Phase Space AI

Valuation

Intel Corporation [INTC]

Intel Corporation [INTC] — Valuation

Spot $86.24 (2026-07-28 close; $81.92 intraday 2026-07-29). Shares 5,043m. Market cap $434.9bn. Net debt $20,810m. EV $455.7bn. EV/Sales 7.99x. EV/EBIT: not defined — TTM operating income is −$77m.

---

1. Cost of capital

A 17.6% WACC is aggressive and it is not the load-bearing input. §4 runs the full WACC sensitivity: at 9.0% — a generous normalisation implying Intel's cost of capital collapses back to mega-cap levels — the answer still fails. The conclusion is robust to the discount rate, which is stated explicitly so that the high beta cannot be blamed for it.

---

2. Terminal EBIT margin: UNIDENTIFIED

The brief instructs: "If terminal margin is underivable from INTC's own history, declare it UNIDENTIFIED rather than borrowing a peer's." It is underivable. Here is the record.

Span: 46.6 percentage points. Mean −0.6%. Standard deviation ~15pp on five observations. There is no central tendency to estimate and no trend to extrapolate — the series is a step function around a structural break.

Three independent reasons the margin cannot be identified from outside:

1. The distribution has no centre. Any point estimate inside a 46.6-point range is a choice, not a derivation.
2. The consolidated margin is a transfer-pricing output. Q2 2026: Intel Products +31.8%, Intel Foundry −36.2%, intersegment eliminations +$254m. The consolidated figure depends on the internal wafer price between them, which is set by Intel and disclosed nowhere.
3. Intel says the allocation is impracticable. 10-Q, verbatim: "We estimate that the substantial majority of our consolidated depreciation expense… was incurred by Intel Foundry… As a result, it is impracticable to determine the total depreciation expense included as a component of each Intel Products operating segment's operating income (loss)." The company cannot split the largest cost item; an outside analyst cannot either.

Consequence for method. Because the input is unidentified, it becomes the output. The implied-path test is run solving for the terminal margin — which is exactly what reverse_dcf.py --solve terminal_margin exists for, and it removes the one parameter nobody can observe from the assumption set.

Rejected: the screen's 9.1% "industry median of mature profitable peers." Intel is not a mature profitable peer of anything; its own record spans a range four times wider than the substituted value.

---

3. Exit multiple — declared imperfect, and the imperfection sized

Intel's own multiple history is not usable as an anchor, and this is the honest reading rather than a convenient one:

Three regimes in six years: a profitable near-monopolist (2020–21, EV/Sales 3–4x); a collapse (2022–24, 1.5–3x); a government recapitalisation and AI re-rating (2025–26, 2.5 → 8.0x). valuation.md is explicit that a history spanning a regime change should be declared unusable rather than averaged. Declared.

Growth-matched comparator set, and why it too is imperfect. Two brackets were built from the 3,980-name universe, both restricted to SIC division 36 (electronic components / semiconductors) and market cap >$5bn:

Largest members, bracket A: MU (7%, 38x), AMAT (3%, 45x), QCOM (0%, 14x), MSI (9%, 27x). Bracket B: AVGO (24%, 62x), APH (22%, 26x), VRT (22%, 54x), MPWR (16%, 80x).

Both brackets fail the strict test in the same way: neither brackets Intel's exit-year growth, because Intel's exit-year growth is itself not estimable (five-year CAGR −5.7%, latest quarter +25.4%). The two are presented to show the range is not identified, not to pick one. 28.1x (bracket A median) is used as the reference and 11.9x to 50.8x is carried through the sensitivity, which spans both brackets end to end.

---

4. THE IMPLIED-PATH TEST — the Valuation Criteria

assets/reverse_dcf.py, solving for terminal EBIT margin. Terminal value is 100% of EV in this instrument, so the reverse DCF is the primary output.

Held fixed and named: revenue CAGR (varied across the grid); exit multiple (varied); WACC 17.6%; horizon 5 years; TTM revenue $57,032m; EV $455,714m (7.99x TTM revenue).

>>> THE MARKET REQUIRES: a 39.7% terminal EBIT margin, at a 10% revenue CAGR and a 28.1x exit.

The achievability frame

Even Intel Products in isolation — a hypothetical fabless Intel with no Foundry and no corporate unallocated costs — earns 31.8%, which is 7.9 points BELOW what the price requires of the whole company.

Full grid: required terminal EBIT margin

The requirement only falls inside Intel's historical range (≤24.6%) in the bottom-right corner: a ≥20% five-year revenue CAGR combined with an exit multiple at or above 46.1x, the 75th percentile of a comparator set that does not bracket Intel's growth. Against a five-year demonstrated revenue CAGR of −5.7%.

The test run the other way, for completeness

Holding the terminal margin at Intel's best-ever 24.6% and solving for revenue CAGR:

This is the honest and uncomfortable part: on the run-rate alone the growth requirement is met by 4.3pp. It is reported rather than suppressed.

