Companies Are Capturing Billions in AI Spend – Where’s Your Share?
The team had time. You had time. You had all of 2025. The single biggest shift in IT budgets in a generation is happening right now. AI spending represents roughly 60% of GDP growth in tech.
If you haven’t captured ANY of this massive spending wave by end of 2025, you don’t need half your team. They had 18 months.
There is an unprecedented amount of money flooding into AI right now. Your job as a founder isn’t to become the next Harvey or OpenAI overnight. Your job is to capture at least some of this spend. Even a tiny slice, and see growth accelerate to some extent from it. And if you haven’t done that by now, your team failed.
The Money is Staggering and Real
Look at what’s actually being spent:
- OpenAI is projecting $100+ billion in revenue by 2027. They just raised commitments for over $1 trillion in capex over the next 5 years. Yes, trillion with a T.
- Anthropic is now projecting $70 billion in revenue by 2028. They’ve massively raised their estimates.
- Microsoft can’t build data centers fast enough. They’re capacity constrained because demand is so high. They’ve committed $250 billion to OpenAI alone.
- Oracle committed $300 billion. Broadcom committed $400 billion.
- Amazon Web Services reaccelerated to 20% growth, and they’re late to the party. They’re the 5th or 6th partner to OpenAI for GPUs – behind Microsoft, Oracle, Google – and they still found enough demand to grow 20%.
This isn’t theoretical future spending. This is happening right now, today.
You Don’t Have to BE an AI Company to Capture AI Spend
Here’s what founders miss: You don’t need to be an “AI startup” to capture billions in AI spend. You just need to be selling something that AI companies need.
Look at these perfect examples:
RevenueCat: $1B/Month Through Their Platform
RevenueCat powers in-app purchases and subscriptions. Not exactly “AI” right? But look at their growth trajectory:
- First API Call: Growing from ~1K to 4K+ per day (30-day rolling sum)
- First Invoice: Surging from ~1K to 3K+ per day
- First Purchase: Exploding from ~1K to 3.5K+ per day
Miguel Carranza (their CEO) is hiring aggressively because they’re processing $1B+ per month through their platform with customers — including OpenAI. Why the explosion? AI apps need monetization infrastructure. Every AI consumer app that charges users needs what RevenueCat built. And vibe coding in general has led to an explosion of new mobile apps.
They didn’t pivot to AI. They didn’t rebrand as “AI-powered billing.” They just positioned themselves where the AI money was flowing and captured it.
WorkOS: $20M to $30M ARR in 5 Months
WorkOS provides authentication and enterprise-ready features (SSO, SCIM, audit logs). Michael Grinich announced they went from $20M ARR to $30M ARR – that’s 50% growth in just a few months.
On our podcast, I specifically called out WorkOS: They went from $20M to $40M in 5 months as AI companies scaled. Why? Every AI startup needs OAuth and authentication. Every single one.
Look at their customer logos: HeyGen, Clay, Copy.ai, Bolt, Superhuman, Clari, Scribe – these are all AI-first companies or companies heavily investing in AI. WorkOS positioned themselves perfectly to capture that spend.
The lesson: You don’t need to build AI models. You need to sell to companies building AI.
Every Layer of the Stack is Capturing It
Companies at every level are getting their share. If they’re not, they’re dying. It’s where all the growth it. So, so much of it.
Infrastructure Layer
Twilio is the perfect case study. They went from single-digit growth (and falling) to 15% year-over-year growth, reaching $1.3 billion in Q3 2025.
But here’s the real story: AI customers are contributing $260 million in annual revenue as of Q2 2025. That’s from over 9,000 AI companies building on their APIs.
The acceleration is stunning:
- Revenue from their top 10 voice AI startups increased more than 10x year-over-year in Q3 2025
- Voice revenue hit mid-teens YoY growth – the fastest pace in over three years
- 58% of non-profits and 47% of B2C businesses are using AI with Twilio’s CPaaS solutions
- Twilio boosted throughput by 6,000% for OpenAI’s 1-800-ChatGPT launch, proving they’re critical infrastructure
They have strategic partnerships with OpenAI, Amazon AWS, and Google. AI companies need what Twilio built for communications. That $260M in AI revenue? That’s just the beginning.
MongoDB – same story. Every AI company needs a database with vector search. They’re capturing that spend.
Cloudflare – capturing it. AI companies need CDN and security.
DataDog – capturing it. AI companies need monitoring and observability.
These aren’t AI-native companies. They’re infrastructure companies from the 2010s that found ways to co-attach to the AI spending wave.
Application Layer
Palantir might be the most stunning example of capturing AI spend. They’re trading at 123x revenue because they’ve proven they can monetize AI at enterprise scale. Their AIP (Artificial Intelligence Platform) is driving unprecedented customer demand.
