ai-costs cloud-computing developer-tools code-generation

Why current LLM costs are not sustainable

aditya.patadia.org

June 26, 2026

5 min read

🔥🔥🔥🔥🔥

52/100

Summary

AI implementation is leading to significant cost overruns for companies, with Uber exhausting its annual AI budget in just four months. Despite the financial challenges, AI continues to simplify programming tasks and enhance productivity in areas such as data interpretation, presentation design, and app development.

Key Takeaways

Companies like Uber have rapidly exhausted their AI budgets, with Uber spending its entire annual budget in just four months.
GPT 5.5 is currently the most expensive AI model, costing $5 per million input tokens and $30 per million output tokens.
The emergence of open-weight models, such as GLM-5.2, offers significantly lower costs compared to frontier models, with GLM-5.2 costing one-tenth of GPT 5.5.
Specialized AI chips, like TPUs, can reduce inference costs by 30-70% compared to traditional GPUs, contributing to a potential decrease in AI model prices.

Read original article

View story →

Coverage & discussions

Hacker News · aditya.patadia.org

Why current LLM costs are not sustainable ↗

Community discussions

Hacker News discussion ↗

Historical count: 5 · 5 saved in sample · Historical sample

Analysis ready

Community Sentiment

Mixed

Positives

Switching to alternative models like DeepSeek Flash can drastically reduce AI spending, potentially making advanced AI more accessible to a wider audience.
Self-hosting options for LLMs could ensure privacy and reduce reliance on external services, which is increasingly important in today's data-sensitive environment.
The cost savings from subscription models compared to pay-per-token systems highlight the need for more affordable AI usage options, promoting broader adoption.

Concerns

Current LLM costs are unsustainable, and without significant cost optimization, many users may be forced to abandon or limit their AI usage.
The perception of heavily subsidized LLM subscriptions may lead to overuse, which could inflate costs and create a false sense of affordability.
Concerns about the long-term viability of subscription models suggest that only wealthier entities may benefit from advanced AI, leaving others behind.