Intel: Too big to turn, too vital to fail

Intel’s disastrous second-quarter report this month has put the struggling chip maker in a whole new light—and not a good one. Falling sales in key markets and rising costs for its ambitious manufacturing turnaround have forced the company to take more drastic steps to conserve cash. Those include laying off 15% of the workforce, cutting the capital expenditures used to build and equip production facilities and suspending the dividend that Intel has paid since 1992.

The latest reboot has investors unplugging en masse. Intel lost a little more than a quarter of its market value the day after its Aug. 1 earnings report, and the stock has shed another 8% since—worse than the drubbing seen by most other chip stocks in this past week’s global selloff. Intel shares have now lost about 68% since Chief Executive Pat Gelsinger first articulated his turnaround plan after rejoining the company in early 2021. The S&P 500 is up 39% over that time.

Intel is also now trading below the company’s book value for the first time since at least 1981, which is as far back as data from FactSet goes. This means investors are now valuing one of the world’s largest chip manufacturers for less than the value of its facilities and other assets on its balance sheet.

But those facilities are also the key to Intel’s staying power. The pandemic’s disruptions and other forms of global instability over the past few years have awakened political leaders on both sides of the aisle to the importance of shoring up domestic production of such a key component of modern life. And chip fabrication plants can’t be spun up overnight—or on the cheap. Modern fabs require years to build and equip and now cost around $20 billion.

That is why the U.S. government is kicking in. The Chips Act passed in 2022 outlined $39 billion in direct grants to chip makers to help with the cost of building new facilities. Intel has been the largest beneficiary, landing $8.5 billion of that pot that is helping to fund the building of new fabs in Arizona and Ohio. Intel is already the biggest fish in the American pond; the company’s current fabs account for about 41% of the country’s 300 mm wafer production capacity—the type of chip production most commonly used in key market segments—according to market research firm TechInsights.

But a big part of Intel’s current problem is that the chips it makes for itself aren’t selling as well as they used to. The company’s once-booming data center business, in particular, has been hit hard as it has lost share in server CPU chips to Advanced Micro Devices. It also has seen a sharp shift in data center budgets to Nvidia’s GPU accelerators, which are key to powering generative artificial-intelligence services. Intel’s data center revenue is projected to hit $12.6 billion this year, less than half its peak just four years ago, according to consensus estimates from Visible Alpha.

The rapid shift to spending on AI—and Nvidia in particular—wasn’t something contemplated when Intel laid out an ambitious and expensive plan to catch up its manufacturing processes to those of Taiwanese rival TSMC three years ago. “The issue, in our view, is that the server-centric data center Intel built their fabs to serve no longer exists, replaced by AI spending that Intel missed," wrote Chris Caso of Wolfe Research in a recent note to clients.

That has also contributed to Intel’s fabs being underutilized—an expensive problem for chip manufacturers who have high fixed costs. Underutilization charges contributed to Intel’s adjusted gross margin coming in at a shocking 38.7% in the second quarter, 5 percentage points lower than Wall Street expected.

Wall Street is divided on where Intel should go from here. Some analysts think it needs to focus on regaining product leadership, even if that comes at the expense of the foundry business the company is building to serve other chip designers. Others think the company needs to focus on getting more major customers for the foundry side, since Intel’s chances of getting competitive in key markets such as data center GPUs look slim.

None of those paths will be quick, though. And loss of the dividend leaves Intel investors with little to hang on to; it is now one of only three Dow components not paying one. But Intel’s large role in what is now deemed an industry vital to national security also provides a floor of sorts. In his report, Caso noted that “given the sensitivity toward domestic U.S. semiconductor production, we doubt the U.S. government would allow for Intel’s problems to become terminal."

Uncle Sam might be the biggest bull in Intel’s pen for a long while.