Assessing the value of AI companies and the hype cycle

From: redpointai

A discussion among Redpoint Ventures partners Scott Rainey, Patrick Chase, Alex Bard, and Jacob Efron at their annual meeting delves into the complexities of investing in AI, particularly focusing on where value is accumulating and how to navigate the current high-valuation environment [00:00:12].

The Staggering Scale of AI Investment

The projected revenue for the NVIDIA data center division, which powers AI training and inference, is $177billion,5.5timesthesizeofthepersonalcomputingmarket<aclass="yt-timestamp"data-t="00:01:47">[00:01:47]</a>.Thisimpliesthatcompaniesmakingthesehardwareinvestmentsanticipategeneratingrevenuesof$1.2 trillion by 2030 and $1.5trillionby2032fromtheAIlandscape<aclass="yt-timestamp"data-t="00:02:29">[00:02:29]</a>.Thisprojectedgrowthof$1.5 trillion in 10 years is staggering when compared to the world enterprise software market, which is about $1.1 trillion built over 50 years [00:02:47].

Alex Bard expresses hope that this number is “understated” due to the immense excitement from technology leaders like Marc Benoff, Bill Gates, Jeff Bezos, and Eric Schmidt about AI’s potential [00:03:24].

Strategic Imperative over Immediate ROI

Companies are compelled to make massive AI investments, regardless of immediate ROI, because it’s a “strategic imperative” to avoid being left behind [00:04:11].

Economic Impact: Service as Software

AI is shifting the market from “software as a service” to “service as software,” where AI performs human jobs rather than merely making humans more efficient [00:04:36]. This shift unlocks significantly larger budgets, as labor budgets are often an order of magnitude larger than historical software budgets [00:04:47]. For example, the customer service software market is $35billion,whilethehumancostis$450 billion [00:04:58]. AI also expands into previously under-penetrated markets where traditional software models were too small or users lacked sophistication [00:05:22].

The AI Landscape: Layers of Value

The AI landscape can be viewed in three layers:

1. Model Layer: This includes LLMs and other models that power AI applications [00:06:08]. The value of these companies increasingly lies in the products built on top of them [00:06:52].

   • High Cost of Entry: Building state-of-the-art LLMs is prohibitively expensive, leading to few new entrants in this specific space [00:07:20].
   • Commoditization: Models are becoming cheaper, with inference and training costs dropping by approximately 10x per year [00:08:13]. The release of Deepseek demonstrated that “scale is not an enduring moat” [00:08:36], as portfolio companies could switch from Anthropic to Deepseek within days, achieving 80-90% cost reductions [00:09:18].
   • Building Moats: Model companies will likely build moats through distribution (e.g., OpenAI launching apps and agents) or specialization (e.g., robotics, biology, material sciences) [00:08:48].

2. Infrastructure Layer: These are the “picks and shovels” that bridge models and applications, enabling developers to build AI applications [00:06:19].

   • Slower Growth: This layer has been slower to develop than anticipated due to the rapid pace of change at the model layer and an initial focus on using powerful, brand-name models for use case discovery [00:10:54].
   • Emerging Opportunities: The emergence of agents is expected to create common patterns for web access and tool usage, opening new opportunities for infrastructure investments [00:11:37].

3. Application Layer: This layer includes horizontal and vertical SaaS solutions leveraging AI to deliver unique capabilities [00:06:28].

   • Disruption Opportunity: AI presents a similar disruption moment to the cloud wave, driven by a technological shift and a fundamental “business model change” where companies charge for “work rather than for a seat” [00:13:34].
   • Horizontal Applications: Opportunities exist to compete with tech giants like HubSpot, Salesforce, and Koopa by building AI-native solutions with a business model advantage [01:13:52]. The attack vector for startups is speed, as incumbents struggle with legacy systems and business models [00:14:30].
   • Vertical Markets: There’s been a “Cambrian explosion” of vertical AI SaaS companies (e.g., 500-600 companies in recent years), targeting markets traditionally underserved by software [00:15:00]. Key questions for assessing these opportunities include:
     • Is there an effective “wedge” into the market with viral end-user love and meaningful growth [00:16:09]?
     • How much more can the company do beyond replacing a single FTE to become a large standalone business [00:16:33]?
     • How much does quality matter in the chosen use case? In industries like healthcare, law, or finance, where quality is paramount, a “race to the bottom” on price is less likely [00:17:00].
   • Market Size: AI’s ability to unlock labor budgets makes traditionally smaller markets more attractive, potentially generating much larger Annual Contract Values (ACVs) by replacing human labor [00:17:23].

