Mike, your read is most of the way there. Carnitine behaves like a recycling system more than a fat burner, and that reframe does most of the work. It has two jobs. The famous one is the shuttle, long-chain fatty acids cannot cross the inner mitochondrial membrane on their own, so CPT1 swaps the CoA for carnitine, a translocase carries it across, and CPT2 swaps it back. That part is real and non-negotiable, and breaking any of those three steps is a genuine metabolic disease. But the shuttle is gated upstream by malonyl-CoA sitting on CPT1, so in a fed state fat transport is throttled no matter how much carnitine is standing around, which is why carnitine availability is almost never the rate-limiting step in a healthy, well-fed person. The second job matters more and gets discussed less. Cells have a finite pool of free coenzyme A, and when CoA is stuck holding acetyl and acyl groups the Krebs cycle jams, like a kitchen where every pan is dirty and nothing new can be cooked. Carnitine strips those groups off, forms acylcarnitines, and parks or exports them, keeping free CoA available. That is the recycle system you were describing, and it explains the entire pattern of who responds: carnitine helps when acyl groups are backing up, and does close to nothing when they are not. On Simon's TMAO concern, it is legitimate but it is context-dependent and usually attached to the wrong variable. You do not make TMAO. Gut bacteria convert unabsorbed carnitine to trimethylamine, your liver enzyme FMO3 oxidizes that to TMAO, and your kidney clears it, so three separate players are involved and only one of them cares what you swallowed. Oral supplements absorb at roughly 14 to 18 percent, so most of a gram never gets absorbed and arrives in the colon as bacterial substrate, which is where the whole TMAO story comes from. Injectable carnitine goes straight into plasma and never meets the colon, so that pathway is bypassed entirely, which answers Mike's question directly. And circulating TMAO is largely a kidney readout, freely filtered with fractional excretion near 100 percent of GFR, so it climbs as clearance falls. In an acute coronary syndrome cohort the TMAO-to-event association lost significance once kidney function was in the model, with eGFR mediating 58 percent of it, and carnitine itself showed no significant association with events. Saltwater fish and shellfish also deliver preformed TMAO that skips the microbes and the liver completely, so a tuna steak moves your number more than a carnitine capsule does, and nobody tells cardiology patients to stop eating fish. If TMAO worries you, the two variables worth knowing are your kidney function and who is living in your colon, the bottle is a distant third, and if you are injecting it is not a variable at all. The injectable specifics are where this gets concrete. Muscle carnitine uptake runs through a sodium-dependent transporter that is already saturated at normal plasma concentrations, a turnstile running at capacity, and pushing more people into the lobby does not make it spin faster. Five hours of IV infusion in healthy men, plasma carnitine driven to roughly ten times normal, and muscle carnitine content did not change at all. Repeat that infusion while clamping insulin at about 150 mIU/L and muscle total carnitine rises about 15 percent, with the threshold sitting somewhere above 50 mIU/L. So a 500 to 600 mg shot fasted before morning cardio raises plasma and extracellular carnitine sharply and the kidney clears most of it, while the muscle pool, the one that would actually change fat oxidation, barely moves. It still has a job, just a narrower one, as a plasma-compartment tool: acylcarnitine buffering, hepatic and renal handling, endothelial and red cell membrane exposure. Those are real compartments with real effects, they are simply not the compartment the marketing is selling. If you genuinely want to load muscle, the honest protocol is unglamorous, roughly 1.36 g of carnitine tartrate with about 80 g of carbohydrate twice daily for six months to raise the store about 20 percent, which sits awkwardly next to selling it as a fat-loss aid. For fat loss specifically, the supportive human data comes mostly from populations that were metabolically impaired to begin with, with pooled RCT work in type 2 diabetes showing around 2 g/day of oral carnitine reducing body mass and reaching most of its effect within about two weeks. That is the tell. It helps the person whose fat oxidation machinery is congested and does very little for the person whose machinery is clean and simply wants it to run faster, so I would consider it in the stalled dieter with a high acylcarnitine-to-free-carnitine ratio, the client grinding long low-intensity volume and clearing acyl traffic poorly, or the insulin-resistant client where mechanism and population evidence line up, and I would skip it as a fasted pre-cardio shot in someone who eats well and trains