Detailed Analysis
Anthropic's reported $30 billion funding round at a $900 billion valuation — surpassing OpenAI's paper valuation — serves as the backdrop for a pointed empirical challenge to the assumptions underlying that capital raise. The author, operating a database of 6,000+ AI tools across 28 categories with 22,000+ real SMB user reviews tagged as WORKED, MIXED, or FAILED, used Claude to interrogate where AI actually delivers measurable returns for small businesses. The findings reveal a striking gap between investor narrative and ground-level performance: the five highest-performing categories for SMBs are Sales Management (74% WORKED), Scheduling & Calendar (67%), SEO & AI Visibility (64%), Customer Retention (62%), and Meeting Notes & Transcription (61%) — tools largely predating the generative AI wave and associated with companies like Pipedrive, Calendly, Ahrefs, and Otter.ai.
The bottom of the performance database is occupied by precisely the categories commanding the largest valuations and venture capital attention: CRM & Customer Data (37% WORKED, 33% FAILED), Tax & Compliance (37% WORKED), Customer Support (42%), Marketing Campaigns (42%), and Email Marketing (42%). The author notes that AI CRM in particular maintains the highest failure rate in the entire dataset, a pattern they describe as "THE CRM GRAVEYARD" — a persistent underperformance that has not shifted despite significant capital investment and product iteration. The $900 billion valuation thesis, the article argues, is structurally built on categories where real-world SMB users report the worst outcomes.
The author frames this as an instance of what they call "THE BORING PREMIUM" — a systematic pattern in the database where narrow, focused AI tools performing a single function consistently outperform broad-platform AI tools promising comprehensive workflow replacement. This pattern, the piece argues, has moved from being a niche observation to the dominant signal across the entire dataset. The practical implication drawn is that SMBs spending $20–$50 per month on category-leading tools in scheduling, sales pipeline, or meeting transcription are generating more verified return than the AI agent and AI worker platforms absorbing billions in institutional capital.
The analysis carries broader significance for how investors and enterprises are evaluating AI ROI. The tension between what gets funded and what gets used effectively is not unique to AI — the article explicitly analogizes it to television network upfronts, where flashy programming captures orders while quieter, consistent content earns its audience — but the magnitude of the valuation gap makes the divergence particularly consequential in 2026. Anthropic's own Claude is deployed here as an analytical instrument to surface this critique, which adds a layer of irony: the model trained and funded by a company at the center of the frontier AI valuation boom is being used to produce evidence that the boom's commercial thesis has a significant SMB validation problem.
The dataset's rapid shift — from only one category above 50% WORKED in early May 2026 to 18 categories above that threshold by late May — does suggest meaningful acceleration in AI tool quality across the board. However, the relative positioning of categories has not changed: utility-focused, single-purpose tools retain their lead, and the flagship AI platform categories remain in the bottom tier. For the companies and funds whose returns depend on SMBs adopting AI CRM, AI customer support, and AI marketing automation at scale, this data represents a durable structural challenge rather than a temporary adoption lag.
Read original article →