OpenAI expected to lose $5 billion: Computing costs account for 80% of operating expenses

According to reports, OpenAI may face losses of up to $5 billion this year, posing a significant challenge for the company currently valued at $80 billion.

OpenAI's total operating costs this year could reach $8.5 billion, with inference and training costs estimated at $4 billion and $3 billion respectively. However, OpenAI's annual revenue is projected to be between $3.5-4.5 billion, far below operating costs.

The company's costs are divided into:

1. Inference costs: Nearly $4 billion spent on renting Microsoft's server clusters to support ChatGPT and its underlying models.

2. Training costs: May surge to $3 billion this year, including data fees.

3. Labor costs: With about 1,500 employees, labor costs could reach $1.5 billion this year.

To address these challenges, OpenAI may need to:

1. Raise more funding in the next 12 months
2. Reduce inference costs through new model-building technologies
3. Expand revenue streams, including:
   • Charging developers for access to GPT models
   • Launching new products like a search engine (possibly called SearchGPT or Sonic)
   • Potentially introducing advertising

The success of GPT-5 could be a turning point if its performance significantly surpasses GPT-4.

A positive sign is that OpenAI's revenue growth is outpacing cost increases, with monthly revenue reaching $283 million, a 3.5-fold increase from a year ago.