Intel
This is the archived full research. Read the story
The company that designs, and also manufactures, the processors inside most PCs and company servers, now trying to make chips for other designers too.
The scores
Growth2/4 ↑
Client units fell 8% and three prior years shrank; two quarters of price-led growth is too short to call the demand engine repaired.
Business quality2/4 ↓
Real switching cost in the x86 software estate, but the buyer is also the attacker: three customers are 43% of revenue and build their own.
Financial position2/4 →
Net debt near zero after the raise, but the plan needs outside money: capital spending above $20B and three equity raises in 14 months.
Execution record3/4 ↑
Seven straight quarters above guidance and 18A yields ahead of plan, but under two years of it, after a decade of missed nodes.
Per share capture1/4 ↓
Share count up about 21% in 18 months with no buyback, and the warrant takes $19.9B from holders precisely if the foundry is separated.
Potential2/4 ↑
A roughly $77B prize, but nobody has committed to buy: 14A volume production is 2028 and no external customer is named.
Risk profile2/4 →
Not existential with net debt near zero, but binary on one question: whether 14A lands an external customer Intel's own risk factors gate it on.
Valuation1/4 ↓
About 167.3 times operating profit, 320% of a three-stage value near $17.57 at an 8% return: priced for perfection, beyond the 20% margin.
Evidence quality3/4 →
Filed figures throughout, with three competitors' own filings behind the machine block, but the central assumption rests on a management claim about supply.
If it works 0.3× · if it fails you keep about 18%.
The Verdict
We rate Intel Unattractive. It designs and makes the processors in most PCs and company servers and is trying to make chips for other designers too. Sales grew fast, but mostly because a chip shortage let it raise prices, its foundry has no named outside customer, and the share count rose so fast that each share gained little, so the business is weak. The price sits far above our value and our best case, as if its best quarter in years were the floor. As a bet it is fragile: even the best case pays back a fraction, and failure leaves under a fifth.
The long caseWhen the one leading chip factory is full, designers need a second source; Intel's outside foundry works only if they sign up for its wafers and packaging.
How we weighed it
The price against our value
Against it
Price at 320% of our value
The price sits far above our value and above the bull case.
The business
Against it
Weak
A weak business: it keeps less of each sale than it used to.
Growth against what the price assumes
Cuts both ways
Growth leaned on a shortage that is already fading.
What could go wrong
Against it
Prices may ease as supply returns, cloud customers design their own chips and the foundry has no outside customer.
Against competitors
Cuts both ways
No filed rival compares directly; against its own normal it is several times dearer.
How its group is behaving
Cuts both ways
AI chip names move together on spending news.
The latest news and report
Cuts both ways
The last report supported a fading recovery and showed heavy dilution.
The evidence under this - 8 claims
- Part of the price increase is mix rather than like-for-like pricing: Intel's MD&A says 'the majority of the increase in ASPs was driven by a higher mix of premium products sold', with demand-based pricing actions 'contributing to a lesser extent'. A richer mix is a better business, not merely a scarcity rent.
- Intel's CFO said on the call that the like-for-like increases were taken 'because we thought we had seen some inflation on our cost and needed to pass that on', which frames them as cost recovery rather than margin capture - and gross margin at 40.4% is still far below Intel's own 2016-2021 range of 55% to 62%.
- Server pricing is struck per processor core, and core counts rise every generation: Intel's CFO said 'this market is priced on an ASP per core basis. So as you get more cores, you easily get more ASP uplift.' Some of the 48% is therefore selling more silicon rather than charging more for the same silicon.
- Unit growth of 9% is real and positive, unlike the client line - so the server story is not purely price, and the CFO forecasts 'strong double-digit unit growth for the industry this year and next'.
- The CFO qualified the easing sharply on the call: 'while things will improve, we will not catch up. We will be behind in the fourth quarter.' If Intel is still undershipping demand at the end of 2026, the pricing environment may persist past the point this thesis expects it to break.
- The constraint management names is not only its own wafers: substrates, memory and advanced packaging are called 'probably our most challenged parts of our supply chain', and those are third-party shortages Intel cannot fix by adding its own capacity - which cuts both ways, since it also means Intel cannot end them on schedule.
- The CEO's own framing is the opposite of the CFO's easing: 'these shortages will persist for the foreseeable future.' Two statements on the same call, and the thesis leans on the narrower one.
- Management has forecast this easing before and been wrong. The Q4 2025 release of 22 January 2026 said 'we expect our available supply to be at its lowest level in Q1 before improving in Q2 and beyond.' Supply was still short through Q2 2026 and the CFO now expects to be behind in Q4 2026. A forecast that has already slipped two quarters is weak evidence on its third attempt - and it weakens this thesis more than it weakens management, because the thesis needs the easing to arrive.
- 2026-10-29editedthe October 29 release extends the PC slowdown through the first half of 2027 and keeps supply, not demand, the limit on Xeon
- Annualising a single quarter is crude in both directions. Intel's H1 2026 GAAP operating income was negative $1,340M because of $4,240M of restructuring including a $3.9B Mobileye goodwill impairment, so the trailing figure is worse than the annualised one; and if the server ramp management describes arrives in Q4 2026 and 2027, the forward figure is better.
- Operating profit is depressed by a capital programme that is deliberately ahead of revenue - more than $20B of capital expenditure in 2026 against $5,891M of first-half depreciation - so today's operating margin is being charged for capacity that has not yet earned anything. A buyer paying for 2028 is not obviously paying 75 times 2028.
- Intel's Q3 2026 guidance of $0.38 of adjusted earnings per share annualises to $1.52, which at $102.94 is 68 times earnings - a different lens that reaches a similar place, so the conclusion is not an artefact of the operating-profit frame.
- 2026-09-28editedre-struck at the 2026-09-25 close of $123.00 on the daily review's maintenance route -- drift +19.5%, past the 12% trigger. Denominator (operating profit) unchanged; only the price moved.
- External revenue is the wrong measure of a foundry's progress at this stage, because a customer signed today ships wafers in two to three years: 14A risk production is 2027 and volume production 2028, so an external revenue line in 2026 cannot reflect any 14A win that may already exist.
- Intel's own framing is that the internal ramp is the qualification for external business - 'the successful volume ramp of 18A for our internal products provides important validation as Intel Foundry engaged with external customers' - which is a coherent argument that the $293M understates the asset.
- 2026-09-28editedre-struck at the 2026-09-25 close of $123.00 (from $102.94) on the daily review's maintenance route -- drift +19.5%, past the 12% trigger. Trailing sales and book value unchanged since no new 10-Q; matches FMP's own TTM price/sales (10.878) and price/book (7.171) computed fresh at the same close.
