OpenAI Said to Spend $3.30 to Make $1

JAKARTA - OpenAI has long been known as the main face of the artificial intelligence or AI explosion. ChatGPT is widely used. Massive investment comes in. The name OpenAI is almost synonymous with the future of technology.

However, there is another side that is starting to be highlighted: the cost of running AI turns out to be very expensive.

In an analysis published on Substack quoted Thursday, June 4, independent writer Shanaka Anslem Perera assessed that OpenAI's biggest problem is no longer just technology. The problem lies in the ever-increasing operating costs.

According to Perera's calculations, OpenAI spends about US$3.30 to generate US$1 in revenue. Assuming a rate of Rp17,800 per US dollar, it means OpenAI spends around Rp58,740 to get Rp17,800.

Perera estimates that OpenAI burns around US$11.5 billion per quarter, or around Rp204.7 trillion. In a year, its value reaches US$46 billion, equivalent to around Rp818.8 trillion.

Meanwhile, OpenAI's operating income is estimated at around US$20 billion, or around Rp356 trillion.

With that count, the outgoing costs are much larger than the incoming money.

Perera assessed that OpenAI's problem is not just "burning money" to pursue growth. He said there were deeper issues, namely the cost structure.

Every time a user uses ChatGPT, OpenAI has to bear the cost of computing. Computing is the computing power of a computer to run AI. The greater the use, the greater the need for chips, servers, electricity, and data centers.

Here, scale can be a double-edged sword. The more users, the greater the bill that must be paid if the cost per service is not efficient enough.

In short, smart AI is not necessarily winning. In the end, companies still have to be able to calculate profits and losses.

Shanaka Anslem Perera is an independent writer at Substack who covers markets, technology, and investment. The figures and conclusions in his writing are Perera's personal analysis, not official statements from OpenAI or Microsoft.

Even so, his analysis opens up an important question. In the AI industry, model capabilities do determine. But operating costs can be an equally important factor in determining who survives.