👋 Hey, Sahil here - welcome back to Venture Curator, where we explore how top investors think, how real founders build, and the strategies shaping tomorrow’s companies.
Big idea + report of the week :
Anthropic modelled the AI economy of 2030. Who actually wins? The answer might surprise you.
Are AI startups getting acquired too early? Here’s what the data shows.
Frameworks & insightful posts :
Scaling vs. Profitability Problem: Where Venture Capital often goes wrong.
Can vertical AI startups still win when incumbents own the data?
How much should founders actually pay themselves? - NFX framework.
Get access to 150+ premium archive posts, 100+ startup & VC resources, investor databases, fundraising templates, and exclusive startup research - all in one place.
FROM OUR PARTNER - MOREVY
Still rebuilding your recurring revenue analysis in spreadsheets every month?
Morevy gives finance teams a simpler way to understand what changed in the business without digging through spreadsheets every time.
Connect Stripe or upload your billing data, and Morevy turns it into a reconciled view of new revenue, expansion, contraction, churn, and reactivation. Then you can simply ask questions about the numbers in plain language and trace every answer back to the customers and source data behind it.
And unlike typical AI analysis, the AI isn’t calculating your financials. The numbers are computed first, then AI helps you understand and explain what happened.
For finance leads and fractional CFOs managing recurring revenue, it means less time rebuilding reports and more confidence in the numbers you share.
WORK WITH US
🤝 Partnership With Us
Get your product in front of an audience of 125,000+ - Our newsletter is read by thousands of tech professionals, founders, investors and managers worldwide. Get in touch today.
START WITH
🧠 Big idea + report of the week
Anthropic modelled the AI economy of 2030. Who actually wins? The answer might surprise you.
The biggest question around AI may not be whether it creates economic growth. It may be who actually captures that growth.
Anthropic’s new Economic Scenario Explorer models what the US economy could look like by 2030 under three different paths for AI adoption: modest, substantial and extreme. Rather than assuming entire jobs disappear, the model treats each occupation as a bundle of tasks that AI can augment, automate, leave unchanged or help create.
The differences between the scenarios are enormous.
Modest: US GDP reaches $34.1T by 2030, just 1.6% above the economy without AI.
Substantial: GDP reaches $36.3T, an 8.3% boost. AI can perform roughly half of knowledge work, although adoption remains incomplete.
Extreme: GDP reaches $44.4T, 32.4% higher than without AI, with annual GDP growth eventually reaching roughly 15%.
But the most interesting part isn’t GDP. It’s what happens to workers as AI becomes more capable.
In Anthropic’s substantial scenario, knowledge-worker wages are essentially flat at -0.3% relative to a world without AI, while wages for other workers rise 5.9%.
In the extreme scenario, the divergence becomes dramatic: knowledge-worker wages fall 11.5%, while wages for other workers rise 33.6%. Knowledge-worker unemployment rises to 17.9%, pushing overall unemployment to 11.9%.
There is an equally important shift in who captures the economic value.
Today, roughly 60% of economic output goes to labour and 40% to capital. Anthropic estimates capital’s share could rise to 43.9% in the substantial scenario and 54.8% in the extreme scenario, leaving labour with just 45.2%.
So AI could create the strange situation where the economy becomes dramatically more productive and wealthier, while many knowledge workers simultaneously experience weaker wages and fewer job opportunities.
Of course, these aren’t forecasts. Anthropic explicitly describes the model as a simplified scenario tool, and the extreme case requires assumptions such as recursively self-improving AI, rapid adoption and very limited creation of new knowledge-work tasks.
Still, it raises a fascinating question.
If AI creates trillions of dollars of additional economic value but increasingly shifts that value from labour toward capital, owning productive assets may become much more important than simply having a highly valuable skill.
Are AI startups getting acquired too early? Here’s what the data shows.
Founders are usually taught to plan around a long runway to exit - build for years, not quarters. The standard model assumes a decade or so of grinding before an acquisition or IPO even becomes realistic.
But new research from Madrona suggests that timeline has quietly collapsed for AI-native companies specifically and most founders haven’t repriced their own plans around it yet.
Madrona tracked five years of its Intelligent Applications 40 list alongside its own portfolio and exit data, looking at how and how fast - AI companies are actually getting bought.