The company
Intel makes the central processor, the chip that runs the operating system and ordinary programs in a laptop or a server. Almost alone among chip designers, it also owns the factories that print its chips.
It reports three businesses: PC chips, data-centre chips, and the factories, called Intel Foundry, which it wants to rent to outside designers the way TSMC does.
The evidence under this - 2 claims
How it makes money
The chips pay for everything. In the June quarter they earned $4.82B of operating profit, while the factories lost $2.09B and head-office costs took $1.42B more.
Intel now keeps 40.4 cents of each sales dollar after the cost of making the chip, against 55 to 62 cents in every year from 2016 to 2021.
The evidence under this - 2 claims
Where the money comes from, stream by stream
| Stream | Share of sales (Q2 2026) | How it earns it |
|---|---|---|
| Client processors (notebook and desktop) | 47.7% | Core and Core Ultra processors sold to PC makers, mostly a few large ones; prices rose 27% while units fell. |
| Server processors | 32.9% | Xeon processors sold to cloud providers, server makers and enterprises; prices rose 48% on premium mix, units 9%. |
| Other client (CCPG) | 6.8% | Chipsets, connectivity and edge and industrial parts sold alongside the PC platform. |
| Other data centre (DCAI) | 5.9% | Purpose-built chips designed for single customers, network adapters and AI accelerators. |
| All Other (mainly Mobileye) | 4.3% | Mobileye's driver-assistance chips, about 3.1% of revenue, plus IMS mask-writing tools. |
| Intel Foundry external | 1.8% | Wafers and packaging made for outside customers, mostly Altera since Intel sold control of it. |
Who runs it
Lip-Bu Tan, who ran the chip-design software company Cadence for twelve years, became chief executive in March 2025; David Zinsner has run the money since 2022. Insiders own a tiny slice, well under 1%, though Tan bought more stock in the August share sale. His yearly bonus rides on sales, margin, cost control and a scorecard of goals; his big share awards pay only if the stock rises, none on profit per share.
The evidence under this - 4 claims
- The Top Jobs scorecard includes building an external foundry business, so a quarter of the bonus does reach the foundry question the thesis turns on.
- Tan had to buy $25M of Intel stock under his offer letter and hold it through each vesting date for the new-hire PSUs to vest.
- Two operating officers sold into the same strength: the EVP and general manager of Intel Foundry sold 21,024 shares at $118.279 on 29 May 2026, and the chief legal officer sold 40,256 at $99.526 on 1 May 2026. The person closest to the foundry customer pipeline was a seller near the high.
- A chief executive buying into his own company's marketed offering is also a signalling act with a known audience, and $10.0M against a holding already worth well over $100M changes his own exposure less than the headline suggests.
Why it wins, and who could beat it
Why customers come back
Decades of business software is written for x86, the chip design Intel shares only with AMD. Moving it to another design means rewriting and retesting it, so companies stay, and the two compete mainly with each other.
That lock let the price rises stick: in the June quarter Intel raised PC-chip prices 27% while selling 8% fewer, and server-chip prices 48%. But AMD's data-centre sales more than doubled and passed Intel's, so the lock protects x86, not Intel.
The evidence under this - 4 claims
- Part of the price increase is mix rather than like-for-like pricing: Intel's MD&A says 'the majority of the increase in ASPs was driven by a higher mix of premium products sold', with demand-based pricing actions 'contributing to a lesser extent'. A richer mix is a better business, not merely a scarcity rent.
- Intel's CFO said on the call that the like-for-like increases were taken 'because we thought we had seen some inflation on our cost and needed to pass that on', which frames them as cost recovery rather than margin capture - and gross margin at 40.4% is still far below Intel's own 2016-2021 range of 55% to 62%.
- Server pricing is struck per processor core, and core counts rise every generation: Intel's CFO said 'this market is priced on an ASP per core basis. So as you get more cores, you easily get more ASP uplift.' Some of the 48% is therefore selling more silicon rather than charging more for the same silicon.
- Unit growth of 9% is real and positive, unlike the client line - so the server story is not purely price, and the CFO forecasts 'strong double-digit unit growth for the industry this year and next'.
- The two segments are not like for like. AMD's Data Center segment includes its Instinct artificial-intelligence accelerators, its networking cards and its adaptive chips as well as EPYC server processors; Intel's Data Center and AI segment is overwhelmingly server processors. A large part of AMD's lead is a product line Intel barely sells rather than server sockets taken from Intel.
- Intel's own server unit volumes rose 9% in the same quarter, so Intel is not shrinking in units; both companies are growing into a shortage.
The threat
The deeper threat is the buyers themselves. Amazon, Google, Microsoft and others now design their own server chips on Arm's design, so the sale leaves the market altogether, and Intel's annual report names them as rivals. Apple did the same in laptops.
The evidence under this - 3 claims
The rivals: AMD, AMZN, GOOGL, MSFT, NVDA, AAPL, TSM.
Who else does this, rival by rival
| Ticker | Rival | Where it competes | How far along | Priced against it |
|---|---|---|---|---|
| AMD | AMD EPYC server processorsPriced against it: Same product | Intel Products | On sale | Same product |
| AMZN | AWS GravitonPriced against it: Different product | Intel Products | On sale | Different product |
| GOOGL · MSFT · NVDA | Arm Neoverse-based hyperscaler processors - Google Axion, Microsoft Cobalt, NVIDIA GracePriced against it: One segment only | Intel Products | On sale | One segment only |
| AAPL | Apple M seriesPriced against it: Different product | Intel Products | On sale | Different product |
| TSM | TSMC, the merchant leading-edge foundryPriced against it: One segment only | Intel Foundry external customers | On sale | One segment only |
How big it could get
How big this could get
The market the foundry wants is making the most advanced chips for other designers. TSMC, which dominates it, took about $77B for such work in 2025.
Intel Foundry sold $293M to outside customers in the June quarter, 1.8% of Intel's own sales and a sliver of what TSMC takes.
The evidence under this - 3 claims
- External revenue is the wrong measure of a foundry's progress at this stage, because a customer signed today ships wafers in two to three years: 14A risk production is 2027 and volume production 2028, so an external revenue line in 2026 cannot reflect any 14A win that may already exist.
- Intel's own framing is that the internal ramp is the qualification for external business - 'the successful volume ramp of 18A for our internal products provides important validation as Intel Foundry engaged with external customers' - which is a coherent argument that the $293M understates the asset.
Who has signed
Those who have signed bought shares, not chips. The US government took up to 433 million shares at about $20, with warrants over 240.5 million more that pay out only if Intel gives up control of the foundry. NVIDIA bought $5.0B of shares and agreed to design products together, with no volume or term disclosed.
