MC.67: Why are AI companies so stupid in pricing?

The first warning sign was Sam Altman's tweet about losing money on ChatGPT Pro users ($200/month). This candid admission revealed fundamental economic challenges in the AI business model that many suspected but few acknowledged publicly.

Then Deepseek R1 entered the arena, offering API calls 23x cheaper than OpenAI. This isn't mere creative destruction—it's creative annihilation that threatens to reshape the entire market's economic foundation.

As the market panics, incumbents are responding with increasingly puzzling strategies that appear more reactive than strategic:

Anthropic released Claude 3.7 at the same price as Claude 3.5, while keeping 3.5 at full price. While they're positioning as premium and need to maintain revenue, the messaging is confusing. Why use 3.5 if 3.7 costs the same? This contradiction suggests internal conflict between innovation and revenue preservation.

Perplexity increased prices by 2.5x with just 30 days' notice, while keeping low-tier API pricing unchanged. Their anchor pricing strategy seems disconnected from market realities and risks alienating early adopters who fueled their growth.

OpenAI takes the crown with GPT-4.5 priced 15-30x higher than GPT-4o (aptly dubbed "GucciAI" by one user). This premium positioning comes when independent benchmarks show most models—both proprietary and open-source—perform nearly identically on practical tasks. The question remains: what sustainable value justifies this premium?

The pricing volatility indicates both market recalibration and strategic uncertainty as companies seek equilibrium between growth and sustainable operations, especially after OpenAI's announcement that it will not be cashflow positive until 2029.

It could also reveals growing hostility toward application-layer companies (hello Manus!). While they can't prevent feature cloning, they can make it prohibitively expensive, suggesting subscription revenue might be more valuable than previously anticipated.

Meanwhile, Google's Gemini 2.0 is dominating performance rankings on OpenRouter. With their infrastructure advantages and diversified revenue streams, they may be playing a longer game—potentially undercutting the market while competitors focus on short-term revenue maximization.

And Google is just starting.

Looking ahead, these chaotic pricing strategies may have significant second-order effects: accelerating adoption of open-source alternatives, creating opportunities for middleware solutions that abstract away model selection, and potentially triggering consolidation among smaller players.

We still believe premium models have their place, and strategic positioning matters beyond token pricing. But Google, Deepseek and Qwen have demonstrated you can deliver excellent quality at competitive prices. The winners in this market may not be those with the most powerful models, but those who most effectively translate AI capabilities into consistent user value.

Let's hope market leaders refocus on serving customers rather than just positioning themselves against competitors.

Like this newsletter? Forward it to a friend and have them sign up here.

Until next Thursday 🎉