Fat Burning & Ketosis: What Actually Happens

Fat does not “melt.” It is dismantled, ferried, and burned by named enzymes, and every step has a switch. Insulin decides whether fat leaves the fat cell at all. Carnitine and the CPT1 shuttle decide whether it can get into a mitochondrion — and CPT1 wears a malonyl-CoA padlock that snaps shut whenever carbohydrate is plentiful. That padlock, not willpower, is the molecular reason a high-carb, high-insulin body burns very little fat. Press Play, then jump to Fasted 72 h and watch the whole 72-hour movie run: glycogen drains, insulin falls, the padlock drops off, fatty acids stream through the gate, and the liver starts shipping ketones to your brain.

Try this: sit on Fasted 72 h until ketones are high — then press 🍞 Carb load. The malonyl-CoA padlock slams back onto CPT1, fatty acids pile up outside a locked door, and fat oxidation collapses within seconds. Then try 🥤 Ketone drink and watch blood BHB rise while the fat you actually burn falls.

The Science in Plain Language

Insulin is the switch — “carbs” are not a moral category

A fat cell is a bag holding a droplet of triglyceride: one glycerol backbone with three fatty acids hanging off it. To get fat out, two enzymes have to cut it apart — ATGL (adipose triglyceride lipase) takes the first fatty acid off, and hormone-sensitive lipase (HSL) takes the next. Both are switched off by insulin.

This is the single most important fact on this page, and it is not a metaphor. Insulin activates an enzyme (phosphodiesterase-3B) that destroys the cyclic-AMP signal your lipases need. High insulin, lipases off, fat locked in. Insulin falls, lipases on, fat comes out. That is the padlock you see drop off the droplet in the animation.

Suppressing fat release is the most insulin-sensitive thing insulin does. Lipolysis is half-shut-down at insulin levels well below those needed to move glucose into muscle. In practical terms: it takes only a little insulin to keep fat in storage, which is why a big carbohydrate load stalls fat burning for hours, and why the fasting insulin level on a lab report is a more revealing number than most people realise.

But the honest version of this is not “carbs are evil.” Insulin is the switch, and protein raises insulin too. Whey protein produces an insulin rise comparable to some starchy foods despite containing essentially no sugar. Beef, fish and eggs all raise insulin. So a ketogenic diet is not a licence for unlimited protein — and, symmetrically, the panic about protein “kicking you out of ketosis” is overdone. Protein modestly blunts ketone levels; it also protects your muscle and keeps you full. Do not cut protein to chase a higher number on a ketone meter. That is optimising the meter, not your body.

The gate: carnitine, CPT1, and a padlock made of malonyl-CoA

A fatty acid released from a fat cell is not water-soluble. It cannot simply dissolve in blood, so it travels bound to albumin, the most abundant protein in your plasma — the pale carriers ferrying orange dots in the animation. Albumin drops it at a muscle cell, a transporter called CD36 takes it in, and there it hits a wall.

Fatty acids cannot cross the inner mitochondrial membrane on their own. They must be handed to a carrier molecule — carnitine — by an enzyme sitting in the membrane called CPT1 (carnitine palmitoyltransferase 1). CPT1 is the gate. Nothing burns until CPT1 lets it through.

And CPT1 has a lock on it. The enzyme is inhibited by malonyl-CoA — a molecule your cells make from acetyl-CoA whenever carbohydrate is plentiful, using the enzyme acetyl-CoA carboxylase. Malonyl-CoA is the cell's chemical announcement that “there is plenty of fuel coming in from outside, do not start burning the reserves.” When it is high, CPT1 is shut, and fatty acids stack up outside a door they cannot open.

So the reason a high-carbohydrate, high-insulin state burns very little fat is not mysterious and it is not a matter of willpower. It is two locks in series: insulin keeps fat inside the fat cell, and malonyl-CoA keeps whatever escapes from entering the mitochondrion. Press Carb load in the animation and you can watch both slam shut in real time.

An honest verdict on carnitine supplements

The gate needs carnitine, so surely taking carnitine opens the gate wider? For most people, no — and it is worth being straight about who it does help.

