"History doesn't repeat… but it rhymes." — Mark Twain
◉ THE PRESENT
Intel closed Thursday at $69.28 and then printed a first-quarter earnings report that stopped the scrolling on every trading desk in Manhattan. Revenue came in at $13.58 billion against a Street estimate of $12.42 billion. Earnings per share hit 29 cents against a penny. Data Center and AI revenue rose 22% year over year to $5.1 billion, with segment operating margin nearly doubling. The stock jumped 19% in extended trading, adding roughly $64 billion in market value to a company that had been priced for a funeral eighteen months ago.
Intel Q1 Rev: $13.58B (vs $12.42B est) | Stock: +19% after hours | YTD under Lip-Bu Tan: +78%
Wall Street has seen this movie before. A written-off American chip icon. A new CEO on his one-year anniversary. A surprise quarter that turns the narrative inside out. The last time the script ran this clean was the spring of 1994.
◉ THE ECHO — APRIL 19, 1994
The dinosaur learns to run.
Lou Gerstner had been running IBM for exactly one year. When he took the job on April 1, 1993, the obituaries were already written. The stock had just traded through $10 a share, its lowest price in two decades, and the board had fired his predecessor in the kind of public mess that only happened to companies no one believed in anymore. Business Week had run a cover called "The Fall of IBM." Analysts were demanding a breakup — sell the mainframe business, sell services, sell the PC unit, sell the chip line, and let the pieces be worth more than the whole.
Gerstner didn't break it up. He told a room full of reporters that the last thing IBM needed was a vision. It needed a quarter. Customers had stopped returning his salespeople's calls. Engineers were walking out the door for Sun and Oracle. And the mainframe — the machine every consultant in America had been calling obsolete since 1988 — was quietly running the back office of every bank, airline, and insurance company that mattered, because none of them could figure out how to migrate off it without blowing up.
He spent his first twelve months cutting costs and re-signing customers one executive at a time. He flew to Tokyo, to Frankfurt, to Charlotte. He listened more than he talked. He stopped the bleeding in Q4 1993 with the first clean quarterly profit IBM had posted in more than a year. The Street noted it and moved on. Too early to call, the analysts said. One quarter doesn't make a turnaround.
Then Q1 1994 came in. Mainframe orders were running ahead of what IBM could ship. Services revenue was compounding. The company had made real money, not cost-cut money, and the stock that had bottomed below $11 the summer before was now trading near $28. By the end of 1995 it was above $50. By 1999 it was above $180. The analysts who had demanded a breakup in 1993 spent the next six years explaining why they had missed the single best turnaround of the decade.
The lesson buried in the chart wasn't about one quarter. It was about the shape of the trade. IBM didn't die. It was quietly selling picks and shovels to the gold rush while the headlines stayed negative. That's the part the Street missed in 1993. And that's the part the Street almost missed again in 2025.
◉ THE RHYME — WHAT'S IDENTICAL
Same script, thirty-two years apart. An American computing icon left for dead. A new CEO handed a company that couldn't sell its own future. And an enterprise customer base that kept quietly signing checks while the covers stayed negative.
◉ THE DIVERGENCE — WHAT'S DIFFERENT THIS TIME
The shape matches. The math does not.
The fab bill is bigger than anything Gerstner ever signed. IBM in 1994 was asset-light by the standards of heavy industry. Intel is spending roughly $25 billion a year on fabs and has an open question about whether the 18A process actually delivers at scale against TSMC. Gerstner inherited a services lever he could pull immediately. Tan inherited a foundry he may not be able to save without another decade of capex.
The competition has a network effect this time. Nvidia does not just sell chips. It sells CUDA, a developer ecosystem fifteen years deep with millions of engineers trained on it. IBM's rivals in 1994 — Sun, Oracle, Compaq — had good products but no moat at the programmer level. Nvidia has one, and it cannot be out-engineered with a single strong product cycle.
The balance sheet is thinner. IBM in early 1994 still carried an enormous installed base of multi-year maintenance contracts that paid like an annuity. Intel's cash position is tighter, which is why the U.S. government took an $8.9 billion equity stake last August to backstop the foundry build. That is rescue money, not strength money, and the market should remember the difference.
The cycle moves in quarters, not years. IBM had the luxury of a slow enterprise refresh window — mainframes got replaced every five to seven years and customers didn't shop around. AI capex is being reallocated in real time by three or four hyperscalers who can switch vendors inside a twelve-month roadmap. If Intel misses once, the window closes faster than it did in 1994.
◉ THE RECKONING — WHAT HAPPENS NEXT
The 1994 path is the tell. IBM's stock ran from around $28 in April 1994 to $74 by the end of 1995, hit $100 in 1996, and climbed past $180 by 1999. Over six years, an "obsolete" American technology company returned roughly five times the market. The money was not made by the traders who caught the absolute bottom in August 1993. It was made by the investors who bought the confirmation — the boring second good quarter — and held through the doubt.
Turnarounds rhyme in thirds. A stock doubles when fear leaves. It doubles again when the first clean quarter lands. It doubles a third time only when the market gets a second clean quarter that proves the first wasn't a one-off. Between the first pop and the second confirmation is always a long, flat, nervous summer. That summer is the opportunity. It is also the graveyard for turnarounds that couldn't follow through.
Intel is right in the middle of that window now. Q1 2026 was the shock. Q2 guidance — $13.8 to $14.8 billion in revenue, 20 cents in EPS against a 9-cent consensus — looks clean. The real test is Q3, which will report in late October. If hyperscaler orders firm up into multi-year Xeon contracts, and if the Google cloud deal and the Austin Terafab project with SpaceX and xAI start showing up as committed backlog instead of press releases, the 1994 script is running and the stock has another leg in front of it. If Q3 softens, this was a short cover dressed up in a turnaround uniform.
The smart money in 1994 did not chase the earnings pop. They watched customer behavior. They tracked which banks were signing System/390 renewals for five years versus one. They looked at service contract durations. The equivalent read for Intel is whether Google, Microsoft, and Amazon commit Xeon across the full inference refresh cycle, or sign quarter-by-quarter stopgaps while they wait to see what AMD's next server chip looks like. Contract length is the signal. Revenue beats are the noise.
One more thing worth remembering. In the fall of 1994, Fortune ran a cover story asking whether Gerstner was too cautious to save IBM. The stock had doubled off its lows and most of the Street was still calling the rebound a trap. That cover is framed in a lot of trading floors now.
Turnarounds pay in three acts. Intel has finished the second. The third leg does not depend on April's earnings, which are already in the tape. It depends on October's — and on whether the largest buyers of AI infrastructure sign multi-year or month-to-month.
◉ TOMORROW’S WATCH
Keep one eye on commercial real estate loan losses quietly piling up on regional bank balance sheets — Texas ratios at mid-sized lenders are creeping higher than they have since 2009. The last time a slow drip in regional bank loan books turned into a run, it was the spring of 1990, and the name on the tombstone was Bank of New England.
