Generic AI Startups Becoming 'Unfundable'

The venture capital landscape has shifted dramatically, with one analysis suggesting the percentage of AI startups considered "investable" dropped from around 80% in early 2023 to just 40% by mid-2023. This is largely due to the high failure rate of "wrapper" applications; some estimates predict 80-95% of these businesses will ultimately fail. The core issue is a lack of a defensible moat, as their features can be quickly replicated by the foundational model providers they rely on, such as OpenAI or Anthropic. A prime example of this vulnerability was Jasper AI, which at one point had a valuation of $1.5 billion for its GPT-3 based content generation tools. However, its value reportedly plummeted when ChatGPT offered similar capabilities for free. This trend of "platform encroachment" is a significant risk, as entire categories of startups can be made obsolete overnight by a single update from a major AI lab. This has led VCs to become wary of startups with no proprietary data or deep workflow integration. The unit economics for simple AI wrappers are also a major concern for investors. Unlike traditional SaaS companies with gross margins of 70-80%, AI wrappers often have margins as low as 25-60% due to the high cost of API calls for every user interaction. This makes it difficult to justify high valuations and achieve long-term profitability, especially when foundational model providers can offer similar services at a lower cost. For engineers in San Francisco's vibrant AI scene, this means the most valuable opportunities are in companies with deep vertical expertise. Startups in sectors like legal tech with companies like Harvey, or those with unique datasets are attracting significant investment. The local ecosystem is buzzing with events like the "SF AI Engineers" meetup and the "AI Tinkerers" build nights, which focus on practical applications and production systems, offering a great way to connect with others in the field. This evolving landscape is also reshaping engineering career paths. The debate between being a generalist versus a specialist is nuanced; generalists may find it easier to adapt in fast-moving startups, while specialists with deep expertise in a high-demand area can command senior roles. The traditional distinction between Individual Contributor (IC) and management tracks is also blurring, as AI tools empower smaller teams and place a greater emphasis on system design and product judgment over just writing code.