The evidence under this - 3 claims
Where the growth is coming from, business by business
Where the growth is coming from
| Business | Share of revenue | Growth: last quarter | Last half-year | Last year |
|---|---|---|---|---|
| Intel Products | 94% | +28.2% | +18.5% | - |
| Rest of the company | 6% | -5.7% | -8.9% | - |
| Whole company | 100% | +25.4% | +16.4% | - |
The filed split through 2026-06-27. These rates are not carried forward: the reading below says why.
How we read it: The blend, and neither projection: Intel Products' 18.5% is price, client up 27% on units down 8% and server up 48%, under supply limits management expects to ease. Per share, about 7%.
The results
The numbers
Sales grew 12% over the twelve months to Sep 2026, faster than the −0.5% in the year to Dec 2025.
Operating margin is 2.9%, up from 0.2% in the year to Dec 2023.
Running costs take 37¢ of each sales dollar, up from 31¢ in the year to Dec 2021.
Shareholders' money lost 11% a year, down from 23% in the year to Dec 2021.
It turned 6.3% of sales into free cash, $3.77B over the last twelve months, after spending 23% of sales on buildings and equipment.
It holds $54.0B of cash against $50.1B of debt, $3.87B more cash than it owes.
The share count rose 30% in 4.8 years, and pay in shares costs 4.1% of sales.
All the numbers
Filed figures from the company's 10-K and 10-Q, read 2026-11-10. Fiscal years end in December.
| Dec 2021 | Dec 2022 | Dec 2023 | Dec 2024 | Dec 2025 | 12 mo to Sep 2026 | |
|---|---|---|---|---|---|---|
| Size and growth | ||||||
| Sales | $79.0B | $63.1B | $54.2B | $53.1B | $52.9B | $59.9B |
| Sales growth | +1.5% | −20% | −14% | −2.1% | −0.5% | +12% |
| Profit | ||||||
| Gross margin | 55% | 43% | 40% | 33% | 35% | 40% |
| Operating margin | 25% | 3.7% | 0.2% | −22% | −4.2% | 2.9% |
| Net margin | 25% | 13% | 3.1% | −35% | −0.5% | −20% |
| What growth costs, per sales dollar | ||||||
| Research and development | 19% | 28% | 30% | 31% | 26% | 22% |
| Selling, general and admin | 8.3% | 11% | 10% | 10% | 8.7% | 7.4% |
| Pay in shares | 2.6% | 5.0% | 6.0% | 6.4% | 4.6% | 4.1% |
| Capital spending | 24% | 39% | 47% | 45% | 28% | 23% |
| Cash | ||||||
| Cash from operations | $29.5B | $15.4B | $11.5B | $8.29B | $9.70B | $17.5B |
| Free cash flow | $10.7B | −$9.41B | −$14.3B | −$15.7B | −$4.95B | $3.77B |
| Balance sheet | ||||||
| Cash and short-term investments | - | - | - | $22.1B | $37.4B | $54.0B |
| Debt | $38.1B | $42.1B | $49.3B | $50.0B | $46.6B | $50.1B |
| Net cash | - | - | - | −$27.9B | −$9.17B | $3.87B |
| Returns and shares | ||||||
| Return on capital | 14% | 1.7% | 0.1% | - | - | - |
| Shares (diluted) | 4.09B | 4.12B | 4.21B | 4.28B | 4.53B | 5.31B |
What growth costs
For years Intel's sales shrank while its research and factory bills did not, so each dollar of sales carried more cost. That has turned since 2024, but mostly because Intel cut jobs and spending, not because growth made selling cheaper. Factory spending is about to rise again, so profits only widen if sales keep growing faster.
The evidence under this - 3 claims
The fall since 2024 came mainly from restructuring, a core workforce cut of about 15% (C36), not from scale; and capital spending is guided above $20B for 2026 and higher in 2027 (C31).
- The CFO stressed that a large offset exists and is not in the gross number: the U.S. advanced manufacturing investment credit returns '$0.35 on the dollar back ... on everything we invest in the U.S.', and the lion's share of the spend is domestic - so net capital intensity is materially lower than $20B suggests, on a lag.
- Intel has guided capital expenditure down as often as up in recent years, and the 2024 and 2025 plans were cut. A raised plan is not a commitment.
The last report (29 Oct)
Quarter ended September 2026, filed figures.
In the quarter ended September 26, 2026 sales rose 21% to $16.5B, led by data-centre chips up 78% while PC chips fell 5%. Adjusted earnings were $0.52 a share, above the $0.41 expected. The foundry still lost $1.85B.
Intel guided December sales to $15.0B to $16.0B and adjusted earnings to $0.25 a share, with the share of each sales dollar left after production costs falling to 38% from 44%. The finance chief expects PCs to stay weak into mid-2027. The shares fell about 20%. Intel also raised $22.6B selling new shares, and the count is 16% higher than a year ago.
What it did to the thesisIt supported the view that the recovery leaned on a shortage and is already fading, and that heavy share sales dilute it. The key question, whether prices hold as supply returns, is still unread, and the foundry showed no outside customer, so the view stays Unattractive.
The quarter in detail
Sales growth from the previous quarter, filed figures.
Earnings per share $0.52 against $0.41 expected.
The evidence under this - 5 claims
Intel's guides have been beaten before: Q3 revenue, gross margin and EPS exceeded guidance (release), so the weak margin guide may be conservative.
- The CFO qualified the easing sharply on the call: 'while things will improve, we will not catch up. We will be behind in the fourth quarter.' If Intel is still undershipping demand at the end of 2026, the pricing environment may persist past the point this thesis expects it to break.
- The constraint management names is not only its own wafers: substrates, memory and advanced packaging are called 'probably our most challenged parts of our supply chain', and those are third-party shortages Intel cannot fix by adding its own capacity - which cuts both ways, since it also means Intel cannot end them on schedule.
- The CEO's own framing is the opposite of the CFO's easing: 'these shortages will persist for the foreseeable future.' Two statements on the same call, and the thesis leans on the narrower one.
- Management has forecast this easing before and been wrong. The Q4 2025 release of 22 January 2026 said 'we expect our available supply to be at its lowest level in Q1 before improving in Q2 and beyond.' Supply was still short through Q2 2026 and the CFO now expects to be behind in Q4 2026. A forecast that has already slipped two quarters is weak evidence on its third attempt - and it weakens this thesis more than it weakens management, because the thesis needs the easing to arrive.
- 2026-10-29editedthe October 29 release extends the PC slowdown through the first half of 2027 and keeps supply, not demand, the limit on Xeon
- 2026-10-29editedSeptember-quarter weighted shares replace the June-quarter figure; the offering has now closed
- The same answer qualifies itself twice. The CFO says 'some of that was mix related' and that the like-for-like increases were taken to pass on input cost inflation - so his own framing is mix plus cost recovery, not scarcity rent, and this thesis is using the sentence for more than its author put in it.