Roughly half of GenAI acquisitions in 2026 involve companies under five years old. In 2022, that number was close to zero.
This isn’t a fluke sitting off to the side of the data - it shows up directly in Madrona’s own exit examples. Statsig was acquired by OpenAI for $1B; Neon was acquired by Databricks for $1B. Neither is a decade-old company grinding toward a traditional outcome.
Hyperscalers aren’t waiting for AI companies to mature into obvious acquisition targets anymore. They’re buying teams and capability years earlier than the old M&A timeline assumed.
There’s a second trend that explains why this is happening.
Staying independent in AI right now is unusually hard: capital is concentrating at the very top, with 90% of the $366B raised by the last two years’ top AI companies going to just three names - OpenAI, Anthropic, and Databricks - while being a “category leader” on Madrona’s own list lasts about a year before 68% of it turns over.
In a market that reshuffles that fast, a two-year-old company with real technical differentiation may already represent as much strategic value to an acquirer as it ever will.
So why are hyperscalers moving so much earlier than they used to?
It’s not that young AI companies are worth less and getting bought cheap out of desperation. It’s closer to the opposite - in a market moving this fast, waiting five extra years for a company to “mature” the way SaaS companies used to often means watching it get displaced instead. Buying early is a bet on capability and speed, not a discount play.
And this is where founders need to be careful.
Faster exits sound like good news on the surface - less time to liquidity, less risk of getting displaced before you cash out. But it isn’t automatically better, and it isn’t automatically available to you just because you’re AI-native. Cap tables, vesting schedules, and investor return timelines are still mostly built around the decade-long hold period this data says is disappearing. A cap table structured for a long grind can leave both founders and early employees with weaker terms if an acquisition offer shows up in year three instead of year eight.
The real risk isn’t that your company gets acquired too early. It’s building and negotiating, and structuring your cap table - for a timeline the market has already proven it no longer runs on.
📄 Must-Read Post
SOMETHING MORE
🧩 Frameworks & insightful posts
Scaling vs. Profitability Problem: Where Venture Capital often goes wrong.
Startups are constantly told to scale. Raise more capital. Hire faster. Enter new markets. Grow revenue. Figure out profitability later.
But Aswath Damodaran makes an important point: not every business becomes better when it gets bigger.
The real question is whether the economics improve as the company scales.
A business is naturally easier to scale when it has a large and growing market, low capital requirements, strong unit economics, limited customer inertia, and real competitive advantages.
If those conditions are missing, growth can create the illusion of progress while making the underlying business worse.
Damodaran describes a few very different outcomes -
Lightning in a Bottle: Revenue and profits scale together. Early Google and Facebook are examples.
Field of Dreams: The company loses money while scaling, but eventually the economics improve. Amazon is the classic example.
Field of Nightmares: Revenue scales, capital keeps coming in, but the business model never becomes profitable.
Niche Star: The company deliberately stays smaller because scarcity, pricing power, or specialisation creates more value than maximum scale.
The problem is that venture capital naturally pushes companies toward the first three paths.
VC returns follow a power law. A small number of investments generate most of the returns, so investors are incentivised to look for companies capable of becoming enormous.
That can create a mismatch.
A founder might have a healthy $50M business opportunity, but the VC model may need that company to pursue a $5B outcome.
More capital then makes aggressive expansion possible, even when the underlying economics do not justify it.
And this has become easier over time.
Private companies can now raise much more money and stay private longer. They can reach huge revenue numbers and valuations before proving that the business can consistently generate profits.
So capital can sometimes delay the moment when a weak business model has to confront reality.
So -
Scaling is valuable only when growth improves the economics of the business. Otherwise, you are not fixing the problem. You are financing a larger version of it.
For founders, a better question than “How fast can we grow?” is:
“What happens to our unit economics, competitive advantage, and profitability when we become 10x bigger?”
If those things improve, scale aggressively. If they don’t, staying smaller may actually create more value.
Can vertical AI startups still win when incumbents own the data?
One of the biggest questions in vertical AI right now is surprisingly simple:
If Salesforce, DocuSign, Atlassian, ServiceNow and other incumbents already own the customer data and workflows, why does a startup need to exist?
The argument has become stronger as general AI agents improve. A customer could theoretically connect its existing systems to Claude or another general-purpose agent and let that agent work across them.
