By Dan Gallagher

It's good to have Nvidia's deep pockets these days. If you don't, getting memory and other necessary components for AI computing is going to keep getting harder.

When the AI chip giant filed its last quarterly update three months ago, the company reported $95.2 billion in purchase commitments with its supply vendors. That dwarfed the $11.4 billion in commitments related to Nvidia's many investments, and represented an 89% jump in just three months' time.

The reason isn't mysterious: key components like memory are surging in price and in short supply. Nvidia CFO Colette Kress said at the time that the company has "strategically secured inventory and capacity to meet demand beyond the next several quarters."

Purchase commitments are an obscure-enough metric that most analysts don't make forecasts for them. So, unlike revenue, earnings and free cash flow, there isn't a consensus on where purchase commitments will land when Nvidia posts its fiscal first-quarter report next week.

But memory costs alone have jumped even more since the last report, and even staying static at $95 billion is no small matter. That is just over half of the company's projected free cash flow for the current fiscal year, during which Nvidia expects the bulk of its purchase commitments will be paid out.

Nvidia isn't hurting for money, as evidenced by the world's largest companies continuing to empty their own coffers on AI infrastructure. But the makers of that infrastructure also have to live in a world of scarcity.

Memory chips, optical components and advanced manufacturing lines at TSMC are all in great demand. And all the money in the world can't grow the supply of those products and services quickly. A memory-chip factory alone takes years to build and get up and running.

Nvidia isn't the only one trying to lock down the necessary resources. Broadcom told investors during its last report in March that it has "secured the supply chain required" to hit $100 billion in AI chip revenue next year. That is a little over triple the current annual revenue of that business segment.

And Nvidia's smaller, direct rival AMD disclosed a little over $21 billion in purchase commitments through next year in its first-quarter filing last week. That was more than double what the company reported three months ago in commitments for that time period.

Using big pocketbooks to lay claim to the supply chain is all well and good for those who can afford it. But there are some who can't, raising questions about fairness and competition in a period of shortage.

Cerebras, for example, a maker of mega-sized chips that is set to go public later this week, boosted the price range of the offering by nearly 30% on Monday and plans to raise about $4.8 billion.

That is a good sign of demand for the offering. But even with that bounty, Cerebras will still be deep in a long line of bigger players who need components like memory and the manufacturing services of TSMC.

The company's IPO filing includes a lengthy warning that it procures nearly all of its necessary services and components on a purchase-order basis without long-term agreements.

These days, the AI chip business is about the art of managing scarcity.

This is an edition of the WSJ AI & Business newsletter, a weekly digest to help you make sense of AI's impact on business with news, insights and data from our global team of technology journalists. If you're not subscribed, sign up here.

OpenAI's Employees Are Cashing Out

Some of OpenAI's longtime employees got big payouts -- up to $30 million each -- after the company allowed them to sell shares to investors in an October financing round. The windfalls were unusual for Silicon Valley, where employees usually can only convert shares into cash after a company goes public or is sold. More money is likely to flow to employees of hot AI startups soon as OpenAI and Anthropic prepare for initial public offerings as soon as this year.

The Number

The unemployment rate among IT workers in March, according to the U.S. Bureau of Labor Statistics. The rate ticked up during the month as large tech companies including Meta Platforms announced large layoffs -- but tech-employee unemployment remains lower than its 7% peak around mid-2025.

What the Humans Are Saying

AI in Charts

The appetite for chips hasn't been this strong in a while. Tech giants are racing to get memory chips and central processors, in addition to the AI processing chips that have been the workhorses of the boom so far.

Taiwan Semiconductor Manufacturing Co. is in prime position to benefit in this setup. The company, the largest contract chip maker in the world by revenue, is enjoying beefy gross profit margins, with sales rising faster than expenses. Margins reached more than 66% in its most recent quarter -- a huge number for a chip manufacturer.

Margins are likely to compress in the coming few quarters because of the cost of introducing new cutting-edge chip-making in its factories. But its long-term outlook for revenue and profit growth looks bright. It has a diverse portfolio of customers, including Apple and Nvidia but also a big basket of other chip designers. And it is expanding geographically in the U.S. and Japan to help mitigate against its most significant risk: disruption caused by a conflict with China.

AI in the Wild

The Trump administration's focus on border security is creating a miniboom in AI-driven surveillance. Companies are marketing cameras that try to identify people carrying weapons and drones that autonomously monitor border threats, among many other gizmos. And U.S. agencies that handle border security, principally Immigration and Customs Enforcement and Customs and Border Protection, have big budgets.

Other Highlights From the Week in AI

   -- Testimony continued in the court battle between Elon Musk and OpenAI, 
      with Microsoft chief Satya Nadella and former OpenAI board member Shivon 
      Zilis taking the stand. 
 
   -- AI chip designer Cerebras raised its IPO price and is now expected to 
      seek $4.8 billion when it lists in the coming days. 
 
   -- Nvidia agreed to invest $500 million in Corning to help expand 
      fiber-optic manufacturing, a move that sent Corning's shares up sharply. 
 
   -- SpaceX is massively raising spending ahead of its IPO after it bought 
      Elon Musk's xAI in February. 
 
   -- Elon Musk's Grok is losing ground in the AI race, as adoption by business 
      and consumer users has slowed. 

About Us

WSJ AI & Business is a weekly look at AI's transformation of the business world. This newsletter was curated and edited by Dan Gallagher and Asa Fitch. Reach them at dan.gallagher@wsj.com and asa.fitch@wsj.com (if you're reading this in your inbox, you can just hit reply). Got a tip for us? Here's how to submit.

(END) Dow Jones Newswires

May 12, 2026 12:00 ET (16:00 GMT)

Copyright (c) 2026 Dow Jones & Company, Inc.