- A chief financial officer has an incentive to attribute upside to pricing decisions rather than to a shortage he does not control: the first is management skill and the second is luck. That incentive runs towards 'it was pricing' whether or not it was.
What management did with the money
It sold $20.0B of new shares at $95.00 in August, more than the $15B announced, taking the count up about 21% since the end of 2024. Earlier it paid $14.2B in cash for the half of an Irish factory it did not own. The chief executive bought about $10.0M of the new shares himself.
Insiders in 2026Q2: 30 purchases and 16 sales, from their Form 4 filings.
The evidence under this - 4 claims
- Roughly 735 million of those shares were issued to the U.S. government, SoftBank and NVIDIA in exchange for $8.9B of disbursements and $7.0B of cash - consideration, not a giveaway, and the cash strengthened a balance sheet that needed it.
- Intel paid no dividend over the period and has not repurchased stock, so none of the issuance was offsetting a buyback that flattered earlier per-share figures.
- Two operating officers sold into the same strength: the EVP and general manager of Intel Foundry sold 21,024 shares at $118.279 on 29 May 2026, and the chief legal officer sold 40,256 at $99.526 on 1 May 2026. The person closest to the foundry customer pipeline was a seller near the high.
- A chief executive buying into his own company's marketed offering is also a signalling act with a known audience, and $10.0M against a holding already worth well over $100M changes his own exposure less than the headline suggests.
The price, and where we disagree
In short. Priced for perfection. The price is about 320% of what we think the business is worth, at an 8% yearly return.
The shares returned 315.6% over the year to the autumn, while the share of analysts rating them positively rose only from 18% to 32% on the same panel of 50. After the 29 October report they fell about 20% in a day, but they were priced for the best quarter to be the floor.
The next test is the call and the quarterly filing in early November, which show whether selling prices held as supply returned. The December guide of lower margin is consistent with scarcity fading.
The evidence under this - 4 claims
- The CFO qualified the easing sharply on the call: 'while things will improve, we will not catch up. We will be behind in the fourth quarter.' If Intel is still undershipping demand at the end of 2026, the pricing environment may persist past the point this thesis expects it to break.
- The constraint management names is not only its own wafers: substrates, memory and advanced packaging are called 'probably our most challenged parts of our supply chain', and those are third-party shortages Intel cannot fix by adding its own capacity - which cuts both ways, since it also means Intel cannot end them on schedule.
- The CEO's own framing is the opposite of the CFO's easing: 'these shortages will persist for the foreseeable future.' Two statements on the same call, and the thesis leans on the narrower one.
- Management has forecast this easing before and been wrong. The Q4 2025 release of 22 January 2026 said 'we expect our available supply to be at its lowest level in Q1 before improving in Q2 and beyond.' Supply was still short through Q2 2026 and the CFO now expects to be behind in Q4 2026. A forecast that has already slipped two quarters is weak evidence on its third attempt - and it weakens this thesis more than it weakens management, because the thesis needs the easing to arrive.
- 2026-10-29editedthe October 29 release extends the PC slowdown through the first half of 2027 and keeps supply, not demand, the limit on Xeon
Intel's guides have been beaten before: Q3 revenue, gross margin and EPS exceeded guidance (release), so the weak margin guide may be conservative.
How we value it
- Method
- Cash the business will produce, in three stages
- Required return
- 8%
- Long-run growth
- 4%
- Moat
- No moat
| Bear | Base | Bull | |
|---|---|---|---|
| Sales growth, years 1 to 3 | 6.4%, 9.7%, 9.5% | 12.7%, 19.4%, 18.9% | 15.9%, 24.2%, 23.6% |
| Normal operating margin | 14.7% | 18.4% | 20.3% |
| Return on new capital after year ten | 8% | 8% | 11% |
| Value a share | $12.58 | $17.57 | $33.42 |
- Price
- $56.28
- Priced on
- 6 Nov
- Price as a share of the base value
- 320%
- Yearly return the price offers on the base case
- 6%
- Risk check: valued at 10%, the rate for a cyclical or indebted business
- $9.83
Where each figure comes from
| Figure | Source | Where exactly |
|---|---|---|
| Sales, the last twelve months | Filed report | Quarterly report (10-Q) for the period to 26 Sep 2026, filed 30 Oct 2026; the four quarters to that date. |
| Operating margin today | Filed report | Quarterly report (10-Q) for the period to 26 Sep 2026, filed 30 Oct 2026; the four quarters to that date. |
| Tax rate | Our method's rule | The normal rate where no rate is filed to read: 25% for a US company, 21% elsewhere. |
| Sales per dollar of new capital | Filed report | Annual report (10-K) for the period to 27 Dec 2025, filed 23 Jan 2026: the year's sales over its operating capital. |
| Shares | Filed report | Quarterly report (10-Q) for the period to 26 Sep 2026, filed 30 Oct 2026; the diluted share count. |
| Cash and investments | Filed report | Quarterly report (10-Q) for the period to 26 Sep 2026, filed 30 Oct 2026; the balance sheet on that date. |
| Debt | Filed report | Quarterly report (10-Q) for the period to 26 Sep 2026, filed 30 Oct 2026; the balance sheet on that date. |
| Sales growth, years 1 to 3 | Analysts' forecasts | FMP's consensus for the years to 27 Dec 2026, 27 Dec 2027, 27 Dec 2028 and 27 Dec 2029: the average of 18 to 37 analysts' revenue estimates, 12.7%, 19.4%, 18.9%. FMP publishes how many analysts, not who they are. |
| Normal operating margin | Analysts' forecasts | FMP's consensus operating margin for the year to 27 Dec 2028, the average of 36 analysts' estimates, 25.4%, less stock pay of 4.1% of sales and less the 2.9% the analysts add back on their adjusted basis. FMP publishes how many analysts, not who they are. |
| Return on new capital after year ten | Our method's rule | Set by our reading of the moat: the 8% rate with none, 12% with an average one, 15% with a strong one; the bear case earns the rate. |
| The bear and bull cases | Our method's rule | Our method's spread: the bear case takes half the growth and a fifth off the margin, the bull a quarter more growth and a tenth more margin. |
| Price | Market price | FMP's closing price on 6 Nov 2026. |
What each case rests on
Bear caseGrowth runs at half the analysts' path, 6%, 10% and 10%, the margin settles a fifth lower at 14.7%, and new capital earns only the 8% rate: the downturn test.
Base caseRevenue grows 13%, 19% and 19% over three years, the analysts' path (C72), then fades toward 4%, closing 30% of the gap each year; the margin moves to 18.4%, the analysts' year-three figure after stock pay and on the filed basis (2.9% less).