Challenges and Considerations for AI Startups

• Competitive Carnage: The rapid proliferation of vertical AI solutions will likely lead to significant attrition and consolidation, similar to the early internet e-commerce boom [00:17:59]. Many companies may experience initial growth but “top out as the markets become crowded” [00:18:13].
• Winner-Take-Most vs. Fragmentation: It’s unclear if AI markets will follow the SaaS model of clear market winners or remain highly fragmented due to value being created by underlying LLMs that many companies can leverage [00:20:41]. Velocity and rapid innovation are seen as crucial for startups to outmaneuver competitors and expand into adjacencies [00:21:24].
• Evaluating Founding Teams: When assessing founders, Redpoint typically favors those with “founder market fit” – collective experiences that provide a unique insight into long-term market problems – over pure AI expertise [00:23:31]. While technical acumen to understand model advancements is important, being a model developer isn’t essential [00:26:00]. Domain expertise is key for understanding end-user problems [00:26:21].
• First-Mover Advantage: Rapidly becoming “synonymous with a category” within 6-9 months allows companies to dominate customer conversations, attract partnerships, and secure capital [00:23:42].
• The “Moat”: The enduring competitive advantage in AI applications often lies not in proprietary data assets, but in “the thousand little things” like user experience (UX), product breadth, and overall user experience, much like traditional SaaS products [00:24:41].
• Vulnerability of Incumbents: While incumbents have distribution and customer relationships, their reliance on old databases, infrastructure, and workflows makes them vulnerable to AI-native startups that can fundamentally rethink and eliminate complex, legacy systems [00:31:51]. The more the workflow changes, the greater the opportunity for disruption [00:33:30].
• “Experimental Budget” vs. “Business Line Budget”: Many early AI products are funded by experimental budgets. It’s crucial to assess if a company can translate this into durable revenue from core business line budgets [00:27:58]. Strong user engagement and usage metrics are key indicators of a product’s real value [00:28:50].

The Hype Cycle and Valuation Challenges

AI company valuations are substantially higher, even for Series A rounds, with incrementally larger round sizes [00:35:30]. This is partly due to the belief that AI will access larger markets, including labor budgets, and the unprecedented speed at which some companies are growing [00:36:00].

• Leaner Operations: AI-native companies are expected to build products much more efficiently, potentially reaching hundreds of millions in revenue with only 20-40 employees and requiring less future capital [00:36:40]. This could lead to less dilution for early investors [00:37:10].
• False Positives and Immaturity: The rapid growth in AI can lead to “false positives” where companies spike quickly [00:37:25]. Many AI companies are reaching significant revenue benchmarks (e.g., $8 million ARR) very quickly (e.g., 8 months old, 10 employees) but lack the corporate maturity (systems, people, processes) of traditional SaaS businesses at that scale [00:39:48].
  • This implies that “revenue has been a lagging indicator” in traditional venture but might be a “misleading indicator” in AI [00:41:11]. A $50millionAISaaSbusinessisverydifferentfroma$50 million traditional SaaS business in terms of maturity [00:42:05].
• Investment Discipline: Given the high valuations, preemptions, and inherent market risks, Redpoint focuses on “tail opportunities” in large end markets where a massive potential exists if the product succeeds [00:37:50]. It requires “incredibly disciplined” investment to avoid overpaying for companies that may not sustain their early traction [00:39:37]. This has led to co-processing more deals between early and Omega funds and being circumspect about investments [00:40:48].

The firm emphasizes “first principle diligence” and focusing on “core fundamentals” rather than being swayed by hype or inflated numbers [00:42:55].