OpenAI Chief Executive Officer Sam Altman is seeking to raise trillions of dollars in funding for a new chip venture, the Wall Street Journal reported on Thursday day evening.

Rumors of the project first surfaced in late 2023. Last month, the Journal cited sources as saying Altman hopes to raise between $8 billion and $10 billion from investors. Today’s report suggests the executive has a far more ambitious goal: Altman is believed to be seeking as much as $5 trillion to $7 trillion for the chip venture.

The AI executive is said to have held fundraising talks with several potential investors including the government of the United Arab Emirates. According to the report, SoftBank Group Corp. CEO Masayoshi Son may participate as well. It’s believed that the venture may also raise debt financing to support its plans.

Altman reportedly intends to spend the capital on a network of chip plants, or fabs, that will be run by established semiconductor makers. Taiwan Semiconductor Manufacturing Co. has been floated as a potential technology partner. It’s unclear how many fabs the chip venture intends to build, in what time frame or what manufacturing processes they will use.

Today’s report did specify that the facilities would be capable of making, among other products, AI accelerators. OpenAI is expected to become a “significant” customer of those chips. Fabs capable of making the most advanced AI accelerators, such as the two facilities TSMC is currently building in Arizona, can cost upwards of $20 billion.

That price tag will have likely increased considerably by the time Altman’s venture starts building its plants. The semiconductor sector is currently adopting a new type of extreme ultraviolet machine, a system used to etch transistors into silicon wafers. The new system is expected to cost in excess of $300 million, about 50% more than the most advanced lithography machines on the market today.

Given the sheer amount of funding Altman reportedly hopes to raise, it’s possible the planned chip venture will not limit its focus to advanced processors alone. Of the more than 500 active fabs listed by Wikipedia, only a small fraction are dedicated to building cutting-edge processors. The rest make less advanced semiconductor products such as sensors.

But even if Altman’s venture moves into multiple segments of the chip market, realizing a return on investment could still take up to decades. On the high end of the range, the $5 trillion to $7 trillion Altman is said to be seeking is more than ten times higher than the semiconductor sector’s 2023 revenues, which stood at $527 billion. That figure is expected to reach $1 trillion by 2030.

Operating a fab network would require the chip venture to build not only processor plants but also a significant amount of supporting infrastructure.

Many advanced chips comprise multiple semiconductor modules that are fabricated separately and then linked together into a single product. The hardware necessary to link together those chiplets, which is known as packaging, is made at specialized, multibillion-dollar factories. Depending on the number of fabs it plans to build, Altman’s venture may have to invest in a significant amount of packaging manufacturing capacity to support its operations.

Fabs also require an extensive supply chain that can provide them with materials and chipmaking equipment. To accelerate its construction roadmap, Altman’s venture could potentially make investments in key chip industry suppliers to help them expand their manufacturing capacity. That’s an approach Apple Inc. has adopted in recent years to support the production of iPhone parts. 

Photo: Unsplash