Beyond financial resources, Sam Altman’s mission to transform the chip business needs more
In order to transform the global semiconductor business, OpenAI Chief Executive Sam Altman intends to invest enormous sums of money into a problem that is considerably more difficult than money.
Chip manufacturing requires a significant amount of capital. In addition, it is among the world’s most intricately complicated industries, and because of its history of severe cyclical fluctuations, businesses have been cautious to pursue drastic expansion.
The most sophisticated semiconductor manufacturers in the world took decades to ascend to their current positions. During one of the industry’s well-known downturns, such as the early and middle 2010s, a few semiconductor businesses failed. Some gave up on creating cutting-edge chips along the route because to the exorbitant expense and high failure rate.
These days, just three companies in the world—Taiwan Semiconductor Manufacturing Co., Samsung Electronics, and Intel—are able to produce the most advanced chips in huge quantities, including the processors that drive AI systems.
Altman has talked to chip manufacturers about partnering with them to use billions of dollars to construct and run new factories, in addition to energy and other AI infrastructure investments. Many of the biggest semiconductor companies in the world, like Nvidia, develop their circuits but contract out their manufacturing to firms like TSMC.
A state-of-the-art chip factory normally takes at least $10 billion to build. However, even with that, Altman is talking about an excessive scale: According to Bernstein Research analyst Stacy Rasgon, the industry has historically spent somewhat over $1 trillion on semiconductor fabrication equipment.
But to flourish in one of the most intricate manufacturing industries on the planet, money isn’t the only requirement.
Industry executives say there are doubts about acquiring the machinery to staff the plants, finding the experts to run a flurry of new factories, and gaining enough orders to support them. Despite decades of capital infusion from the Chinese government, the development of the semiconductor sector was impeded by a few other problems.
Jimmy Goodrich, a senior adviser to Rand Corp. and industry expert in semiconductors, stated that “the semiconductor industry doesn’t lack funding.” “China is a prime example—they have put over $150 billion at their sector, with no success. The primary obstacle lies in the extreme difficulty of this technology.”
Building a lot of new chip manufacturers wouldn’t guarantee that Altman’s short-term issue—the lack of AI chips required to create products like OpenAI’s ChatGPT—would be resolved. Following the imprinting of circuits on silicon, the manufacturing process known as packaging has proven to be the largest source of bottlenecks in the production of Nvidia’s AI processors.
Altman has also voiced concerns about the price of Nvidia’s processors, which Srini Pajjuri, a Raymond James analyst, said more chip plants might not be able to address.
“More competition against Nvidia is necessary for AI chip prices to decrease,” he stated.
There are currently more factories being built. Tens of billions of dollars are being spent on them by all of the major chip manufacturers, which are timing their expansions in anticipation of worldwide chip sales exceeding $1 trillion by 2030. From roughly 10 million silicon wafers annually in 2016 to 16 million last year, TSMC has increased production.
According to chip insiders, if Altman’s proposal succeeds, it will probably generate oversupply in the market, which would drive down prices and force businesses to operate facilities much below their capacity, which would be a financial death knell for the industry due to its enormous fixed costs.
According to Bernstein’s Rasgon, the sector grows in spurts and as they invest now, online demand declines. It always occurs because, more often than not, the desire you are creating turns out to be unfounded in the first place. Thus, the never-ending loop.”
Governments all around the world are contributing to the industry’s current boom because they understand how crucial chips are to their ability to dominate in the fields of technology, economy, and military might. With the passing of the Chips Act two years ago, the United States is investing $39 billion in subsidies into new plants in an effort to revive a sector that had fled to Asia in recent decades.
The amount of cash Altman is considering would be far smaller than the incentives offered in the United States and other countries, such as China, Taiwan, South Korea, and Japan.
According to those familiar with the negotiations, Altman has met in recent weeks with U.S. officials, including Commerce Secretary Gina Raimondo, to discuss his goals. Raimondo is in charge of distributing grants made possible under the Chips Act.
Altman is scheduled to speak at an Intel contract chip-making business event later this month, following a flurry of recent engagements with chip firms. It is anticipated that Chips Act grant funds, which will be distributed in the upcoming weeks, would go largely to Intel.
How Altman would recruit personnel to staff dozens of additional plants is a mystery. By the end of the decade, 115,000 new employment are predicted by the Semiconductor Industry Association, with 58% of those positions potentially going unfulfilled. Furthermore, there’s no guarantee he could find adequate production equipment in a timely manner. Certain chip-making machinery has two-year lead times.
Industry sources believe that making chips presents different obstacles than those that Altman faced in his previous enterprises, all of which involved computers and software.
“With software, anything is possible—the only things that stand in the way are money and coding,” Goodrich stated. In the realm of hard technology, on the other hand, one must truly contend with the rules of physics. It is difficult to consider engineering difficulties and the real world at the same time.
