New calculations from semiconductor and AI research firm SemiAnalysis suggest that individual subscription plans offered by OpenAI and Anthropic may provide far more computing value than their monthly prices indicate, raising fresh questions about how long leading AI companies can continue subsidizing heavy consumer usage while expanding enterprise-focused businesses.
Subscription Value Far Exceeds Monthly Price
SemiAnalysis found that a $200 (₹19,100) monthly subscription plan can deliver substantially more token value than previously assumed when measured against API pricing. While such plans were widely believed to top out at around $2,000 (₹1.91 lakh) worth of monthly token usage, the firm calculated that the actual value could be closer to $14,000 for ChatGPT and $8,000 (₹13.37 lakh) for Claude.
FCRF’s Flagship Cyber Law Certification Returns With a New Four-Week Cohort
The gap is also visible in lower-cost subscription tiers. For plans priced around $20 (₹1,910) a month, the estimated token value available to users was calculated at about $700 (₹66,850) for ChatGPT and $400 (₹38,200) for Claude. The figures underline the extent to which individual users may be receiving heavily subsidized access compared with the cost of similar usage through enterprise or API channels.
Heavy Use Can Push Margins Deeply Negative
SemiAnalysis said the profitability of subscription plans depends heavily on average utilization. It noted that if both companies are assumed to have 75% API gross margins, a heavy user on Anthropic’s $200 monthly Claude Max 20x plan could generate a gross margin of minus 900% for the company.
Even moderate usage can strain the economics of such plans. According to the analysis, 20% utilization could push margins to minus 100%, while very low usage of around 1% could still allow the same plan to deliver a positive gross margin of about 90%.
The firm said this makes consumer subscriptions less attractive than API usage from a margin perspective. However, it also noted that openly reducing subscription limits could trigger strong public backlash, making it more likely that AI companies may restrict access to newer features or models rather than sharply weakening existing plans.
Enterprise Growth May Sustain Subsidies
The analysis does not necessarily suggest that OpenAI and Anthropic will be unable to continue supporting consumer subscriptions. The broader business model may still allow subsidized individual plans if revenue growth from enterprise users and API partners remains strong.
The Wall Street Journal reported in May that Anthropic had seen rapid growth that could help the company move toward profitability, supported by a high-margin, enterprise-first model and strong revenue from API partnerships rather than dependence on consumer subscriptions.
Subsidized subscriptions also serve a strategic purpose. By keeping individual plans attractive, AI companies can expand their user base, encourage product adoption and gather usage patterns that may help improve models and attract enterprise customers. SemiAnalysis suggested that rapidly falling AI costs could eventually make it possible to profitably offer advanced models at lower monthly prices, although access to the newest tools may increasingly be reserved for higher-value customers.
