Interview, Fireside Chat
Alex Kolicich, 8VC: Exodus of Venture Capital, AI Predictions
- AI is anticipated to be overhyped and face infrastructure struggles or deployment disappointments within the short term of one to three years, with a specific "despair period" predicted around two years where the market views AI as limited hype.
- Over the next decade, AI is expected to dramatically transform knowledge work, fundamentally alter every vertical software company, and drive extensive automation in back-office jobs and document processing for industrials and vertical industries.
- Open source models are projected to eclipse closed source models while coexisting in a full ecosystem, with some predictions suggesting open source capabilities will reach state-of-the-art levels within one year.
- Significant challenges in reliability are expected due to the "90-10 problem," where the cost of creating software remains high until the "requirements to code" goal is achieved; once reached, creation costs are predicted to plummet, though high-reliability enterprise deployment will take years.
- Enterprise customers are expected to be less tolerant of AI errors and hallucinations than consumers, and internal workflows are viewed as more vulnerable to LLM attacks than new software protocols.
- The venture capital landscape is forecast to see a general shrinking over the next five years unless AI achieves rapid success, with large funds from 2021 undergoing retrenchment rather than complete closure despite a persisting oversupply of capital.
- Despite a lower VC headcount trend compared to 2019, market conditions remain robust for founders prioritizing value creation and product-market fit, with 2024 expected to be a "boom year" for capital deployment and valuations driven by a returning grassroots Silicon Valley energy reminiscent of 2013-2015.
- Incumbents hold a competitive advantage due to existing user data and distribution, while innovation at major players like OpenAI is expected to remain extraordinary enough to disrupt businesses before term sheets close.
- Value from AI is expected to flow to customers through business automation, but use cases are projected to be far narrower in the near term compared to five years from now, with the AI industry taking longer than anticipated to solve enterprise problems in production.