Carnitine supplements reliably help people with a genuine carnitine deficiency. That is not marketing; it is medicine. Primary carnitine deficiency (a genetic defect in the OCTN2 transporter) is treated with levocarnitine and the treatment is life-changing. So are the secondary deficiencies seen in some people on long-term dialysis, in some people on valproate, and in certain inherited disorders of fatty-acid oxidation. In those situations the tank is genuinely empty and filling it works.

In people who already have plenty of carnitine, it has not shown a convincing fat-loss benefit. Healthy muscle is already close to carnitine-saturated, and swallowed carnitine barely raises muscle carnitine at all unless it is taken with a large carbohydrate load, repeatedly, for weeks — because it needs an insulin signal to be pulled into muscle, which is a slightly absurd requirement for a fat-burning aid. Pooled analyses of weight-loss trials find an average difference on the order of a kilogram, from short, small, heterogeneous studies, and the effect tends to shrink as trial quality rises. If you are carnitine-replete, carnitine is not the missing piece. The gate is not short of keys; it is padlocked.

Inside the matrix: sawing a fatty acid two carbons at a time

Once through the gate, a fatty acid meets β-oxidation: a four-step loop that chews the chain from the business end, two carbons at a time. Each pass snips off one acetyl-CoA and hands the mitochondrion one NADH and one FADH₂ — the electron carriers that feed the electron transport chain and turn into ATP.

Take palmitate, the 16-carbon fat that dominates most diets. Seven passes of the saw reduce it to nothing, yielding 8 acetyl-CoA, 7 NADH and 7 FADH₂. (The last pass splits a 4-carbon fragment into two acetyl-CoA at once — which is why 16 carbons give 8 acetyl-CoA but only 7 turns of the loop.) Feed all of that into the machinery, and one molecule of palmitate is worth on the order of ~106 ATP, against roughly ~30–32 ATP for one molecule of glucose. Per gram: 9 kcal of fat against 4 kcal of carbohydrate. Fat is a vastly denser fuel, which is exactly why your body stores energy as fat and not as sugar.

There is a real cost, and it deserves to be said out loud: fat needs more oxygen per unit of ATP. Burning palmitate consumes 23 O₂ for those ~106 ATP; glucose consumes 6 O₂ for ~32. Do the division and carbohydrate delivers roughly 10–15% more ATP per litre of oxygen. When oxygen is not limiting — at rest, walking, sleeping, fasting — that costs you nothing. When oxygen is the limiting factor, which is precisely what happens at high exercise intensity, the body shifts back toward carbohydrate on purpose. It is not confused. It is being efficient with the resource that has actually run short.

Where ketones come from: an acetyl-CoA traffic jam in the liver

Acetyl-CoA cannot simply be burned on its own. To enter the Krebs cycle it has to be handed to a partner molecule — oxaloacetate — and the two are condensed into citrate. No oxaloacetate, no entry.

Now fast for a day. Your liver glycogen — about 100 g, roughly a day's worth of blood sugar — is gone within about 24 hours. (Muscle holds far more glycogen, around 400 g, but muscle cannot export it: it lacks the enzyme glucose-6-phosphatase, so muscle glycogen is muscle's own private fuel and can never become blood sugar for your brain.) Your brain still demands a large amount of glucose, so the liver starts building glucose from scratch — gluconeogenesis — and the raw material it drains for that job is oxaloacetate.

So in the fasting liver, acetyl-CoA is arriving in a flood from β-oxidation while its partner molecule is being siphoned off to make sugar. The queue backs up. And the liver does something elegant with the surplus: it condenses acetyl-CoA into ketone bodies — acetoacetate first, then mostly β-hydroxybutyrate (BHB), plus a little acetone. Acetone is volatile and simply leaves through your lungs, which is the fruity, nail-polish smell on the breath of someone deep in ketosis, and it is what a breath ketone meter detects.

The liver, notably, does not burn the ketones it makes — it lacks the enzyme to reactivate them. It manufactures them for everybody else: heart, muscle, kidney, and above all the brain.

