos. Card '$1B+ ARR' from top: OpenAI, ANTHROPIC, databricks (superscript 1), xAI mark, CURSOR, Canva, surge (superscript AI). Card '$200M+ ARR' from top: perplexity, synthesia, Lovable, ElevenLabs, glean, SUNO, GONG (superscript 1), Fireworks AI, dataiku (superscript 1). Card '$100M+ ARR' from top: clay, Gamma, SIERRA, Harvey, Cognition, HeyGen, EliseAI (superscript AI). Logos are flat, single-colour or full-colour brand marks with no photographs.

Slide 18

AI is on the cusp of unlocking a multi-trillion-dollar market

AI outcomes will be larger than those of any prior platform shift

($ in billions)

Application Software

Infrastructure Software

Software TAM

Additional Disruption Opportunity with AI

AI Software TAM

$4,100

$1,460

$921

$431

$490

$1,719

Cloud Software

+ Services

Automation

+ Incremental

Human Labor

Displacement

AI Software

18

Sources: Gartner, Battery Ventures estimates.

Battery

Visuals: Waterfall bar chart titled '($ in billions)', no axis lines or tick values printed except a horizontal baseline. Bars left to right: Software TAM stacked column totalling $921 (bottom segment Infrastructure Software $490 in #0071CE, top segment Application Software $431 in #32DB77); a floating cyan bar $1,719 labelled '+ Services Automation' spanning roughly from the top of the $921 column to about $2,640; a floating orange bar $1,460 labelled '+ Incremental Human Labor Displacement' spanning from about $2,640 to about $4,100; a dark navy total bar $4,100 labelled 'AI Software'. Bar colours: cyan #03CBFD for Services Automation, orange #FFB145 for Incremental Human Labor Displacement, navy #01308F for AI Software total. Legend at top right: green square #32DB77 Application Software, blue square #0071CE Infrastructure Software. Value labels are printed inside segments in dark text except white on blue $490. Bar heights are proportional to values; the navy total bar top is at the same level as the orange bar top. Dashed grey vertical dividers separate the three header groups. No photographs, icons or illustrations.

Slide 19

Operational Best / Practices / 19 / This presentation includes proprietary information of Battery Ventures / Battery

Visuals: Photograph, full-bleed, covering the whole slide: a tall freestanding column of small white translucent cubes on a white plinth, centred right at about x 49% to 80%. The cubes are dense and solid at the base (around y 70% to 81%), then break apart and dissolve upward into a fine haze of blurred cubes and thin vertical transparent threads that run to the top edge. A few cubes near the middle and base are tinted pale blue (#3D8EDC). Behind the column are pale grey walls and a recessed panel. Thin translucent curved threads with small blue and red dots run along the left wall and the white floor in the lower half, fading into the floor at the bottom. Lighting is soft, diffuse and even with no hard shadows. Dominant colours are white, pale grey and light blue, with small red dots. The image is not cropped, is not boxed, and carries a white gradient fade at the left edge (about 0 to 24% width) and a white fade across the bottom strip (about y 72% to 100%) that lightens the floor threads.

Slide 20

Pick the right entry wedge and expand into adjacent products and verticals

A targeted land motion with quick time-to-value unlocks distribution while product depth and breadth create durable moats

CURSOR

ANTHROP\C

Application

Proprietary Models

01

Wedge

(Land + Distribution)

Code / Completion

CI/CD, testing, / code review, etc.

Claude Sonnet / Models

More / Models

Proprietary Models

Application

02

Moat

(Durability)

Open source/proprietary models to diversify / away from AI Lab models

Claude Code

Vertical Expansion (Product Depth)

Horizontal Expansion (Product Breadth)

20

Battery

Visuals: Diagram (2x2 matrix-style flow). Shapes: two column logos at top (Cursor logo, Anthropic wordmark). Four header bars: 'Application' (light blue #C2E4FF, Cursor column, top row), 'Proprietary Models' (light blue, Anthropic column, top row), 'Proprietary Models' (pale cyan #CDF5FF, Cursor column, bottom row), 'Application' (pale cyan, Anthropic column, bottom row). Row 01 band (light blue #C2E4FF) contains two white rounded cards: Cursor card 'Code Completion' with a dashed navy arrow pointing right to 'CI/CD, testing, code review, etc.'; Anthropic card 'Claude Sonnet Models' with a dashed navy arrow pointing right to 'More Models'. Row 02 band (pale cyan #CDF5FF) contains two white rounded cards: Cursor card 'Open source/proprietary models to diversify away from AI Lab models'; Anthropic card 'Claude Code'. Four solid navy #01308F chevron-down arrows, each connecting a row 01 card bottom edge to a row 02 header: two under the Cursor card (left chevron under Code Completion, right chevron under CI/CD), two under the Anthropic card (left under Claude Sonnet, right under More Models). Navy #01308F numbered circles '01' and '02' sit at the left edge of each row. A thick navy #01308F vertical arrow on the left (pointing down, label rotated 90 degrees reading upward: 'Vertical Expansion (Product Depth)') and a thick navy #01308F horizontal arrow at the bottom (pointing right, label 'Horizontal Expansion (Product Breadth)' in white bold). Logos: Cursor (black cube icon with 'CURSOR' wordmark in dark charcoal, about #231F20) and Anthropic ('ANTHROP\C' wordmark in near-black). Battery logo bottom right (dark navy wordmark with a blue corner triangle).