They’re not just talking about AI – they’re showing actual revenue impact from enterprises adopting their AI agents. When a company trades at 123x revenue and the market believes it, that tells you everything about the value of capturing AI spend.
Salesforce built Agent Force. It’s actually good – not vaporware. They have 2,000 people building it and it’s competitive with any agent platform you can buy.
HubSpot is shipping AI products. They haven’t fully monetized yet, but they’re positioned.
Even companies in slower-adoption verticals are shipping products and preparing to capture this spend.
The New AI Companies
Harvey: $150M ARR, $8B valuation, growing like crazy. 170% NDR. They’re capturing law firm AI spend.
Open Evidence: 300,000 doctors in one year (took Doximity 10 years to hit that). They’re capturing healthcare AI spend.
Hundreds of AI-native companies are absorbing billions in new budget.
The Budget is There – Go Get Your Share
There are trillions being spent on AI infrastructure, models, and applications.
A million companies are using voice AI agents right now. AI is driving massive database usage. Companies are deploying AI agents for sales, support, research, legal work, coding – everything.
At SaaStr, we track this obsessively. On our SaaStr.ai/agents page where we share the agents we use, we get 12,000 views per month – and it came out of nowhere. We’ve sent millions in revenue to vendors like Artisan and Qualified in just a couple months because we actually use their products and they work.
The demand is insatiable. When you can genuinely replace work with AI – not pretend, not for demo day, but actually – customers cannot buy fast enough.
You Had 18 Months to Capture SOMETHING
This is where I lose patience with founders and teams.
You had 18 months since the AI spending wave started. ChatGPT launched in late 2022. Enterprise AI budgets exploded in 2024. By now, you should have:
- Shipped AI features that customers actually want
- Seen some revenue acceleration from AI adoption
- Captured at least a few basis points of the massive AI spend
I’m not expecting you went from 30% growth to 300% growth. That’s unrealistic for most companies.
But going from 15% to 25% growth? From 10% to 20%? That should be table stakes if you’re remotely positioned in a category where AI spending is happening.
As I said on our 20VC pod this week:
“Going into next year, you sure better have seen re-acceleration because of AI because there’s so much money there. If you got none of it guys, you had 18 months to not launch a single feature, tap into a single trend to reaccelerate. If you’re not growing faster at the end of 2025 than the start as a founder, I give you an F minus. There’s so much money and you don’t have to be Harvey to get a little piece of it. Twilio and MongoDB and Cloudflare and tons of folks have a little piece of it. A little piece of that massive pie.”
The Point Spread Between Winners and Losers
Rory O’Driscoll nailed this on our show: There’s a massive point spread between companies that captured AI spend and those that didn’t.
Companies with NO AI lift:
- Getting taken private at 3x revenue multiples
- Getting absorbed by PE firms and destroyed
- Declining relevance in their markets
- Struggling to retain top talent who want to work on AI
Companies with SOME AI lift:
- Trading at 6-7x revenue with growth stories (or 123x if you’re Palantir)
- Reaccelerating to mid-20%+ growth rates
- Attracting top talent
- Positioned for strong exits or IPOs
That difference – from 3x to 7x (or 123x), from declining to growing – comes down to whether you captured AI spend.
Even Navan (TripActions), which just IPO’d at $700M+ revenue growing 32%, was seen as “struggling” at a $5B valuation. Why? Because in 2025, if you’re not capturing massive AI budget, you’re leaving money on the table.
Where the Money is Flowing (And How to Capture It)
Let me break down where this spend is actually going, so you can see if you’re positioned to capture any of it:
1. AI Infrastructure ($100B+ annually)
- Cloud compute for AI training and inference
- GPUs and specialized AI chips
- Networking and data center buildout
- If you sell infrastructure, you should already be seeing massive growth
2. AI Development Tools ($10B+ annually)
- Model development and fine-tuning platforms
- AI coding assistants (Cursor, Replit, etc.)
- MLOps and model deployment
- Vector databases and embedding stores
3. AI Applications Building on Your Platform ($50B+ annually)
- This is the WorkOS and RevenueCat play
- Authentication, billing, payments, compliance
- Developer tools, APIs, infrastructure
- Any horizontal service that AI companies need
- You don’t have to build AI – just sell to AI builders
- Twilio captured $260M annually from 9,000+ AI companies this way
4. AI Agents and Applications ($50B+ annually and growing)
- Industry-specific AI agents (legal, healthcare, finance)
- Horizontal AI tools (sales, support, operations)
- AI-powered productivity software
- This is where most B2B SaaS companies should be capturing spend
- Palantir is the gold standard here – 123x revenue multiple
5. AI Services and Integration ($20B+ annually)
- Consulting and implementation
- Custom model development
- Integration and workflow automation
- If you have services revenue, you should be capturing some of this
What “Capturing AI Spend” Actually Looks Like
Let me give you concrete examples from companies actually doing this:
RevenueCat’s Approach:
- Provides subscription and billing infrastructure
- Every AI consumer app needs to charge users
- Positioned as the default billing system for mobile AI apps
- Result: Processing $1B+ per month, explosive growth across all metrics
WorkOS’s Approach:
- Built authentication and OAuth layer
- Every AI company needs enterprise-ready auth
- All the hot AI startups use them (HeyGen, Clay, Copy.ai, etc.)