The brain runs on ketones — and that is what saves your muscle

Fatty acids do not cross the blood–brain barrier well enough to fuel the brain. Ketones do — there are dedicated monocarboxylate transporters for them, and they are up-regulated during fasting. In the animation, watch the fatty acids bounce off the barrier while ketones sail through.

This is not a curiosity. It is the reason you are alive. If the brain could burn only glucose, a fasting body would have to manufacture around 120 g of glucose a day, and the only large-scale raw material for that is your own muscle protein. A human without ketones would burn through their skeletal muscle in a couple of weeks. Instead, as BHB climbs over the first few days, the brain switches over: by roughly day 3–4 ketones supply a large fraction of its fuel, and at steady state in prolonged fasting they cover roughly two-thirds of it. The demand for new glucose collapses, protein breakdown falls sharply, and your muscle is spared. Ketosis is not a hack. It is a survival system, and it is very old.

“Fat burning” is not “fat loss” — the myth that has to die

This is the most consequential misunderstanding in the whole subject, so here it is bluntly: you can be in deep ketosis and gain body fat.

Ketosis tells you which fuel you are burning. It says nothing about how much fat you keep. If you eat more energy than you spend — and it is entirely possible to do that on butter, cream, nuts, cheese and MCT oil, all of which are extremely easy to over-eat — then the dietary fat you are so efficiently oxidising is simply substituting for the body fat you would otherwise have used, and the excess goes into storage. A ketone meter reading of 2.5 mmol/L is not a receipt for fat loss. It is a fuel gauge, not a scale.

The same error powers the “fat-burning zone” on gym cardio machines. At a gentle pace, a higher percentage of your calories comes from fat. Total calories are lower, and total fat oxidised over the whole day — which is what actually matters — is dominated by your energy balance, not by the intensity of one workout. Burning fat and losing fat are two different verbs.

What ketogenic eating does reliably do for many people is blunt appetite and make an energy deficit easier to sustain without counting anything. That is a genuine, useful, unglamorous benefit. It is also the whole mechanism.

Keto-adaptation and the “keto flu” — it is mostly salt

The first week on a ketogenic diet or a long fast often brings a headache, fatigue, light-headedness on standing, muscle cramps and irritability. This gets called the “keto flu,” as though it were a mysterious detox. It is not mysterious at all, and it is fixable.

Insulin tells the kidney to hold on to sodium. When insulin falls, that instruction stops, and the kidney dumps sodium — and water follows sodium. On top of that, every gram of glycogen you store is holding roughly 3 g of water with it, so emptying several hundred grams of glycogen releases well over a kilogram of water. That is the dramatic “first week” weight drop, and it is not fat. And the symptoms — the headache, the dizziness on standing, the cramps — are largely the symptoms of a person who is short of sodium and has lost fluid volume, with potassium and magnesium shifting too.

The fix is genuinely boring and genuinely works: deliberately add salt (clinicians supervising ketogenic diets routinely tell patients to increase sodium, which is the opposite of the advice most people expect), eat potassium-rich food, and address magnesium. Most people who quit keto in week one because they “felt terrible” were not failing to adapt. They were low on electrolytes.

Exogenous ketone drinks: they raise BHB and suppress your own fat burning

Ketone salts and ketone esters do exactly what the label says: they raise blood BHB, quickly, sometimes to levels a multi-day fast would take days to reach. What they do not do is accelerate fat loss — and the reason is a beautiful piece of physiology.

BHB is not just a fuel. It is a signal. It binds a receptor on your fat cells (HCAR2, the same receptor niacin acts on) and shuts lipolysis down. That is a negative-feedback loop: ketones tell fat cells “enough, there is fuel in circulation, stop releasing.” So drinking ketones raises your ketone number while lowering your free fatty acids and reducing the fat you release from storage. You have bought a nicer meter reading with a supplied fuel, and simultaneously told your body to stop dipping into the reserve you were trying to spend. They are also not calorie-free.

Press 🥤 Ketone drink in the animation: BHB climbs, and “fat burned” falls. That is not a bug in the model. That is the finding.