Bull caseGrowth runs a quarter above the analysts' path, 16%, 24% and 24%, the margin reaches 20.3%, and new capital earns 11%.
What moves the base value
| Lower | Higher | |
|---|---|---|
| Growth in years 1 to 3, a quarter lower or higher | $17.80 | $17.32 |
| The normal margin, 15% lower or higher | $11.81 | $23.33 |
| Capital per dollar of growth, a quarter more or less | $10.20 | $21.99 |
| Return on new capital after year ten, at the rate or one step up | $17.57 | $25.99 |
| The rate, a point higher or lower | $10.59 | $29.63 |
| Long-run growth a point lower | $17.32 | - |
Value a share as growth and margin change
| Normal margin ↓ · growth → | 50% | 75% | 100% | 125% | 150% |
|---|---|---|---|---|---|
| 14.7% | $12.62 | $11.35 | $9.90 | $8.23 | $6.31 |
| 16.6% | $15.32 | $14.58 | $13.73 | $12.77 | $11.67 |
| 18.4% | $18.01 | $17.80 | $17.57 | $17.32 | $17.03 |
| 20.2% | $20.71 | $21.03 | $21.41 | $21.86 | $22.39 |
| 22.1% | $23.40 | $24.25 | $25.25 | $26.41 | $27.75 |
Growth is shown as a share of the base case's. Green: the price is at or under 80% of that value. Red: at or over 120%.
How we read the multiple
Intel lost money in 2025. Buyers now pay about $647B for the whole company, debt included: about 90 times a year of operating profit if the June quarter repeated all year, or 58 times on Intel's adjusted profit.
On sales the shares cost over three times their six-year normal. That extra pays for a chip-making service for outside customers that made 1.8% of Intel's sales in June, with no outside customer named. The shares are up 19.5% since we first wrote this, on the same quarter.
Against its own history. Several times its own normal on every measure, while it keeps less of each sales dollar than it did then. The gap is a bet on the future.
The evidence under this - 3 claims
- Annualising a single quarter is crude in both directions. Intel's H1 2026 GAAP operating income was negative $1,340M because of $4,240M of restructuring including a $3.9B Mobileye goodwill impairment, so the trailing figure is worse than the annualised one; and if the server ramp management describes arrives in Q4 2026 and 2027, the forward figure is better.
- Operating profit is depressed by a capital programme that is deliberately ahead of revenue - more than $20B of capital expenditure in 2026 against $5,891M of first-half depreciation - so today's operating margin is being charged for capacity that has not yet earned anything. A buyer paying for 2028 is not obviously paying 75 times 2028.
- Intel's Q3 2026 guidance of $0.38 of adjusted earnings per share annualises to $1.52, which at $102.94 is 68 times earnings - a different lens that reaches a similar place, so the conclusion is not an artefact of the operating-profit frame.
- 2026-09-28editedre-struck at the 2026-09-25 close of $123.00 on the daily review's maintenance route -- drift +19.5%, past the 12% trigger. Denominator (operating profit) unchanged; only the price moved.
- 2026-09-28editedre-struck at the 2026-09-25 close of $123.00 (from $102.94) on the daily review's maintenance route -- drift +19.5%, past the 12% trigger. Trailing sales and book value unchanged since no new 10-Q; matches FMP's own TTM price/sales (10.878) and price/book (7.171) computed fresh at the same close.
- External revenue is the wrong measure of a foundry's progress at this stage, because a customer signed today ships wafers in two to three years: 14A risk production is 2027 and volume production 2028, so an external revenue line in 2026 cannot reflect any 14A win that may already exist.
- Intel's own framing is that the internal ramp is the qualification for external business - 'the successful volume ramp of 18A for our internal products provides important validation as Intel Foundry engaged with external customers' - which is a coherent argument that the $293M understates the asset.
Priced against its rivals
Only Advanced Micro Devices sells the same thing to the same buyers as a large part of what it does. A price comparison needs at least two, so the price is judged against its own history instead.
The bet: what it pays if it works
This bet is on one part of Intel only: making chips for other companies. We count Intel's own processors separately, at the price the market paid for Intel over six years. That leaves most of today's price resting on a foundry that sells little to outsiders, mostly to a former subsidiary, with no named customer for its next process. Even the best case we can defend pays back only a fraction of today's price.
If it fails, shareholders keep the processor business, worth far less than today's price.
| Intel Foundry external customers revenue in two to three years | Points of the leading-edge merchant wafer market needed | At the stated multiple |
|---|---|---|
| $1.17B | about 2 at 100% kept | 0.2× |
| $4.00B (the big win) | about 5 at 100% kept | 0.3× |
| $7.70B | about 10 at 100% kept | 0.3× |
| $9.17B (market forecast) | about 12 at 100% kept | 0.3× |
TSMC's guide puts leading-edge wafers near $175 billion by 2029 with packaging short; at 3% of wafers, half Samsung's place, and a tenth of packaging, Intel's outside revenue is about $9 billion. On that forecast, $9.17B pays 0.3× from today's value: an Expensive bet on the market case. Intel Foundry external customers would be worth about $41.27B, 0.1× what the price pays for it.
To pay 2× (a Fair bet) it would need about $254.12B of Intel Foundry external customers revenue (about 330 points of the leading-edge merchant wafer market); to pay 5× (a Cheap bet), about $683.53B of Intel Foundry external customers revenue (about 888 points of the leading-edge merchant wafer market), at the same multiple.
Each row is Intel Foundry external customers alone, plus the rest of the company (Intel Products) counted at about $144.65B.
The steps it has to take
- Done External foundry revenue reported, mostly Altera after its deconsolidation
- Not started A named external 14A customer with disclosed committed volume due 2027-07
- Not started Intel 14A in high-volume production due 2028-12
- Not started Intel Foundry external revenue above $1B in a single quarter due 2028-12
The detail behind the label
The evidence under this - 5 claims
- External revenue is the wrong measure of a foundry's progress at this stage, because a customer signed today ships wafers in two to three years: 14A risk production is 2027 and volume production 2028, so an external revenue line in 2026 cannot reflect any 14A win that may already exist.
- Intel's own framing is that the internal ramp is the qualification for external business - 'the successful volume ramp of 18A for our internal products provides important validation as Intel Foundry engaged with external customers' - which is a coherent argument that the $293M understates the asset.
- The CFO argued the capital commitment is itself the signal: 'you can read that now inversely, given our confidence around next year that we must have pretty significant confidence in our customers that we wouldn't be putting the POs in place today.' Intel is spending more than $20B in 2026 and more in 2027, which is expensive behaviour for a company with no demand.
- Customers frequently forbid suppliers from naming them. Absence of a public name is not absence of a contract, and Intel has non-disclosure obligations it cannot unilaterally waive.