- Result: $20M to $30M ARR in 5 months (50%+ growth)
Twilio’s Approach:
- Built AI-first voice capabilities and CPaaS solutions
- Partnered strategically with OpenAI, AWS, Google
- Voice AI products integrated with AI agent platforms
- Result: $260M annually from 9,000+ AI companies, top 10 voice AI customers up 10x YoY
- Outcome: Reacceleration from single digits to 15% growth, $1.3B quarterly revenue
Palantir’s Approach:
- Built AIP (Artificial Intelligence Platform) for enterprises
- Focused on real deployment and ROI, not demos
- Proved AI agents can drive measurable business value
- Result: Trading at 123x revenue because market believes in AI monetization
Our Approach at SaaStr:
- Deployed 20+ AI agents throughout 2025
- Generated $1.5M in revenue while investing $500K in AI
- Built AI tools that are used 500,000+ times
- Result: We’re capturing AI spend and we’re an 8-figure events business, not even a software company
Harvey’s Approach:
- Built AI specifically for lawyers (huge TAM: 1M lawyers in US)
- Focused on genuinely replacing associate work
- Result: $150M ARR, 170% NDR, $8B valuation
The common thread? They found where the AI money was flowing and they built products to capture it.
Three Ways to Capture AI Spend (Pick One)
If you’re not sure how to position your company, here are the three proven strategies:
Strategy 1: Build AI Products for Your Existing Market
- What Salesforce is doing with Agent Force
- What HubSpot is doing with AI features
- What Palantir did with AIP
- Requires significant product investment
- Takes longer but defends your core business
Strategy 2: Sell to AI Companies
- What WorkOS and RevenueCat are crushing
- What Twilio did to capture $260M from 9,000+ AI companies
- Position your existing product as essential infrastructure for AI builders
- Often just requires go-to-market shift, not major product changes
- Can happen very fast (WorkOS: 50% growth in 5 months; Twilio: 10x growth from top AI customers)
Strategy 3: Enable AI Use Cases in Your Platform
- What Twilio did with voice AI capabilities
- What MongoDB does with vector search
- Add features that make you the obvious choice for AI workloads
- Moderate product investment, big upside
If you’re not pursuing at least ONE of these strategies, what are you doing?
Your Team’s Job is Simple: Find the Money
I don’t care if you’re in infrastructure, applications, vertical SaaS, horizontal SaaS, or services.
Somewhere in your TAM, companies are spending money on AI. Your customers have AI budgets. Those budgets are growing faster than any other category in IT spend.
Your team’s job is to:
- Figure out where the AI spend is in your category
- Build products that capture some of that spend OR position existing products for AI buyers
- Ship those products in 2025 OR shift your GTM to AI companies
- Show revenue acceleration by year-end
If your team hasn’t done this by now, what exactly have they been doing?
They’ve had 18 months. The opportunity was obvious. The budgets were real. The customer demand was there.
Twilio found 9,000+ AI companies to sell to. WorkOS repositioned to capture AI startup spend. Palantir built enterprise AI that commands 123x revenue multiples. What’s your team’s excuse?
The “It’s Too Late” / “It’s Too Early” Paradox
Here’s the mind-bending thing about AI in late 2025:
It feels late because we’ve been talking about it for years. ChatGPT feels old. Everyone has an AI strategy. The Cursor and Replit AI coding stories feel played out.
But it’s actually still incredibly early. It’s only been 3 years since ChatGPT shipped. Most enterprise adoption is just beginning. We’re maybe 10-20% into the total AI transformation of enterprise software.
The problem isn’t that it’s too late to build AI products. The problem is it’s too late for YOUR SPECIFIC TEAM to prove they can execute on this opportunity.
If you haven’t shipped by now – or haven’t repositioned to sell to AI companies – you’ve proven you’re too slow. You’ve proven you don’t have the right people. You’ve proven you can’t capitalize on the single biggest shift in IT spend in a generation.
Fire Yourself or Fire Half Your Team
This is where I get harsh, but it needs to be said:
If you haven’t captured AI spend by now, fire half your team. Give them a turkey and three months of severance before the holidays, but they failed.
