You're in the middle of something and the computer just shuts off. No warning, no blue screen. Most people settle on “it overheated.” This article answers three questions instead: how to tell which cause it is, whether leaving it is dangerous, and what to do first.
1. One question splits it — did it slow down just before? Per Intel, heat cuts speed before it cuts power. A shutdown with no slowdown is probably not thermal.
2. The shutdown itself is protection. Intel writes that the machine turns off “to prevent permanent damage.” The genuinely risky moves are overclocking and opening a laptop.
3. Cost follows the same fork. Thermal usually ends at cleaning and thermal paste; non-thermal depends on the part. On a finished product, the logic board is covered for two years.
Each of the three, worked through below.
One question first — did it slow down before it died?
Intel's temperature document gives the sequence.
“If the cores exceed the set throttle temperature, the processor will reduce power to maintain a safe temperature.”
And the sentence after it —
“If that fails, the processor will shut down automatically to prevent permanent damage.”
— Intel, temperature information for Intel® processors
So a computer that shuts down from heat always passes through a slow stretch first — the one where frame rates drop and work stalls. “It was fine and then it just died” means that stage never happened.
| Just before it died | Which document | Do this first |
|---|---|---|
| It noticeably slowed down | Intel — consistent with thermal protection kicking in | Cooler fit and mounting → thermal paste → fans → airflow |
| It died with no slowdown | Microsoft — hardware, drivers, software | Undo the most recent change |
| Our fork from the two documents. Use it to decide where to look first, not as a diagnosis. | ||
If it slowed down, it's heat — Intel's order
Intel defines throttling this way: “Throttling is a mechanism where the Intel® processor lowers its clock speed when the system's temperature rises above TJ Max (or Tcase). This is intended to protect the processor and to alert the user that the system has a cooling issue that needs to be looked at.”
In other words, the slowdown is itself a signal the machine is sending. Intel's checklist when overheating is suspected, in order:
| # | What Intel lists | How to read it |
|---|---|---|
| 1 | Confirm the cooler is rated for this CPU | Parts compatibility comes first |
| 2 | Confirm the cooler is installed correctly | A loose mount will not perform |
| 3 | Check that thermal interface material was applied in the proper amount | “Proper amount” — more is not better |
| 4 | Confirm the system fans are spinning | |
| 5 | Check airflow | Space around the case and dust belong here |
| Laptops are not the target of this list. For laptops Intel says power and current limits are set by the manufacturer (OEM) and directs you to them. | ||
Intel also puts a number on the limit. Tjunction max is “the maximum junction temperature the processor will allow prior to using internal thermal control mechanisms to reduce power and limit temperature,” and the value varies by product, typically between 100°C and 110°C. The easy mistake here is worth naming: that is not the temperature at which something breaks — it is the temperature at which protection intervenes.
If it died with no slowdown — Microsoft's six steps
Microsoft's sentence on unexpected restarts reads: “Unexpected restarts can be caused by a hardware device, its driver, or software.” Three causes, and heat is not among them. Nor is it in the six steps that follow.
| # | What Microsoft lists | What it rules out |
|---|---|---|
| 1 | Remove newly installed hardware | The first suspect is whatever changed most recently |
| 2 | Start the PC in safe mode | Boots with minimal drivers and software |
| 3 | Check Device Manager | Look for flagged devices |
| 4 | Check for enough free disk space | The document suggests 10–15% free |
| 5 | Install the latest Windows 11 updates | |
| 6 | Restore Windows 11 | Last resort — “choose the recovery option closest to your situation” |
| The document also notes: “In most cases, the restart resolves the issue and there's nothing more you need to do.” | ||
The box that most often turns out to be the answer is probably the first. If you recently swapped memory, changed a graphics card or attached a new peripheral, undoing it is the document's own order — and it is a step that costs nothing and takes minutes.
Is leaving it dangerous, and should you open it?