Seema Amble from a16z recently shared a useful framework for thinking about where vertical AI startups can still win.
The key distinction is between owning the record and owning the job.
An incumbent might own the CRM record, contract, support ticket, or employee data.
But the actual work customers need completed is usually much bigger.
A contract is not the entire legal matter. A CRM opportunity is not the sale. A support ticket is not the complete customer resolution.
Those jobs involve information, decisions and actions spread across multiple systems and people.
That creates three potential layers:
Incumbents own the record: data, permissions and actions inside systems like CRM, HR or legal software.
AI labs can own the front door: general agents can potentially coordinate work across several applications.
Vertical AI startups can own the job: they can specialise in completing one specific workflow from beginning to end.
The opportunity becomes clearer when you look at how AI agents are evolving.
Amble describes an agent hierarchy with four levels.
Retrieval: The AI reads. It searches information, summarises documents, analyses data and drafts responses.
Process: The AI does. It updates records, invokes tools, routes approvals and completes predefined workflows.
Policy: The AI decides within rules. It uses company playbooks, precedents and thresholds to handle situations requiring some judgment.
Principal: The AI decides what should be done. It makes open-ended decisions involving strategy, risk and resource allocation.
Most enterprise AI products started around retrieval.
But incumbents are moving upward.
DocuSign is moving from contract review toward applying legal playbooks. Salesforce is building specialised sales and quoting agents. ServiceNow is moving from workflow agents toward governed execution.
That means simply building a better chatbot on top of an incumbent system is becoming a much weaker startup strategy.
The stronger opportunity is learning how to perform the entire job better than either the incumbent or a general-purpose agent.
And this is where vertical specialisation becomes important.
A vertical AI startup can observe not only the final result, but also the decisions, corrections and expert feedback that produced it.
Consider legal work.
A signed contract tells you what everyone eventually agreed to.
It doesn’t necessarily tell you which alternatives were considered, why a lawyer rejected certain language, which risks required escalation, or why an exception was approved.
Capturing those decisions creates a learning loop.
The AI performs work → an expert reviews it → corrections are captured → outcomes are measured → the system improves on the next job.
That can become much more valuable than simply having access to historical documents.
Harvey offers an interesting example.
Instead of waiting years to accumulate customer data, it created roughly 1,750 simulated legal-task environments using synthetic data, public legal information and expert-created examples. Each environment represented realistic legal work and included detailed criteria for evaluating whether the result was good.
In other words, Harvey manufactured its own curriculum.
That suggests a useful framework for founders evaluating vertical AI opportunities.
Ask four questions:
Can an expert quickly identify what the AI did right or wrong?
Does the work require real judgment rather than simple rules?
Does the job happen frequently enough for the system to keep learning?
Can you start with one task and gradually expand until you own the entire job?
Markets where all four are true could be particularly attractive.
Legal, accounting and tax are obvious examples, but the same pattern can exist in manufacturing, healthcare operations, logistics, construction, and many other industries where work crosses multiple systems and expert judgment still matters.
The bigger takeaway is that access to data alone may not be the moat vertical AI founders should chase.
Incumbents already have enormous amounts of data. General AI agents will increasingly be able to access that data.
The harder advantage may be owning the workflow deeply enough to understand what good work looks like, why experts make particular decisions, and how to improve the next result.
So the emerging AI stack may look something like this:
Incumbent owns the record.
AI lab owns the interface.
Vertical AI startup owns the job.
For vertical AI founders, that last layer may still be very much up for grabs.
How much should founders actually pay themselves? - NFX framework.
Founder salary is one of those decisions that sounds simple until you actually have to make it.
Pay yourself too little and personal finances start becoming a distraction. Pay yourself too much and you burn runway while slowly starting to think more like a salaried executive than an owner.
NFX recently shared a useful framework for thinking about founder compensation, and the interesting part is it starts with one principle:
Pay yourself enough that money doesn’t distract you from building, but keep equity and the long-term outcome as the real financial reward.
That means the right salary will be different for every founder.
Someone supporting a family and paying a mortgage has different needs from a 22-year-old founder living cheaply. Trying to copy another founder’s salary misses the point.
NFX breaks the philosophy into three ideas.
