Intel (INTC): The Turnaround That’s Been “Turning Around” for Five Years
The $100B foundry bet Wall Street is already calling a success.
For: Contrarian / speculative investors
Horizon: 3–5 years — this is a bet, not a hold
Intel is up 146% from its 52-week lows. Wall Street is calling it a comeback story.
I’ve been watching this one for a while. The question: is the turnaround real, or am I just paying for hope?
At $43, I’m paying for hope. The stock is above my trim zone — and none of the proof has arrived yet.
TL;DR
Intel is up 146% from its lows on hope, not proof.
The foundry business loses $2-3B/quarter and has almost no external customers.
18A yields are in low double digits vs TSMC’s 60%+ on comparable nodes.
At $43, the stock is above my $40 trim zone — priced for a turnaround that hasn’t turned.
My buy zone: $22-25 — that’s a 45-48% pullback from here.
If proof comes (Panther Lake ships, foundry wins customers), I’ll reassess. Until then, pass.
Core Thesis
Intel is a turnaround story that still hasn’t turned. The stock is priced for success, not execution — and the foundry math doesn’t work yet.
At $43, you’re paying 50-60× earnings for a company with no external foundry customers, yields that don’t work, and a CPU business still losing share. The market is betting the turnaround is already won. I’m not.
What the business actually is
Intel used to be simple: the company that made the chips inside your PC. That business is still there, but it’s shrinking relative to the competition, and management is betting the entire company on something much harder.
Today, Intel is three businesses in a trench coat:
Intel Products Group (~$12.7B/quarter): The legacy CPU business. PCs, laptops, data center processors. This is what actually makes money. Client Computing did $8.5B in Q3 2025, up 8% sequentially. Data Center did $4.1B, down 1% YoY. The PC refresh cycle is helping, but AMD keeps taking server share. Intel’s x86 server market share has fallen from 94% in 2020 to ~73% in 2025.
Intel Foundry (~$4.2B/quarter): The moonshot. Intel wants to become TSMC — manufacturing chips for other companies, not just itself. This is where the $100B+ in capital investment is going. It’s also where the losses are piling up: $2.3B operating loss in Q3 2025, $3.17B loss in Q2. The foundry has almost no external customers — the vast majority of that revenue is Intel making chips for itself.
Other (Mobileye, Altera remnants): Distractions being sold off or spun out.
Valuation: At $43, Intel trades at 50-60× forward earnings — higher than AMD (~25×) or TSMC (~20×) despite worse execution and margins. The market is pricing in turnaround success that hasn’t materialized.
Where the returns actually came from
Intel: ~$68 in 2020 → ~$43 today. Down 37% over five years.
Let’s be honest about what happened:
Earnings collapsed. EPS went from ~$5 in 2020 to barely positive now. The core business got worse, not better.
The multiple compressed from ~14× to essentially meaningless (current P/E is 50-60× on depressed earnings).
Buybacks stopped. Intel suspended repurchases to fund the foundry bet.
The 2025 rally (146% off the lows) is pure narrative. Government money, Nvidia partnership headlines, hope that 18A works. None of it is in the earnings yet.
If you bought at the top in 2020, you’re still down 37%. If you bought the 2024 lows at $17, you’ve more than doubled. Same company, wildly different outcomes based on entry price.
That’s the whole game with turnarounds — entry matters more than thesis.
What’s holding the price up
This is a hope machine, not a cash machine. Here’s what’s propping it up:
Government backstop. Intel is getting ~$8B in CHIPS Act grants plus billions in loans. The U.S. government has decided Intel can’t fail for national security reasons. That’s a floor under the stock — but a floor isn’t a thesis.
The Nvidia narrative. Intel announced a partnership where Nvidia will use Intel’s foundry for some chips. Headlines made it sound huge. Reality: it’s a small deal, testing the waters. But it gave momentum traders something to buy.
Short covering. Intel was one of the most shorted large-cap tech names in 2024. The 146% rally forced shorts to cover, which fed the rally. That’s mechanical, not fundamental.
Sector momentum. Semiconductors had a strong 2025. The tide lifted most boats, Intel included.
The embedded narrative: “Intel is becoming the American TSMC.” That’s what 50-60× is pricing in. The problem: the foundry has almost no external customers, yields don’t work, and TSMC is pulling further ahead.
Dependency mapping: This entire thesis depends on one thing — 18A working at commercial yields and attracting real customers. If that happens, Intel re-rates massively. If it doesn’t, the $100B+ in capex becomes the most expensive failed bet in semiconductor history.
Reflexivity: Weak. Unlike Apple or Adobe, Intel can’t use its stock price to fund the turnaround. They’re not doing buybacks. They’re issuing debt. High stock price just means insiders can sell at better prices. It doesn’t feed the fundamentals.
Where the cracks are
Foundry has no marquee customers: The 18A process is in risk production, but no major external customer has committed to volume. Microsoft has a small deal. Amazon has a small deal. That’s it. The big fish — Apple, Nvidia, AMD, Qualcomm, Broadcom — are all using TSMC.
Yield problems: Intel’s 18A yields are still well below commercially comfortable levels — various leaks put them in low double digits at best, versus roughly mid-60% yields for TSMC’s N2. You can’t run a profitable foundry at that gap.
TSMC is at least a node ahead: TSMC is already in volume production on 3nm and ramping 2nm. Intel is still trying to prove 18A works. In real-world production timing, that’s likely 1-2 years of daylight — and the gap isn’t closing.