Slide 21

Align your KPIs to the modern AI operating metrics

As AI software shifts from tools that assist work to agents that autonomously complete it, company metrics must evolve as well, capturing not just revenue growth and efficiency but also product usage, customer value, and unit economics

SaaS 1.0

Metric

Target

ARR Growth

2x – 3x

NDR

130%+

Magic #

0.8x+

Burn Ratio

< 3x

Commentary

AI businesses are growing faster than SaaS 1.0 as immediate productivity gains and faster adoption cycles compress what used to take years into months.

Early-stage AI companies often trade growth and distribution for margins, which makes validating long-term scalable unit economics essential.

Gross retention is critical in AI because churn signals experimentation, while strong retention shows real adoption and product stickiness.

Value is no longer tied to seats or licenses. Real usage is the clearest measure of whether customers are consistently realizing value.

Magic number and burn ratio still matter because they validate whether the business can scale efficiently.

AI-Native

Metric

Target

ARR Growth

5x – 10x

Gross Margin

20%-40%

Gross Retention

80%+

Usage

DAU/WAU/MAU

Magic #

1.0x+

Burn Ratio

< 2.0x

21

Sources: Battery Ventures estimates.

Battery

Visuals: Two tables and a set of connector lines. Left table SaaS 1.0: columns Metric and Target; rows ARR Growth 2x – 3x; NDR 130%+; Magic # 0.8x+; Burn Ratio < 3x. Centre column Commentary: five grey text boxes, top to bottom: 'AI businesses are growing faster than SaaS 1.0 as immediate productivity gains and faster adoption cycles compress what used to take years into months.'; 'Early-stage AI companies often trade growth and distribution for margins, which makes validating long-term scalable unit economics essential.'; 'Gross retention is critical in AI because churn signals experimentation, while strong retention shows real adoption and product stickiness.'; 'Value is no longer tied to seats or licenses. Real usage is the clearest measure of whether customers are consistently realizing value.'; 'Magic number and burn ratio still matter because they validate whether the business can scale efficiently.' Right table AI-Native: columns Metric and Target; rows ARR Growth 5x – 10x; Gross Margin 20%-40%; Gross Retention 80%+; Usage DAU/WAU/MAU; Magic # 1.0x+; Burn Ratio < 2.0x. Connectors: from the right edge of commentary boxes 1 to 3 and 4 to 5, elbow lines run to navy dots at the left edge of the matching AI-Native rows (ARR Growth, Gross Margin, Gross Retention, Usage, and Burn Ratio); the second box connects to Gross Margin, the fifth to Burn Ratio. Thin dashed grey row dividers in both tables. Battery wordmark logo at bottom right in dark navy with a blue triangular corner mark.

Slide 22

Benchmark your GTM to a more efficient post-AI framework

AI is reshaping GTM by driving higher productivity, faster ramps and fundamentally better unit economics

Pre-AI

Post-AI

Rep Quota

$1.0M - $1.2M

$1.2M - $1.5M

Quota Measurement Period

Annually/Quarterly

Annually/Quarterly/Monthly

Ramp Time

9-12 months

6-9 months

Conversion Rate (Oppty -> Close)

20% - 30%

40% - 50%

Attainment Targets

70% attainment/25% time selling

80% attainment/50% time selling

Team Structure

AE:SE | 2:1

AE:SDR | 3:1

AE:SE | 3:1

AE:SDR | 5:1

CAC Payback

12-18 months

<12 months

LTV:CAC

3:1

4:1

S&M % of Revenue

50% - 60%

30% - 40%

22

Sources: Battery Ventures estimates.

