The End of Traditional VC: Why AI Demands a Complete Investment Rethink

AI is fundamentally reshaping venture capital, forcing investors to abandon spreadsheet-driven strategies for a more artisanal, mystery-focused approach.

Key Takeaways

• Traditional B2B SaaS investing metrics are becoming obsolete in the unpredictable world of AI startups
• The venture industry's obsession with principal-driven markups is creating a packaging mentality that misses real innovation
• AI companies require investors who can navigate "mysteries" rather than solve predictable "puzzles" with raw data analysis
• Product-centric evaluation beats financial metrics when assessing early-stage AI founders and their potential for long-term success
• Service-oriented VCs who deeply understand founders will outperform transactional investors optimizing for quick deployment and markups
• The most promising AI startups fall into three categories: adaptation, evolution, and revolution - with revolution offering the highest returns
• Future AI value will accumulate around taste and execution speed rather than just computational resources or capital deployment
• Early-stage investors need smaller teams making subjective bets instead of large partnerships following industrialized playbooks and frameworks

The Death of Spreadsheet Investing

Here's the thing about AI investing - everything we thought we knew about evaluating startups just got turned upside down. Nabeel Hyatt from Spark Capital puts it bluntly: the spreadsheet-driven SaaS investing that dominated the last decade is making dinosaurs out of investors who refuse to adapt.

• The classic "18 months to $10 million ARR" gold standard has been completely blown apart by AI companies hitting massive revenue numbers in mere months, only to potentially vanish just as quickly when the next model release reshapes their entire market overnight
• Traditional revenue heuristics fail catastrophically when evaluating AI companies because the underlying technology shifts so rapidly that yesterday's breakthrough becomes tomorrow's commodity, making historical growth patterns virtually meaningless for predicting future success
• What used to be "puzzles" that you could solve with raw analytical horsepower have transformed into "mysteries" where you simply cannot work out the answer ahead of time, requiring fundamentally different investment approaches that prioritize intuition and deep founder relationships over data analysis
• The venture industry's evolution toward hiring armies of associates and principals to grind through metrics has created a system optimized for the wrong type of problem-solving, leaving firms poorly equipped to handle the fog-of-war uncertainty that defines AI investing today

The reality is we've moved from late-stage capitalism for startups - where everything was about optimization and minor arbitrages in red ocean markets - to a world demanding rampant creativity and nuanced decision-making that can't be systematized.

The Principal Problem Destroying VC Returns

What's really happening inside most venture firms today would shock you. The industry has become dominated by principals, associates, and junior GPs who aren't actually waiting for exits - they just want promotions or better jobs at competitor firms within two years.

• This incentive misalignment creates a markup-chasing mentality where principals figure out what the next stage investor wants, then invest one month earlier to secure the markup four months later, completely divorcing investment decisions from long-term founder success or genuine market innovation
• The partnership expansion from seven partners to 25 or 30 partners at major firms has necessitated simple heuristics and approval processes that work directly against the subjective, artisanal decision-making required for breakthrough AI investments where pattern matching fails spectacularly
• Most firms are essentially running a packaging business where the goal is getting deals approved internally and shifted to the next investor, rather than building lasting relationships with founders who are genuinely reshaping entire industries through technological innovation
• The explosion of startups requiring simple yes/no frameworks has led to "Brita filter" investing - dumping everything into the top and hoping a few good companies sift out - instead of proactive, conviction-driven investment strategies that identify exceptional founders before they become obvious

When you're optimizing for performance while your competitor optimizes for deployment, you're playing entirely different games. Smart money increasingly goes to common stock at triple the price, leaving traditional VCs wondering why they're losing great deals.

Product-Centric Evaluation in an AI World

The most successful AI investors are focusing intensely on product evaluation, not because they're product masters, but because the product represents the most authentic window into understanding the humans behind the company and their decision-making processes.