Server market share keeps bleeding: AMD keeps taking server share. Arm-based chips are eating into laptops and data centers. Intel’s x86 server share has dropped from 94% in 2020 to ~73% in 2025 — 21 points in five years, and still declining.
The math doesn’t work yet: Foundry needs “low- to mid-single-digit billions” in external revenue to break even (per CFO). External customers are contributing rounding-error tens of millions. That’s not a gap — it’s a business that doesn’t exist yet.
Market expectations
Street view: Consensus expects Intel to grow revenue low-single digits over the next 2 years, with margins stabilizing around 40% and EPS grinding toward $1.50-2.00 by 2027. The market is treating INTC as a “turnaround in progress” that deserves a premium multiple because the foundry optionality is real and the government backstop reduces downside risk.
My view: The Street is too optimistic on timing. The foundry has no meaningful external revenue. Yields are in the low double digits. TSMC is at least a node ahead and pulling away. The CPU business is still bleeding share. Until I see proof — external customers committing volume, yields above 50%, Panther Lake shipping on time — I’m not paying 50-60× for a turnaround that’s “turning around.”
The gap between narrative and execution is where the risk lives.
Tripwires I’m watching
External foundry revenue: ~$50M/qtr now → watching for >$500M/qtr (customers are real)
18A yields: low double digits now → watching for >50% (process works at scale)
Foundry operating loss: $2-3B/qtr now → watching for <$1B/qtr (path to breakeven)
Marquee foundry customer: None committed to volume → watching for Apple, Nvidia, AMD, Qualcomm, or Broadcom to commit (proof the big fish trust the process)
Server market share: ~73% now, declining → watching for stable or up 2 consecutive quarters (bleeding stopped)
None of these are close to triggering bullish yet. I’m watching, not buying.
What could move this
Bull path (6-24 months):
Panther Lake ships on time in Q1 2026 and performs well → proves 18A works
External foundry customer announces volume commitment
18A yields hit 50%+ by mid-2026 → foundry losses narrow
Server share stabilizes as new Xeon chips regain competitiveness
Bear path (6-24 months):
Panther Lake delayed or underperforms → 18A credibility damaged
No major external foundry customer by end of 2026
TSMC 2nm ramps successfully → gap widens further
AMD continues taking server share
Timing: The next 12-18 months are critical. Panther Lake in Q1 2026 is the proof-of-concept moment.
How it trades
Up 146% from the 52-week low of $17.67. Currently at $43.47, within 0.5% of the 52-week high.
The stock is above my trim zone. Momentum traders have front-run the turnaround story. Now they’re waiting for proof.
If proof doesn’t come (Panther Lake delays, foundry disappointments), there’s a lot of air underneath this.
Long-term math
Bull (25% probability): Foundry works, external customers, margins expand → $55-60
Base (50% probability): Muddle through, foundry breaks even by 2028, CPU share stabilizes → $28-32
Bear (25% probability): Foundry fails, CPU share keeps bleeding → $18-22
Probability-weighted fair value: ~$30-32
At $43 today, I’m paying 35-40% above fair value. The stock is pricing in bull-case outcomes.
Decision Framework
Buy zone: $22-25 — paying for broken business with free turnaround optionality. That’s a 43-48% pullback from here.
Watch zone: $25-35 — would need proof before buying.
Avoid zone: >$40 — where we are now. Priced for success that hasn’t happened.
If that pullback sounds crazy, remember: this stock was at $17 less than a year ago.
What would change my mind
I’d get more constructive if:
External foundry revenue hits $500M+/quarter
18A yields reach 50%+
Intel wins a marquee foundry customer (Apple, Nvidia, AMD, Qualcomm)
Server market share stabilizes for two consecutive quarters
Foundry operating losses narrow to under $1B/quarter
Any of those would tell me the turnaround is actually turning — and I’d pay up for proof.
Kill switch
Foundry losses exceed $3B/quarter for three consecutive quarters
18A volume production slips into 2027
A major foundry customer publicly cancels or downgrades
CHIPS Act funding gets clawed back materially
Final take
Intel is the kind of turnaround I find genuinely interesting. The foundry bet is bold. The government backing is real. If it works, this becomes one of the most strategically important tech companies in America.
But “if it works” is doing a lot of heavy lifting.
At $43, I’m paying above my trim zone for turnaround success that hasn’t happened yet. The foundry has no customers. The yields don’t work. The CPU business is still losing share. The stock has already run 146% on hope and government money.
My buy zone is $22-25. That’s a 45%+ pullback — and it might never come. If I miss it, I miss it. I’d rather miss a winner than buy hope at a premium.
Turnaround stories reward patience, not FOMO. I’ve got time.
If this is how you like to think about stocks — filings first, cash over stories, clear kill switches — that’s what I do here every week.
I’m building a small group of investors who care more about coverage math, downside risk, and opportunity cost than hype and victory laps. We read the filings, stress-test the story, and keep each other honest.
If you want to be part of that, hit Subscribe to Phaetrix Investing and stay in the loop for the next deep dive.
Disclaimer: This content is for informational and educational purposes only and is not financial advice. Nothing here is a recommendation to buy or sell any security. I’m explaining how I analyze companies, not what you should do with your money.
Sharp risk framing on Intel's foundry bet. The yield gap is the real killshot, low double digits versus TSMC's mid-60s on comparable nodes means Intel can't economicaly compete even if they win customers. What stands out is the reflexivity asymmetry, unlike software companies where a high stock price funds buybacks and compensation leverage, Intel's burning $100B in capex with no stock price feedback loop. The government backstop creates a floor but not a busines model. At $43 the market is pricing in something closer to the bull case whenit should be discounting the muddle-through base case.