Verdict: FAIL

The reason is jointness, and it is stated as the operative logic rather than as a slogan:

• The run-rate growth requirement (21.1%) is met only if the terminal margin is simultaneously the best in Intel's modern history (24.6%).
• The margin requirement at any growth rate Intel has actually sustained over five years is 39.7% to 50.1% — off the scale of its own record.
• Both conditions must hold together, for five years, while Intel Foundry loses $2,089m per quarter on $293m of external revenue and the CFO has just guided capex up.
• And the terminal margin — the parameter doing all the work — is UNIDENTIFIED, so the "pass" case rests on assuming the single number the company itself says is impracticable to determine.

Stated margin for the strategy to rank on: −15.1pp (demonstrated maximum operating margin minus the margin required at a 10% CAGR and a 28.1x exit).

Implied multiple compression: not statable in the usual form, because Intel has no current EV/EBIT. On EV/Sales the price today is 7.99x against an own six-year median of 2.99x — the implied requirement is that Intel sustain a 96th-percentile multiple, not compress to one.

WACC sensitivity (10% CAGR, 28.1x exit) — the conclusion is not a beta artifact

At a 9.0% WACC the requirement is still 27.2% — above Intel's best-ever 24.6%. The FAIL does not depend on the 17.6% discount rate.

---

5. The 12-month target: UNIDENTIFIED

valuation.md: "If the history is too short or spans a regime change, declare it UNIDENTIFIED rather than substituting a peer median."

Intel fails three separate preconditions for a 12-month target:

1. No usable earnings base. TTM GAAP EPS is negative and the loss is caused by the share price rising (§Research 5). A GAAP P/E cannot be formed. Guided Q3 2026 GAAP EPS is $0.31 and non-GAAP $0.38 — a $0.07 gap on a $86 stock, i.e. the multiple would be built on rounding.
2. No usable multiple history. Three regimes in six years (§3). The own EV/Sales median of 2.99x describes a company that no longer exists in either direction.
3. No named, dated product event with quantified revenue. 18A and 14A are named in the releases but no filing read for this memo attaches a dated volume ramp or a customer commitment to either, and external foundry revenue is $293m a quarter.

Declared UNIDENTIFIED. What is reported instead is the bound, because the bound is the useful output. NTM revenue is taken as ~$67bn (Q3 guided midpoint $16.3bn, held roughly flat-to-up through Q2 2027 — an extrapolation, labelled as one), net debt $20,810m, shares 5,043m:

A range of −71% to +108% is not a target. It is a demonstration that the multiple, not the earnings, determines the answer — which is the finding, and the reason a number is refused.

One anchored cross-check, on the only forward earnings figure Intel has published: annualising the guided Q3 non-GAAP EPS of $0.38 gives $1.52, a forward non-GAAP P/E of 56.7x, against Intel's own trailing-P/E 90th percentile of 17.9x and median of 9.0x. Even on the most flattering earnings measure available, the multiple is 3.2x the 90th percentile of its own history.

---

6. Downside case with a named cause (MEASURED — logged, does not reject)

Primary named cause: Intel Foundry never wins a material external customer.

The evidence is not a projection, it is the current disclosure: external third-party foundry revenue was $293m in Q2 2026 and $467m in H1 2026, against an Intel Foundry segment operating loss of $2,089m in the quarter. If 18A and 14A do not convert — and the mention-frequency series shows Intel itself de-emphasising both by 60–84% over nine quarters (§Research 7) — then the $16.1bn of subsidised gross PP&E and the ~$105bn net PP&E support only internal demand, the transfer-price subsidy from Products to Foundry persists, and the consolidated margin stays in the 0–12% band it has occupied since 2022 rather than returning to 24.6%.

Quantified on Intel's own multiple distribution, NTM revenue $67bn:

Base named downside: −57% to −59%. Probability: 35% (judgement, logged for Brier scoring).

Is it a going-concern case? No, and it must be argued rather than asserted. Liquid assets $29.7bn; H1 2026 free cash flow positive $1.9bn; Q2 operating cash flow $7.0bn; the US government is now an equity holder with 143m escrowed shares and 241m warrants, which is a strong (if uncomfortable) statement about political willingness to support the balance sheet; and the AMIC credit rate was just raised to 35%. This is a permanent-impairment-of-value case, not a solvency case.

Second named downside, mechanical and specific to this name: the escrow derivative liability is marked to Intel's own share price. A further rally increases the reported GAAP loss. It is not a driver of permanent loss, but it is why any screen, index rule, rating methodology or covenant that keys off GAAP earnings will read Intel's improvement as deterioration — and it is the reason no GAAP-earnings multiple appears anywhere in this memo.

Third, and it cuts the other way — the strongest contrary fact, stated fairly. Q2 2026 was, in the CEO's words, "our strongest revenue growth in more than fifteen years." Gross margin +12.9pp, operating margin +35.8pp, DCAI +59% with segment margin 16.1% → 39.5%, free cash flow positive for the first time since 2021, and non-GAAP EPS +$0.42 against −$0.10. If the operating turn continues at this rate the Valuation FAIL becomes a timing question rather than a structural one, and the Quality Criteria PASS is the record of that. What the price requires is nevertheless not what the business has ever delivered, and that is the Criteria's question.

---

7. What would change the analysis