• Product evaluation serves as the primary method for separating genuine executors from sophisticated hustlers who deliver compelling pitches but lack the deep thinking and taste required to build lasting institutions in rapidly evolving technological landscapes
• The questions you ask founders about their product decisions - why they made specific choices, what they're most proud of building, what they're embarrassed about, what they'd build if constraints didn't exist - reveal thinking patterns that predict long-term success better than traditional metrics
• AI founders need to balance execution speed with exceptional taste, creating a dynamic tension where pure execution players get lapped by taste-driven competitors, while pure taste players never ship anything meaningful to market
• In competitive AI markets with 25+ similar companies, success depends entirely on a founder's ability to aggregate innovation happening across all competitors while maintaining enough taste to avoid building me-too products that offer marginal improvements

The founder evaluation process has fundamentally shifted from checking boxes about pedigree and market size to understanding how someone thinks about the craft of building products that users genuinely cannot stop thinking about once they try them.

The Three Categories Reshaping AI Investing

Understanding where AI startups fit in the adaptation-evolution-revolution framework helps investors identify which opportunities offer genuine venture-scale returns versus quick arbitrage plays that'll get commoditized within months.

• Adaptation represents the obvious approach where existing companies add AI features to current products - think Adobe Firefly or Spotify DJ - but these initiatives typically lack defensibility since incumbents with distribution advantages will dominate this category through superior resource allocation and customer relationships
• Evolution creates new workflows and behaviors native to AI capabilities, exemplified by companies like Granola (AI meeting notes) or Descript (AI-native video editing), where the user experience fundamentally changes but the core job-to-be-done remains recognizable to existing market categories
• Revolution builds entirely new platforms that could only exist because AI technology enables previously impossible capabilities, similar to how Uber revolutionized transportation by making real-time coordination feasible through mobile technology and GPS integration
• Most current AI startups fall into weak adaptation or evolution categories because the incubator-driven startup industrialization pushes founders toward quick arbitrage opportunities rather than the patient innovation required for revolutionary breakthrough products

The highest returns historically come from revolution and strong evolution plays, but these require investors willing to take higher risks on longer development timelines rather than chasing obvious near-term opportunities.

Foundation Models vs. Application Layer Value

The debate over where value accrues in AI - foundation models versus application layer - misses the nuanced reality that both can win, but for different strategic reasons related to data collection and customer relationship ownership.

• Foundation model companies like Anthropic and OpenAI benefit from direct user interfaces that generate consumer data exhaust, providing insights into how people actually want to use AI capabilities that purely academic or compute-focused competitors cannot access or replicate effectively
• Application layer companies sitting on expert workflow data - like Descript capturing every editing decision that transforms rough cuts into polished productions - are internalizing wisdom of experts rather than average crowd behavior, which becomes increasingly valuable as AI systems aim for expert-level performance
• The key competitive advantage isn't just model quality or interface design, but the ability to continuously iterate based on direct customer relationships while maintaining execution speed and taste that keeps you ahead of competitors copying your innovations
• Vertical AI applications in previously uninteresting small markets become massive opportunities when you're replacing human labor rather than just improving efficiency, turning broker-calling-truckers businesses into multi-billion dollar addressable markets

Success requires being full-stack enough to own meaningful customer relationships while staying focused enough to build genuinely differentiated capabilities that compound over time.

Navigating Uncertainty with Conviction

The most challenging aspect of AI investing is making large bets in a world where fundamental assumptions can change overnight, requiring investors to develop entirely new muscles for evaluating opportunities under extreme uncertainty.

• Traditional venture heuristics like market size analysis become meaningless when you're creating new markets that didn't exist before, forcing investors to evaluate founder capability for continuous innovation rather than execution within known parameters and established competitive dynamics
• The sustainability of value in AI requires finding founders who can reinvent continuously rather than building defensible moats, since barriers to entry that seem solid today may dissolve completely when the next major model release changes the entire competitive landscape
• Smart investors focus on picking hard jobs that will take a decade to solve properly rather than problems that current AI models can address within a three-month incubator timeline, ensuring their portfolio companies have room to grow and adapt as technology evolves
• Service-oriented investing - being genuinely useful to founders through deep product understanding and industry expertise - becomes increasingly important when founders face unknowable challenges that require experienced sounding boards rather than prescriptive advice

The firms winning in AI are those that treat venture capital as an artisanal business requiring subjective judgment and deep founder relationships rather than an industrial process optimized for pattern matching and rapid decision-making.

This isn't just another market cycle - it's a fundamental reshaping of how innovation happens and how smart capital should position itself. The investors who recognize this shift and adapt accordingly will capture the outsized returns that define the next generation of technology winners.