SaaStr Founder: Growth Is the Only AI Litmus Test
Jason Lemkin, founder of SaaStr, argued that the true measure of an AI company is whether its growth is re-accelerating, typically to 80-110% year-over-year. He stated that simply adding AI features is insufficient, as investors are now scrutinizing companies for tangible, AI-driven revenue impact rather than just technological adoption.
- Venture capital funding for AI startups has become highly concentrated, with global funding reaching $270.2 billion in 2025, over half of all VC investments. Megadeals are now common, such as Anthropic's $13 billion Series F, Databricks' $4 billion Series L, and xAI's $5.3 billion equity funding. - The market for AI infrastructure is projected to hit $758 billion by 2029, with hyperscalers leading the investment surge. In 2025 alone, Amazon plans to spend over $100 billion, Microsoft up to $93 billion, and Google $75 billion on AI infrastructure, primarily for data centers and specialized chips. - To optimize for specific AI workloads and reduce long-term costs, major tech companies like Google (TPUs) and Meta (MTIA) are increasingly designing their own custom silicon, or ASICs. However, the complexity and cost are substantial, with the design of a 5nm chip potentially exceeding $500 million. - The AI Go-to-Market (GTM) tooling landscape is rapidly expanding, with the AI marketing tools market projected to grow from $4.9 billion in 2023 to nearly $13 billion by 2032. Key GTM categories include AI-driven prospecting and sequencing tools like Apollo.io and Outreach, and revenue intelligence platforms like Clari. - For AI startups and enterprise ML teams, the Machine Learning Operations (MLOps) market is critical for scaling models and is projected to grow from $1.84 billion in 2025 to over $84 billion by 2035. This ecosystem includes end-to-end platforms from major cloud providers like AWS SageMaker and Google Vertex AI, which help manage the entire machine learning lifecycle. - The decision for AI companies to "build vs. buy" their core infrastructure involves a trade-off between control and speed-to-market. While buying pre-built platforms from vendors can accelerate deployment, building a custom stack offers greater control over architecture and can be more cost-effective at scale, a crucial consideration for teams with proprietary workflows. - Investment in the AI hardware sector is surging, with multiple $100M+ deals for AI chip and robotics startups in late 2024 and over $1 billion flowing into AI-specific hardware in Q4 2025 alone. Notable recent funding rounds include Unconventional AI's $475 million seed round for co-designing analog AI chips and software. - The growth of AI is driving a massive expansion of data center infrastructure, with the market for data center servers projected to grow nearly fivefold from $204 billion in 2024 to $987 billion by 2030. This is creating new challenges and opportunities in server design, rack-scale architecture, and liquid cooling technologies to handle the increased power density of AI workloads.