One reassurance first: the shutdown is not damage but an attempt to avoid it — Intel's phrasing is “to prevent permanent damage.” If it repeats, though, the cause is still in place, so one of the two branches above needs following.
| Action | What you accept by doing it |
|---|---|
| Overclocking | Intel writes that altering clock frequency or voltage may void the warranty and reduce stability, security, performance and the life of the processor |
| Opening a laptop | Intel's five-item list is not for laptops. That territory is directed to the manufacturer |
| Applying extra thermal paste | The document says “proper amount.” Nothing suggests more is better |
| Forcing a restart | Microsoft says most restarts resolve it, but unsaved work is lost |
On cost, a built PC is covered part by part, but a finished product has clear terms.
| Item | Warranty | Within the period |
|---|---|---|
| Computer | 12 months | Performance or function failures in normal use are free |
| Logic board | 24 months — core component | Free for two years from purchase |
| Battery and adapter | Consumables | End of service life is chargeable |
| Samsung's published warranty terms for Korea. Expired warranties, ignoring the manual's cautions, impact damage and non-specified consumables are listed as chargeable. | ||
If it slows without shutting down, see a slow computer. If it won't start at all, when the computer won't turn on; if only the display or sound is out, no sound and no display. If only the internet drops, when the internet won't work; for running costs, the electricity bill calculator.
Questions that come up
It runs above 90°C — is that dangerous?
Intel's document cannot settle that. All it states is that the Tjunction max limit varies by product and typically sits between 100°C and 110°C. You need your own CPU's figure from its spec page, and that temperature is where protection intervenes, not where something breaks.
Will cleaning the dust fix it?
Item five on Intel's list is airflow, so the direction is right. But items one to four come first — cooler compatibility, mounting, thermal paste and fan operation. And opening a laptop is outside this list.
Could it be the memory?
Microsoft names “a hardware device” as the first cause and makes step one “remove newly installed hardware,” so if you recently changed memory, undoing that matches the document's order. The memory diagnostic procedure itself is not in this document.
Does dying without a blue screen mean something different?
Neither document draws that distinction. Microsoft's document treats blue screens, black screens and stop codes as one group. Its note that “in most cases, the restart resolves the issue” is worth remembering.
What should I tell the service centre?
Whether it slowed down just before shutting off. The two documents cover different causes, so that one sentence changes where they look. The pattern helps too — only during games, or a few minutes after startup.
What we read into it
Read Intel's two sentences in order and you get “reduce power to hold a safe temperature → shut down to prevent permanent damage if that fails.” From which follows the conclusion that there is a slow stretch before the shutdown.
And the three causes Microsoft names for unexpected restarts — hardware, drivers, software — do not include heat. The two documents are covering different territory, and neither says so. The fork built on “did it slow down” is ours.
Sources
- Intel (tier 3, manufacturer document) — Temperature information for Intel® processors (checked August 2026). Source for the definitions of Tjunction max and Tcase max, “if the cores exceed the set throttle temperature, the processor will reduce power to maintain a safe temperature… if that fails, the processor will shut down automatically to prevent permanent damage,” and the maximum junction temperature varying by product between roughly 100°C and 110°C.
- Intel (tier 3, manufacturer document) — What is throttling and how can it be resolved (checked August 2026). Source for the definition of throttling, “intended to protect the processor” and to alert the user to a cooling issue, the five checks (cooler rating, installation, proper thermal interface material, fans, airflow), the direction to contact the OEM for laptops, and the overclocking warning.
- Microsoft Support (tier 3, manufacturer document) — Troubleshooting Windows unexpected restarts and stop code errors (checked August 2026). Source for “a hardware device, its driver, or software,” the six steps and their order, the 10–15% free space guidance, “in most cases, the restart resolves the issue,” and “choose the recovery option closest to your situation.”
- Samsung Electronics (tier 3, manufacturer document) — Warranty information (checked August 2026). Source for the 12-month computer warranty, the 24-month core-component warranty on the logic board, the classification of batteries and adapters as consumables, and the list of chargeable cases.
Where to check further
- Your own CPU’s Tjunction max — it varies by product. Intel publishes the figure per model on its product specification pages (ARK), and explains how to find it in its own ARK guide. Laptop makers can set a lower limit, so the ARK figure and the actual behaviour may not match.
- AMD processors — the temperature material here comes only from Intel's documents. AMD's values and behaviour belong to AMD's own documentation.
- The power supply — none of the documents opened here covers PSU capacity or age. On a built PC with repeated shutdowns, that belongs on the list alongside the PSU maker's material.
Written as of August 2026. Temperature and throttling come from Intel's documents, the causes and six steps for unexpected restarts from Microsoft Support, and the warranty terms from Samsung's warranty page; the fork drawn from placing the documents together is marked as ours. If you suspect a hardware failure, do not open the machine — take it to the manufacturer or a service centre. Intel directs laptop owners to the manufacturer too.