The Q2 2026 10-Q's equivalent forward-looking list drops the pause language and instead reads 'our pursuit of Intel 14A and other next-generation leading-edge process technologies ... and our efforts to secure product design wins with and demand commitments from potential significant external customers'. The conditional abandonment risk was softened between January and July 2026, which is evidence the company's own assessment improved.
- 2026-09-28editedre-struck at the 2026-09-25 close of $123.00 (from $102.94) on the daily review's maintenance route -- drift +19.5%, past the 12% trigger. Trailing sales and book value unchanged since no new 10-Q; matches FMP's own TTM price/sales (10.878) and price/book (7.171) computed fresh at the same close.
Where we disagree with the market
Intel's turnaround is under way, and its factories will become a second TSMC.
The recovery is real but bought with price in a shortage, and the foundry has sold almost nothing.
One line of the segment note says outside foundry revenue was $293M, and it grew mainly because Altera, a business Intel used to own, now counts as an outside customer. Nothing else in the results separates it out.
Not in dispute: Intel has $29.7B of cash and short-term investments, and the US government behind it. What is disputed is what that is worth at this price.
What analysts say, and the evidence
The evidence under this - 4 claims
- External revenue is the wrong measure of a foundry's progress at this stage, because a customer signed today ships wafers in two to three years: 14A risk production is 2027 and volume production 2028, so an external revenue line in 2026 cannot reflect any 14A win that may already exist.
- Intel's own framing is that the internal ramp is the qualification for external business - 'the successful volume ramp of 18A for our internal products provides important validation as Intel Foundry engaged with external customers' - which is a coherent argument that the $293M understates the asset.
What could go wrong, and how we would know
The risks
- Prices fall back. The PC and server gains came from a shortage management expects to ease in the second half.
- Customers building their own. Amazon, Google and Microsoft design their own server chips, and AMD's data-centre sales now exceed Intel's.
- A foundry without customers. Intel's risk factors contemplate pausing its next process, 14A, if no significant outside customer signs.
- Heavy spending, cash going out. Capital spending above $20B this year and more in 2027, with adjusted free cash flow minus $8.42B in the June quarter.
- A price on separating the foundry. The government's warrants, about $19.9B above their strike, pay out if Intel gives up control of the foundry.
The evidence under this - 7 claims
- The CFO qualified the easing sharply on the call: 'while things will improve, we will not catch up. We will be behind in the fourth quarter.' If Intel is still undershipping demand at the end of 2026, the pricing environment may persist past the point this thesis expects it to break.
- The constraint management names is not only its own wafers: substrates, memory and advanced packaging are called 'probably our most challenged parts of our supply chain', and those are third-party shortages Intel cannot fix by adding its own capacity - which cuts both ways, since it also means Intel cannot end them on schedule.
- The CEO's own framing is the opposite of the CFO's easing: 'these shortages will persist for the foreseeable future.' Two statements on the same call, and the thesis leans on the narrower one.
- Management has forecast this easing before and been wrong. The Q4 2025 release of 22 January 2026 said 'we expect our available supply to be at its lowest level in Q1 before improving in Q2 and beyond.' Supply was still short through Q2 2026 and the CFO now expects to be behind in Q4 2026. A forecast that has already slipped two quarters is weak evidence on its third attempt - and it weakens this thesis more than it weakens management, because the thesis needs the easing to arrive.
- 2026-10-29editedthe October 29 release extends the PC slowdown through the first half of 2027 and keeps supply, not demand, the limit on Xeon
- The two segments are not like for like. AMD's Data Center segment includes its Instinct artificial-intelligence accelerators, its networking cards and its adaptive chips as well as EPYC server processors; Intel's Data Center and AI segment is overwhelmingly server processors. A large part of AMD's lead is a product line Intel barely sells rather than server sockets taken from Intel.
- Intel's own server unit volumes rose 9% in the same quarter, so Intel is not shrinking in units; both companies are growing into a shortage.
The Q2 2026 10-Q's equivalent forward-looking list drops the pause language and instead reads 'our pursuit of Intel 14A and other next-generation leading-edge process technologies ... and our efforts to secure product design wins with and demand commitments from potential significant external customers'. The conditional abandonment risk was softened between January and July 2026, which is evidence the company's own assessment improved.
- The CFO stressed that a large offset exists and is not in the gross number: the U.S. advanced manufacturing investment credit returns '$0.35 on the dollar back ... on everything we invest in the U.S.', and the lion's share of the spend is domestic - so net capital intensity is materially lower than $20B suggests, on a lag.
- Intel has guided capital expenditure down as often as up in recent years, and the 2024 and 2025 plans were cut. A raised plan is not a commitment.
- Intel may elect to settle net in cash or net in shares, so the exercise need not be a $19.9B cash outflow - but either election transfers the same value away from existing holders.
- The warrants expire in August 2030, so a separation after that date costs nothing. A five-year deterrent is not a permanent one, and the foundry is not ready to separate before then on any published timeline.
- A separation that unlocked the Products multiple could be worth more than $19.9B, in which case the warrant is a toll rather than a blocker. Nothing here establishes which.
What we still do not know
- Who, if anyone, will commit volume to 14A. Intel speaks only of potential significant customers.
- How much of the price rises was a richer mix of products rather than scarcity. Intel has not said.
The evidence under this - 3 claims
- The CFO argued the capital commitment is itself the signal: 'you can read that now inversely, given our confidence around next year that we must have pretty significant confidence in our customers that we wouldn't be putting the POs in place today.' Intel is spending more than $20B in 2026 and more in 2027, which is expensive behaviour for a company with no demand.
- Customers frequently forbid suppliers from naming them. Absence of a public name is not absence of a contract, and Intel has non-disclosure obligations it cannot unilaterally waive.
- Part of the price increase is mix rather than like-for-like pricing: Intel's MD&A says 'the majority of the increase in ASPs was driven by a higher mix of premium products sold', with demand-based pricing actions 'contributing to a lesser extent'. A richer mix is a better business, not merely a scarcity rent.
- Intel's CFO said on the call that the like-for-like increases were taken 'because we thought we had seen some inflation on our cost and needed to pass that on', which frames them as cost recovery rather than margin capture - and gross margin at 40.4% is still far below Intel's own 2016-2021 range of 55% to 62%.
- Server pricing is struck per processor core, and core counts rise every generation: Intel's CFO said 'this market is priced on an ASP per core basis. So as you get more cores, you easily get more ASP uplift.' Some of the 48% is therefore selling more silicon rather than charging more for the same silicon.
- Unit growth of 9% is real and positive, unlike the client line - so the server story is not purely price, and the CFO forecasts 'strong double-digit unit growth for the industry this year and next'.
