Article 3: The Engineer's Crusade: A War Against a Ticking Clock
(By Michael Apemah, The Fab Analyst)
In our last piece, we explored the geopolitical forces that brokered a peace between Intel and NVIDIA. But that peace was only necessary because the kingdom of Intel was in turmoil, a state that prompted the board to make a dramatic move in 2021: they brought back the prodigal son, Pat Gelsinger.
Pat’s return wasn’t just a CEO change; it was a cultural revolution. He wasn’t a finance guy like his predecessor; he was a fire-breathing engineer who had helped design the legendary 80486 processor. He came back with a singular mission: to restore engineering greatness to Intel and put the company back on the throne of process technology.
His plan had a name: IDM 2.0. And it was powered by a mantra that every employee soon knew by heart: a “torrid pace.”
“Five Nodes in Four Years”: A Declaration of War
For those of us inside the company, the Gelsinger era began like a lightning strike. The slow, cautious, and often bureaucratic culture of the previous years was over. Pat declared war on mediocrity. He printed t-shirts. He demanded speed. His rallying cry of “Five nodes in four years” (Intel 7, 4, 3, 20A, and 18A) was an audacious, almost impossible goal. It was a public declaration that Intel was done making excuses and was going to out-innovate the world.
The ultimate bet, the company’s entire future, was placed on a single node: 18A. This was the process that would leapfrog TSMC and Samsung and restore Intel’s unquestioned leadership.
But as Pat was trying to rally the troops for this monumental charge, the battlefield itself began to shift under his feet in ways no one could have predicted.
Upended by the Tides of History
Pat’s crusade was “upended” by three seismic industry shocks that no t-shirt or slogan could overcome:
The Apple Divorce: Apple’s decision to switch from Intel CPUs to its own Arm-based M-series silicon was a body blow, costing the Client Computing Group an estimated $4 billion in revenue. Internally, it was a scramble. Marketing teams worked overtime to offset the loss, but it was a clear sign that Intel’s x86 dominance was no longer absolute.
The AI Big Bang: The public launch of OpenAI’s ChatGPT changed the world overnight. More importantly, it revealed the deep alliances between AI labs and NVIDIA. Cloud Service Providers (CSPs), Intel’s biggest data center customers, began diverting billions of their budgets away from CPUs and towards NVIDIA’s GPUs to meet the insatiable demand for AI training.
The AMD Offensive: Compounding the pain, Intel’s own execution stumbled. The persistent delays of the “Sapphire Rapids” Xeon server chip created a massive opening, and AMD, with its excellent Epyc processors, charged right through it, eating into Intel’s most profitable market share.
“In the Saddle”: The Optimist in a Storm
Through it all, Pat remained a relentless optimist. He famously used a metaphor from his experience climbing Mount Kilimanjaro. He’d tell us in all-hands meetings that the company was “in the saddle,” his term for the grueling, high-altitude stretch between two peaks. It’s the hardest part of the climb, he’d say, the part where you feel like giving up, right before you make the final push to the summit.
To his credit, he led the sacrifice from the front, taking a significant pay cut himself and requiring senior management to do the same. But while that pay was later restored as conditions improved, other deep cuts to employee benefits, like the much-loved four-week sabbatical (after four years of service) program, were not. This created a complex internal dynamic: a leader demanding a “torrid pace” while taking away one of the key benefits for surviving it.
His spending was colossal, building massive new fabs based on the conviction that if you build it, customers will come. Yet by the fall of 2025, Intel had to publicly acknowledge that 18A, the node upon which the entire company was bet, still had no major external customers signed up.
This is where my insider perspective diverges from the public story. Internally, we saw the public messaging on 20A, the precursor to 18A, quietly shift. It went from being the node for the Arrow Lake compute tile to being a “testing platform.” Arrow Lake’s most advanced chiplet was then slated to be built by our rival, TSMC. For many of us on the ground, this was a major red flag. We wondered: if 18A was truly on track, why was the company’s public confidence so shaky?
From an analyst’s perspective, this suggests the board may have been seeing yield data for 20A and 18A that indicated the “torrid pace” was not materializing as promised. Pat’s relentless spending, coupled with a delayed payoff, may have made him a liability. He was the visionary Moses, destined to lead his people through the desert but perhaps not into the promised land.
He fought back. He established an AI task force to push the Gaudi accelerator. He correctly identified “sensing” as a new superpower for computing. And, critically, he championed the pivot to the AI PC, integrating Movidius VPUs (rebranded as NPUs), a move that is now central to the company’s client strategy.
But it wasn’t enough. The engineer’s crusade, a heroic and necessary effort to restore the soul of the company, had run out of time. The board, impatient for a return, decided a new leader was needed.
But was Pat’s passionate, engineering-first approach the problem, or was he simply trying to fix a deeper cultural issue that had taken root during the tenure of his more financially-minded predecessor? That is the story of Bob Swan—the accountant who inherited an engineer’s crisis, and whose controversial decisions may have been the painful medicine the company needed to survive.”