This is the web version of Data Sheet, Fortune’s daily newsletter on the top tech news. To get it delivered daily to your in-box, sign up here.
It was, as Yogi Berra used to say, deja vu all over again when Intel released its earnings last night. And not in a good way.
At first glance, the news looked good, even very good. Despite the coronavirus pandemic and low expectations from Wall Street, second quarter revenue jumped 20% to almost $20 billion and profits did even better, rising 22% to over $5 billion.
But hiding in plain sight at the top of the release, at the end of the fourth bullet point, was quite the line: “7nm product transition delayed versus prior expectations.”
That short sentence sent fund managers leaping to sell their shares, analysts pulling the rip cord on their ratings, execs at PC makers grabbing for their Tums and Rolaids, and likely more than a few folks at Advanced Micro Devices popping champagne corks. (This morning, Intel’s stock is down 14% in premarket trading and AMD’s is up 8%.)
Why such a reaction? Because it immediately brought back memories of another July earnings release. Back in July 2015, then-CEO Brian Krzanich announced that Intel’s move to 10 nanometers was running behind schedule and new products slated for 2016 would arrive in 2017, a slight delay of only six to nine months, he said. Flash forward to 2020 and we’re still awaiting some of those products and Krzanich is long gone, replaced by his former CFO, Bob Swan.
Swan promised such a short product delay on Thursday night. But as he was named CEO in part to clean up the mess of Krzanich’s delayed 10-nanometer, the news of today’s delay did not inspire confidence. And Swan upended 50 years of Intel strategy when he revealed that the company’s contingency plan, in case the delays worsened, was to outsource chip manufacturing to rivals, an unprecedented move with unknown impact on Intel’s finances.
The announcement also sheds light on other recent events, like Apple’s decision to dump Intel chips from its Mac computers and, possibly, the departure of star chip designer Jim Keller, who lasted just two years.
The success of the entire tech industry has relied on chipmakers’ ability to cram more and more transistors on a silicon wafer. First highlighted by Intel co-founder Gordon Moore in 1965, Moore’s Law says the industry should be able to fit twice as many transistors on the same size chip every two years or so. It’s what has enabled computers to go from the size of a house to tinier than a smartwatch and what has unleashed a thousand new industries.
But Moore’s Law has been slowing down for a while as companies like Intel, Samsung, and Taiwan Semiconductor have pushed the transistors to smaller and smaller scales. The 7-nanometer scale Intel was aiming for is about 1/20th the width of a human hair. Apple’s A13 Bionic processor in the iPhone 11, made at 7-nanometer scale by Taiwan Semiconductor, fits 8.5 billion transistors on a chip smaller than a dime.
Even if Intel can’t be the best chip manufacturer, it could still be the best chip designer. That outsourcing strategy, and some big decisions Keller pushed to compartmentalize chip design, may ultimately save Intel’s market share. Keller likes to quote Steve Jobs, whom he worked for a few years at Apple, and one quip he has repeated to me is this: “Why screw up two things when you can only get one thing right?”
Intel, its investors, and the rest of the industry are probably wondering the same thing right about now.