What would change our view
- Unattractive→NeutralThe price falls back within 20% of fair value
It needs:
- ○The share price falls to about $21.08, back within 20% of what we think the business is worth.about ≈$20.62 today, 83% below the last close
- Unattractive→AttractiveThe price falls 20% under fair value
It needs:
- ○The share price falls to about $14.06, 80% of what we think the business is worth.about ≈$13.49 today, 89% below the last close
- Reopens the researchThe pricing gains turn out to be real
It needs:
- ○Chip prices hold their gains after the shortage eases, and Intel shows how much of the rises was product mix rather than scarcity.checked at the next results, due by 28 Jan
We would also be wrong if
- Intel Foundry breaks even on the profit from its products, not on lower one-off charges.
- The share count stops rising, with no new shares sold through 2027.
What is coming
- 6 November 2026 Q3 2026 call and 10-Q: Client and server price against volume, and Foundry external revenue against $293M.
Every dated event ahead
- 6 November 2026 Q3 2026 call and 10-Q: Client and server price against volume, and Foundry external revenue against $293M.
- January 2027 Q4 2026 results and the FY2026 10-K: The 2027 capital spending plan against 'significantly above' $20B, customer concentration for the year, and any change to the 14A risk factor.
- December 2027 EMIB-T packaging ramps for outside customers: Whether management's packaging backlog becomes external Foundry revenue, the nearest route to outside customers before 14A.
- December 2027 Intel 14A risk production: Whether 14A proceeds on schedule with a significant external customer, or the pause the risk factors contemplate.
Since we wrote this
- 11 NovThesis reviewUnattractive, high conviction: the view holds even in our best caseConviction re-graded on the new anchors: how robust the view is, not whether anything is still open.
- 9 NovThesis reviewStill Unattractive: priced at 286% of a value near $19.67Re-valued in three stages at an 8% return, the method now drawn on every thesis: the price sits beyond the 20% margin of safety.
- 30 OctEarningsThe 10-Q showed prices still rising and a fab stoppageClient prices rose 9% on 14% less volume and one Israeli factory halted for two weeks; the Unattractive view held.
- 29 OctEarningsSales beat, but the December guide cuts margin six pointsRevenue $16.5B up 21% and EPS $0.52 beat; fourth-quarter EPS guide $0.25 and a PC slowdown to mid-2027. The price-versus-volume test waits for the call and 10-Q.
- 27 OctNewsIntel's Classiq purchase, reported at $2.3B, leaves the case unmoved. View heldIntel agrees to buy Classiq, reported at about $2.3B.
- 6 OctNewsMusk says TSMC may join Terafab; Intel falls as its showcase customer is shared. View heldMusk says TSMC could join Terafab, the venture that named Intel 14A.
- 5 OctThesis reviewThe bet now names what it ridesThe owner asked that every bet name the macro trend behind it, after BWXT's bet was missing its chief executive's case for nuclear over gas.
- 4 OctThesis reviewOwner's checklist: revenue streams, cost of growth, management, catalystsThe owner's checklist found the method silent on who runs the company, what growth costs, where the money comes from and what is dated ahead; this pass fills those four blocks from the proxy and filings.
- 28 SepPrice movePrice re-struck, +19.5%; view heldNo print, filing or confirmed development since the thesis. The turnaround is now priced at 90x operating profit rather than 75x on the same June quarter.
- 14 SepThesis writtenUnattractive, medium convictionThe quarter behind a 315% rise was a price rise into a shortage, about two-thirds of the growth went to new shares, and the foundry sold $293M outside.
Also checked 10 times with nothing new to act on.
Sources 28
Sources - 28
Every claim above cites one of these, or states the reasoning that derived it. The quote is frozen at the date it was read: a link can rot, and an agent fetch returns a reading of a page rather than the page. Backs counts the whole claim ledger, so it can name a claim this page does not surface.
- S01primarycompany filing2026-07-24Segment table including the $293M external foundry line, the escrowed-shares note and the $12,529M mark, the Ireland SCIP repurchase, share counts, the liquidity table, and the client and server price-and-volume decompositionbacksC01C02C03C04C05C07C08C12C13C14C16C18C19C20C32C33C34C38C43C52C59C66
- S02primarycompany release2026-07-23The GAAP to non-GAAP reconciliation, Q3 2026 guidance, adjusted free cash flow of negative $8,419M, and the full-year adjusted operating expense figurebacksC02C03C07C18C32C37C44C47
- S03primarytranscript2026-07-23Intel Q2 2026 earnings call transcriptManagement's answers on pricing, capital expenditure and the ASIC line; and separately the 15 analyst turns from seven firms, including Bank of America's direct question on when foundry confidence becomes customer announcements
I mean client obviously exceeded expectations. I would say it was largely ASP, of which some of that was mix related. ... Due to strong customer demand signals, we're raising our outlook for 2026 and now expect our CapEx to be more than $20 billion, which is up significantly versus our expectations entering the year. ... We'll increase tooling in '26 by 40% relative to '25.
backsC06C07C14C31C42C45C57 - S04primarycompany filing2026-01-23Customer concentration at 43%, the segment Competition sections naming AMD, NVIDIA, the hyperscalers, TSMC and Samsung, the 14A abandonment risk factor, the SoftBank and NVIDIA share sales, and the restructuring headcount figurebacksC01C15C22C23C35C36C40C41C46C60C64C66
- S05primarycompany filing2025-08-25The full terms: $8,869.8M of disbursements, 274,583,000 shares at closing and 158,740,000 in escrow, warrants over 240,516,150 shares at $20.00 triggered by foundry ownership falling below 51%, the five-year expiry and the voting covenantbacksC16C35
- S06primarycompany release2026-08-12The announced size and the stated use of proceedsbacksC10
- S07primarycompany release2026-08-12210,526,315 shares at $95.00, upsized from $15B to $20B, net proceeds of approximately $19.7B and a 30-day option over 31,578,947 more sharesbacksC10C34
- S08primarycompany filing2026-08-05Data Center revenue of $6,718M against $3,240M a year earlier, for the identical quarter end - the competitor's own account of the same demandbacksC20
- S09primarycompany filing2026-02-04AMD's own description of EPYC as x86 server processors sold to hyperscalers, OEMs and ODMs - the product-match evidence behind the cohort entryread, but no claim rests on it
- S10primarycompany filing2026-04-16Net revenue of US$121,423M, wafer fabrication at about 86%, advanced technologies at 74% of wafer revenue, 15,022 thousand wafers shipped, and the high-performance-computing platform series - the whole prize build and the driver ratebacksC24C25C26
- S11primarycompany filing2026-05-26Revenue of $4,920M up 23% with royalties up 21%, and the licensor's own statement of an established cloud presence with the largest hyperscalersbacksC39
- S12primaryvendor data2026-09-14AWS Graviton product documentation Amazon Web ServicesThe buyer's own published account of the substitute it built: over 120,000 customers on Graviton and a stated price advantage against x86
AWS Graviton-based instances cost up to 20% less than comparable x86-based Amazon EC2 instances.