Nutritional ketosis is NOT diabetic ketoacidosis — and who should not do this

Nutritional ketosis: BHB roughly 0.5–3 mmol/L. Insulin is low but present. Blood glucose is normal or low-normal. Blood pH is normal. It is stable, and it is self-limiting.

Diabetic ketoacidosis (DKA): BHB commonly 10–20 mmol/L or higher. Insulin is essentially absent. Blood glucose is usually very high (often over 250 mg/dL) because it cannot get into cells. Blood pH falls below 7.30. It causes vomiting, deep laboured breathing, confusion and death. It is a medical emergency requiring intravenous insulin and fluids in a hospital.

Why does one stop at 3 and the other run to 20? The insulin brake. In nutritional ketosis, the small amount of insulin still circulating is enough to restrain hormone-sensitive lipase — and because lipolysis is the most insulin-sensitive process in the body, even a trickle of insulin is enough to cap how fast fatty acids reach the liver, which caps ketone production. Ketosis regulates itself. In type 1 diabetes with no insulin at all, that brake is simply gone: lipolysis runs flat out, the liver is flooded, ketone production has no ceiling, and the acid accumulates faster than the blood can buffer it. Look at the BHB tower in the animation with both bands marked on the same axis. The gap between them is the insulin brake.

One honest caveat, because precision matters here: the ketone number alone does not diagnose DKA — the acidosis does. Guidelines use a BHB threshold around 3 mmol/L together with high glucose and a low blood pH or bicarbonate. A person on a long fast can drift above 3 mmol/L with a completely normal pH and be perfectly fine. It is the collapse of pH, not the ketone reading, that makes it an emergency.

Who should not attempt ketosis or extended fasting:

• Type 1 diabetes — do not do this without specialist supervision. The safety brake described above does not exist in your body.
• Anyone taking an SGLT2 inhibitor (canagliflozin, dapagliflozin, empagliflozin and relatives). These drugs can cause euglycaemic DKA — full ketoacidosis with a blood-glucose reading that looks nearly normal, which is exactly what delays the diagnosis. Combining one with a ketogenic diet or a fast is genuinely dangerous. This is a labelled warning, not a theoretical one.
• Pregnancy and breastfeeding.
• Inherited disorders of fatty-acid oxidation — MCAD deficiency, CPT1 and CPT2 deficiency, primary carnitine deficiency. These people cannot burn fat properly, cannot make ketones on demand, and fasting can drop them into a dangerous hypoglycaemia with no ketones to fall back on. Fasting is the specific thing they are told to avoid.
• A history of eating disorder, and anyone underweight or medically frail.
• Anyone on insulin or a sulfonylurea, without a doctor adjusting the dose first — the hypoglycaemia risk is real and immediate.

What fasting genuinely does well — and what it doesn't

What it does well. It lowers insulin, and it lowers it further and for longer than simply eating less does. For a lot of people it is a far easier way to eat less than counting calories — you close the kitchen instead of negotiating with yourself at every meal. It reliably empties liver glycogen and turns on the fat-burning machinery described on this page. And it is free.

What it does not do. When intermittent fasting or time-restricted eating is compared head-to-head with ordinary calorie restriction matched for calories, the weight loss is broadly similar. The clock is a tool for eating less; it is not an independent fat-burning drug. Fasting also does not selectively remove fat — if protein intake is low and you are not doing resistance training, a meaningful share of what you lose can be lean tissue, which is precisely the thing you want to keep. And whatever the ketone meter says, the same rule applies as on any other diet: ketosis chooses the fuel; energy balance decides how much fat you keep.

↑ Back to the animation

Connections

• All Interactive Visualizations
• Mitochondria — where the fat is actually burned
• Blood Sugar & Insulin — the other half of the switch
• Liver — the organ that makes your ketones
• Fasting
• Caloric Restriction
• Carnitine — the key to the CPT1 gate
• Carnitine Benefits
• Fasting Insulin — the test that shows the switch
• Hemoglobin A1C
• Type 1 Diabetes — where the insulin brake is missing
• Type 2 Diabetes
• Insulin Resistance
• Metabolic Syndrome
• Sodium — the real cause of “keto flu”
• Potassium
• Magnesium