You don’t need people who had 18 months to either:
- Ship AI products
- OR reposition to sell to AI companies
- OR enable AI use cases
…and did none of those things.
Your customers are spending billions on AI. Companies like WorkOS went from $20M to $30M in 5 months. Twilio is getting $260M annually from AI companies. RevenueCat is processing $1B+ per month. Palantir is trading at 123x revenue. Your competitors are capturing that spend. Your team sat there and did… what exactly?
After Thanksgiving 2025, there are no more excuses.
Where’s your AI agent? Where’s your AI feature that’s driving expansion? Where’s your reacceleration from selling to AI companies? Where’s your repositioning to capture AI budget?
If the answer is “we’re still working on strategy” or “we’re evaluating our options” – fire the people giving you those answers.
The Exception: Know Your Adoption Curve
I’ll give you one out: Some verticals and customer segments are genuinely slower to adopt AI.
Marc Benioff said maybe only 2-3% of Salesforce’s enterprise customer base is fully ready for AI agents today. If you’re selling to retail, manufacturing, government, or other slower-moving verticals, you might have more time.
But you still need to be positioned. You still need to have shipped products. You still need a clear plan for how you’ll capture AI spend as your customers’ adoption accelerates.
And honestly? If you’re in a slow vertical, maybe you should be shifting to faster ones. WorkOS didn’t wait for their legacy customers to adopt AI – they went after every hot AI startup. RevenueCat positioned themselves for the wave of AI consumer apps. Twilio found 9,000+ AI companies to sell to. That’s where the money is flowing right now.
What Success Looks Like in 2026
Going into next year, here’s what I expect from any B2B SaaS scale-up that wants to survive:
Minimum bar:
- AI features shipped and in production OR clear GTM motion to AI companies
- At least some customers using those features OR some AI companies as customers
- At least some revenue impact visible in the numbers (aim for Twilio’s $260M from AI customers)
- Clear roadmap for expanding AI capabilities OR AI customer base in 2026
Good execution:
- Measurable revenue acceleration from AI products or AI customers
- 5-10 percentage points of growth rate improvement
- Strong customer adoption metrics (like Twilio’s 58% of non-profits using AI)
- Pipeline building for AI features or AI customers
Great execution:
- Significant reacceleration (10-20+ points of growth improvement)
- AI products or AI customers becoming primary revenue driver
- Clear market leadership in AI for your category
- Ability to price at premium because of AI capabilities (like Palantir’s 123x multiple)
If you’re not at least at the minimum bar by end of 2025, you’re in trouble.
The Brutal Math of Missed Opportunity
Let me paint the picture of what you’re missing:
Scenario 1: You captured 1% of AI spend in your category
- Your category has $10B in total spend
- $1B is shifting to AI products or AI company budgets
- You capture 1% = $10M in new revenue
- Result: Meaningful growth acceleration, higher valuation multiple
Scenario 2: You repositioned to sell to AI companies (the WorkOS/Twilio play)
- You had $20M ARR in traditional customers
- You shift focus to AI startups who need your product
- You add $10M in 5 months from AI companies (or $260M annually if you’re Twilio)
- Result: 50% growth rate, explosive momentum
Scenario 3: You captured 0% of AI spend
- That same $1B goes to competitors and new entrants
- Your growth rate stagnates or declines
- Your valuation multiple compresses
- Result: PE buyout at 3x revenue instead of IPO at 7x revenue (or 123x if you executed well)
The difference between Scenario 1 or 2 and Scenario 3 for a $100M ARR company:
- Success: $700M+ exit valuation (or billions if you’re Palantir), growing company, happy team
- Failure: $300M PE buyout, flat company, demoralized team
That’s a $400M+ difference (or much more) because your team couldn’t capture a tiny fraction of the massive AI spending wave OR couldn’t reposition to sell to AI companies.
The Bottom Line: Go Get Your Share
The single biggest reallocation of IT spend in a generation is happening right now. Trillions of dollars are flowing into AI infrastructure, development, and applications.
Your job as a founder isn’t to capture all of it. It’s not even to capture a big chunk of it.
Your job is to capture SOME of it.
A little piece of the Harvey pie. A fraction of the Twilio voice AI growth ($260M from 9,000 AI companies). A sliver of the MongoDB database expansion. Some of the Cloudflare infrastructure spend.
Or do what WorkOS and RevenueCat did: Position your existing product to sell to the explosion of AI companies. You don’t have to build AI. You just have to sell to AI builders.
If you haven’t done either of these by now – if you can’t point to revenue growth that came from AI budget OR from AI companies – you have a team problem.
Fire half your team, or fire yourself. Because the money was there. The opportunity was there. And your team proved they couldn’t execute.
The age of AI isn’t coming. It’s here. The budget is flowing. Go capture your share or get out of the way.