Feel like an owner, not an employee
Founders shouldn’t benchmark themselves against what they could earn at OpenAI, Anthropic, or another large technology company.
You chose a different economic model. An employee primarily gets paid through compensation. A founder takes more risk because the potential reward sits in the equity.
The goal is therefore not to maximize salary every time the company raises more capital. It is to preserve enough ownership and upside that increasing the value of the company remains far more meaningful than increasing your paycheck.
But don’t make yourself unnecessarily poor
The opposite extreme can be just as damaging.
Some founders believe paying themselves almost nothing demonstrates commitment.
But if you’re accumulating personal debt, worrying constantly about bills, or creating financial pressure for your family, that stress eventually affects how you run the company. It can even change your incentives.
A founder under serious personal financial pressure may become more willing to accept an early acquisition simply because it solves their immediate financial problem.
The better approach is to pay yourself enough to live comfortably, maintain a reasonable personal buffer, and stay focused on building for the long term.
Don’t turn fundraising into a salary increase
Closing a funding round feels like progress, but NFX makes an important distinction:
Fundraising success is not the same as company success.
Raising $5M doesn’t automatically mean the founders should suddenly start paying themselves much more.
Founder compensation can change, but there should be an underlying reason.
NFX suggests that legitimate reasons might include a meaningful life change, becoming heavily diluted, the company genuinely reaching escape velocity, or completing a strategic secondary when the company is performing strongly.
What shouldn’t drive the decision?
“We just raised.” Or: “Other founders are earning more.”
There’s also a second-order effect founders sometimes overlook.
Your own compensation influences how the rest of the organization thinks about compensation. A very high founder salary can make it harder to maintain discipline when negotiating packages with senior hires.
The same thinking applies to secondaries.
Taking some money off the table isn’t necessarily bad. At the right stage, it can actually align founders and investors by removing enough personal financial pressure that the founder doesn’t feel compelled to sell the entire company too early.
But doing a secondary in every round defeats the purpose.
The broader lesson is that founder compensation shouldn’t be optimized around maximizing or minimizing salary.
It should be optimized around alignment.
You should earn enough that personal finances don’t interfere with your ability to build the company, while keeping enough of your economic upside tied to creating something genuinely valuable.
A useful question to ask yourself is:
“Does my compensation make it easier for me to think like the long-term owner of this company?”
If the answer is yes, the exact salary number matters a lot less.
📬 Upgrade to Venture Curator Premium
Get access to 150+ premium archive posts, 100+ startup & VC resources, investor databases, fundraising templates, and exclusive startup research - all in one place.
TODAY’S JOB OPPORTUNITIES
💼 Venture capital & startup jobs
2027 Venture Capital Summer Associate - StepStone Group | USA - Apply Here
Head of Finance - Hivemind Capital Partner | USA - Apply Here
Associate - Marble | France - Apply Here
Senior Associate, Talent Sourcing - Redesign Health | India - Apply Here
Vice President, Investor Relations - Red Cell Partner | USA - Apply Here
Senior Associate - BVP | USA - Apply Here
Investor - Superseed | UK - Apply Here
Associate / Principal, Follow-on Investing - Village Global | USA - Apply Here
Senior Associate - Yamaha Motor Venture | USA - Apply Here
MBA Intern - Montauk Capital | USA - Apply Here
AI Analyst - Array Venture | USA - Apply Here
SMB Program Associate - Plug and Play Tech Centre | USA - Apply Here
VC Fellow - V11 | Remote - Apply Here
Tips to Break Into VC
What should your CV look like for a VC job?
Most VC applicants are not rejected because they lack experience. Their CV fails to show the three things funds actually look for: sourcing, judgment, and hustle.
This deep dive breaks down how to structure a VC CV, rewrite your experience for investing roles, build proof of work, and fix your LinkedIn profile. It also includes before-and-after examples, a 20-point screening checklist, and practical guidance for consultants, bankers, students, operators, and engineers.
Partnership With Us
Get your product in front of an audience of 125,000+ - Our newsletter is read by thousands of tech professionals, founders, investors and managers worldwide. Get in touch today.
Upgrade to Venture Curator Premium
Get access to 150+ premium archive posts, 100+ startup & VC resources, investor databases, fundraising templates, and exclusive startup research - all in one place.
(Subscribe to Venture Curator Premium and get 20% off forever.)