Battery

Visuals: Table with 3 columns and 10 rows (one header row plus 9 data rows). Column 1 is the row label column with bold labels. Column 2 is headed Pre-AI and column 3 is headed Post-AI. Rows, as printed: Rep Quota: $1.0M - $1.2M | $1.2M - $1.5M. Quota Measurement Period: Annually/Quarterly | Annually/Quarterly/Monthly. Ramp Time: 9-12 months | 6-9 months. Conversion Rate (Oppty -> Close): 20% - 30% | 40% - 50%. Attainment Targets: 70% attainment/25% time selling | 80% attainment/50% time selling. Team Structure: AE:SE | 2:1 over AE:SDR | 3:1 | AE:SE | 3:1 over AE:SDR | 5:1. CAC Payback: 12-18 months | <12 months. LTV:CAC: 3:1 | 4:1. S&M % of Revenue: 50% - 60% | 30% - 40%. The Pre-AI header cell is light blue and the Post-AI header cell is saturated cyan, with the Post-AI column body tinted pale cyan. Thin dashed grey horizontal rules separate the rows, and a solid blue rule closes the bottom of the table. No photographs or charts. A small Battery wordmark logo sits in the bottom-right corner.

Slide 23

Unlock a new era of demand generation with AI powered strategies / AI is unlocking a new growth blueprint for modern demand generation / AI in Demand Generation / New data and intent signals to surface / higher quality leads / Agents for deeper account research / and automated plays and sequences / Personalized engagement: right person, / right message, right time / Always-on agents to qualify inbound / AI-Powered ICP Targeting and Qualification / 1. Multi-dimensional segmentation: Blend first-party and third-party signals, including firmographics, growth rate, web traffic, product usage and technographics to generate higher-quality, higher-intent leads. / 2. ICP learned from real wins: Models train on closed-won deals to score account fit and intent, instead of relying on static personas. / 3. Every win triggers a lookalike loop: When a deal closes, agents find and enrich lookalike accounts, identify key buyers, and hand reps a prioritized list with research and messaging. / 4. Agents cover every step of the demand-gen funnel: build and audit pipeline, score, engage and route leads, analyze closed-lost reasons and update CRM fields automatically, allowing reps to focus on closing deals. / 23 / Battery

Visuals: Left: a clipboard-style checklist graphic with a blue (#0071CE) header tab reading AI in Demand Generation and four rows separated by dashed dark-blue dividers. Each row has one text statement on the left and a square blue-outlined box on the right containing a green check mark (Wingdings tick, #008000). Right: a numbered list of four items with underlined lead-in phrases, no icons. Bottom-right: the Battery wordmark logo. No photographs, charts or tables.

Slide 24

Price for value and outcomes

As AI products continue to drive greater human productivity, application and infrastructure companies are adopting value-based pricing to align customer spend with successful business outcomes

Application Software

Value-Based Pricing

Infrastructure Software

salesforce

HubSpot

Figma

baseten

unify

1

SIERRA

snowflake

databricks

1

ClickHouse

1

Seat-based pricing

Predictable, stable revenue

Easy for buyers to understand

Simple to implement and measure

# of seats flat / going down as AI does the work

Human productivity value not captured

Seat-based fee

+ pay per action or outcome

Captures AI value

Aligns pricing with successful customer outcomes

Greater willingness to pay

Revenue naturally scales with customer maturity / growth

Minimum platform fee

Consumption-based pricing

Low-friction land

Aligns pricing with usage

Volatile revenue

Complex billing and unpredictable costs for customers

Under monetizes small customers and early usage

24

Note: 1. Denotes a past or current Battery company. For a full list of all Battery investments, please click here.

Battery

Visuals: Logos: left column salesforce cloud logo in #009EDB blue, HubSpot wordmark in dark slate #33475B with orange sprocket, Figma logo in multicolour with black wordmark. Centre column baseten wordmark in black, unify wordmark in black with a rounded icon, SIERRA wordmark in dark green #1F4E3D with a stylised atom-like icon, each with a superscript 1 for unify and SIERRA. Right column snowflake wordmark in #29B5E8 with a snowflake icon, databricks wordmark in black with a red-orange diamond icon and superscript 1, ClickHouse wordmark in black with a yellow and black bar icon and superscript 1. Diagram: two curly braces, one at left enclosing the left column bullets and pointing right to the centre, one at right enclosing the right column bullets and pointing left to the centre. Dotted navy arrow from Seat-based pricing to Seat-based fee with a solid arrowhead pointing right. Dotted navy arrow from the right column Under monetizes item to Minimum platform fee with an arrowhead pointing left. Check and cross icons: green circles with white checkmarks for positive items, red circles with white crosses for negative items, arranged before each bullet. The centre card is a white rounded rectangle with soft shadow containing the pay per action list.

Slides 25 to 35 continue in the full text file.

Download this deck

The hosted PDF of every library deck is free for readers of the SlideDeck Co newsletter. Join with your email and the viewer and download appear here.

Preparing a results presentation or business update of your own? The SlideDeck Co Standard Deck is a slide library for business reviews, updates and plans.

More Reporting decks

Deck reviews in your inbox.

See what works in real company decks, what we’d change, and get the latest SDC releases.

Expanded slide preview