backsC21 - S13primaryvendor data2026-09-14Research pack, frozen at content/INTC/research-pack.jsonPrice, returns, the analyst consensus series, price-target drift, short interest from FINRA, insider statistics, headcount, the segment XBRL read and the valuation ratios against their six-year mediansbacksC27C29C30C36C43C49
- S14primarycompany release2026-01-22The full-year 2025 baseline this thesis measures the recovery against - $52.9B of revenue, a negative 4.2% reported operating margin, 5.5% adjusted, and $0.42 of adjusted earnings a share for the whole year - plus the Q1 2026 guidance and the January 2026 statement that supply would improve from Q2 2026 onwards, which is the forecast that has since slippedbacksC47
- S15primaryregulator2025-07-0426 U.S.C. 48D - advanced manufacturing investment credit U.S. House Office of the Law Revision CounselThe statutory rate behind the CFO's '$0.35 on the dollar', and the amendment note dating the rise from 25% to 35% to Public Law 119-21 for property placed in service after 31 December 2025
35 percent of the qualified investment for such taxable year with respect to any advanced manufacturing facility
backsC48 - S16primarytranscript2026-01-15TSMC Q4 2025 earnings call transcript TSMthe merchant leader's own five-year revenue and AI-accelerator growth guide to 2029, and advanced packaging's share of its revenue in 2025 and 2026; read through transcript_fetch (FMP)
Based on our planning framework, we raised our forecast for the revenue growth from AI accelerator to approach a mid- to high 50s percent CAGR for the 5 years period from 2024 to 2029. Underpinned by our technology differentiation and broad customer base, we now expect our overall long-term revenue growth to approach 25% CAGR in U.S. dollar terms for the 5-years period starting from 2024. ... Laura, the revenue contribution last year from advanced packaging is close to 10%. It's about 8%. For this year, we expect it to be slightly over 10%. Okay. We expect it to grow in the next 5 years, higher or faster than the corporate.
backsC51C53 - S17primarytranscript2026-07-16TSMC Q2 2026 earnings call transcript TSMTSMC's raised 2026 growth, its added Arizona investment for 2nm and packaging, and its chief executive on packaging shortage and on Intel's EMIB-T as overflow; read through transcript_fetch (FMP)
we would like to announce an additional $100 billion investment in Arizona. This is to build several more semiconductor logical wafer fab for 2 nm and below technologies, as well as advanced packaging fabs to support the strong multi-year demand from our leading U.S. customers. ... we now expect our full year 2026 revenue growth to be slightly above 40% year-over-year in U.S. dollar terms. ... Well, let me say that our packaging capacity is so tight that now it's limiting my customers' growth. We welcome that additional flexibility in the market.
backsC51C53C56 - S18primaryindustry data2026-09-09an independent tracker's quarterly foundry ranking: TSMC's and Samsung's revenue and share and the tenth-place threshold, which Intel's external foundry revenue sits below; read through ir_fetch
Among the top 10 foundries, TSMC maintained its lead with 2Q26 revenue nearing $40.2 billion, a 12.1% increase QoQ, capturing a 72.5% market share. ... Revenue edged up 1.8% QoQ to $3.26 billion. However, Samsung’s market share declined to 5.9% as competitors grew at a faster pace. ... PSMC ranked tenth, with revenue rising 11.9% QoQ to $432 million.
backsC52 - S19primaryindustry data2025-11-25an independent tracker's reading of which buyers are moving packaging to EMIB and why, and the technical limit that keeps GPU vendors on CoWoS; read through ir_fetch
Furthermore, the increasing demand for larger package sizes and US localization requirements are encouraging major North American CSPs, like Google and Meta, to collaborate with Intel on EMIB adoption. ... Consequently, EMIB mainly attracts ASIC customers rather than GPU vendors, who demand very high bandwidth and very low latency. ... As Google plans to implement EMIB in its 2027 TPU v9 and Meta considers it for its MTIA accelerators, EMIB is set to significantly boost IFS's growth.
backsC54 - S20primaryindustry data2026-04-30the tracker's view that the packaging shortage eases from 2027 as CoWoS grows over 60%, and that Samsung and Intel trail TSMC at 3nm; read through ir_fetch
TrendForce expects that the severe shortage in global 2.5D packaging capacity will begin to ease slightly by 2027, aided by order spillover and TSMC’s plan to expand CoWoS capacity by over 60% by 2027. ... From a supply perspective, Samsung and Intel remain behind TSMC in 3 nm foundry progress.
backsC54 - S21secondarynews2026-09-14TrendForce news digest - TSMC Reportedly Targets 22% 2nm, 16%+ 3nm Capacity Boost by Mid-2027; CoWoS to Double by 2028 TrendForce (citing Economic Daily News and Wccftech)press-reported capacity estimates for CoWoS and for Intel's EMIB-T in CoWoS-equivalent wafers, the only figure found for Intel's packaging capacity; secondary, from unnamed analysts; read through ir_fetch
Wccftech , citing the Economic Daily News , reports that CoWoS capacity could double from roughly 130,000 wafers per month by end-2026 to 260,000 by end-2028. ... Analysts cited by the reports estimate Intel’s EMIB-T capacity, measured in 12-inch CoWoS-equivalent wafers, could climb from 15,000-20,000 per month in 2027 to 40,000-45,000 in 2028.
backsC55 - S22primarycompany filing2026-01-29a large US chip buyer's own filing of its collaboration with Samsung to make advanced AI semiconductors in the U.S. - the onshored second source Intel competes withbacksC56
- S23primarycompany filing2026-03-23beneficial ownership as of 20 March 2026 and the 5% holders, the 2025 annual bonus metrics and payout, the 2025 PSU design, and the chief executive's new-hire options and PSUsbacksC60C61C63
- S24primarycompany filing2023-01-27the 2020 to 2022 cash flow statement: dividends paid, the last share repurchases, capital spendingbacksC64C66
- S25primarycompany filing2026-10-29September-quarter income statement, segment revenue and operating income, weighted shares, the $22.6B equity raise and the fourth-quarter guidebacksC07C08C67C68
- S26primarycompany filing2026-10-30client and server price and volume split, Intel Foundry external revenue, the Kiryat Gat stoppage and the market-trends discussionbacksC69C70C71
- S27primarycompany filing2026-11-09trailing revenue, operating profit, stock pay, tax, debt, cash and the diluted share countbacksC73
- S28secondaryvendor data2026-11-09the analysts' average revenue and operating profit for the next three fiscal years, and the closebacksC72C73