well. On performance there are two separate claims and they earn different verdicts. Loading does improve fuel selection and work capacity, shifting fat oxidation at low intensity, improving pyruvate dehydrogenase flux, lowering lactate at high intensity, and raising work output, but the entry fee is six months of daily carnitine plus a large daily carbohydrate load, which almost nobody does and almost nobody selling carnitine mentions. The recovery claim is the more defensible short-term one, with around 2 g/day of carnitine tartrate reducing markers of muscle damage and hypoxic stress after resistance training and increasing androgen receptor content in muscle, at doses and timeframes far too small to change muscle carnitine content, which tells me that effect is vascular and oxidative buffering, meaning better perfusion and less local hypoxia around the set, rather than fuel transport. Same molecule, different mechanism, which is why both claims can stand without contradicting each other. Where carnitine stops being optional is high-turnover terrain, and thalassemia is the clean example because it moves three variables at once. Turnover, because red cells are being destroyed continuously and a red cell constantly repairs its own membrane phospholipids, drawing down acyl supply every day. Loss, because iron overload and chelation therapy both stress the renal proximal tubule, which is exactly where filtered carnitine gets reabsorbed, so you have punched a hole in the bucket. Demand, because chronic anemia forces a high-output cardiac state and the heart is the most carnitine-dependent tissue in the body for fatty acid oxidation, so you have raised the rent on the tissue least able to skip a payment. Three deficits stacking is why measured carnitine deficiency in beta-thalassemia is a documented finding rather than a theory, and the trials line up with it: improved echocardiographic indices in beta-thalassemia major and intermedia including reduced left ventricular dilatation and hypertrophy, improved maximal oxygen consumption, cardiac output and oxygen pulse at peak exercise, a direct link between carnitine deficiency and red cell mechanical impairment, and in thalassemia minor patients on hemodialysis a roughly 39 percent reduction in erythropoietin requirement that reversed within weeks of stopping. Propionyl-carnitine specifically protects thalassemic erythrocytes from oxidative stress, which hints that form choice matters more there than it does in a healthy athlete. That setting is repletion of a measured deficit in someone who often has cardiac and renal involvement, so it belongs in a physician-led plan, and my job is spotting the pattern, bringing the literature, and making sure the hematologist and the cardiologist get asked the question. The same three-question check, turnover, loss, demand, applies to any carnitine-wasting picture: dialysis, valproate therapy, pivalate-conjugated antibiotics, ifosfamide, cyclosporine, or any proximal tubulopathy. The forms are also not interchangeable, and treating them as flavors of the same thing is where most of the money gets wasted. Think of the carnitine molecule as a delivery truck, where the attachment is both what is loaded in the back and where the truck is licensed to drive. Oral free carnitine absorbs at 14 to 18 percent, feeds your colonic bacteria with the rest, raises the systemic pool slowly, and carries the highest TMAO cost per unit of benefit. L-carnitine L-tartrate is chemically the same carnitine with a salt added for stability and dissolution, so it does not change gut fate and it does not change the TMAO question, and its real distinction is that it is the form used in most resistance-training recovery studies, so the recovery evidence sits with it by convention rather than chemistry, and at roughly 68 percent carnitine by weight 3 g of LCLT delivers about 2 g of carnitine, which people consistently miss when comparing doses across studies. Injectable levocarnitine gives essentially complete bioavailability with zero colonic exposure and no TMAO route, clears renally fast, fills the plasma and extracellular compartment, will not load muscle without concurrent insulin, and is an established therapy for primary and secondary carnitine deficiency and for dialysis patients, which tells you exactly which compartment it was designed to serve. Acetyl-L-carnitine crosses the blood-brain barrier far better and delivers an acetyl group into the acetyl-CoA and acetylcholine pools, making it the neural tool with its strongest human evidence in peripheral neuropathy and weaker mixed results in cognition and mood, so using it for fat loss is taking a brain-licensed truck to a warehouse job. Glycine propionyl-L-carnitine is worth flagging by name because people confuse it with alpha-GPC, which is a completely different compound, a choline donor, and choline is itself a TMA substrate, arguably a more efficient one than carnitine, so anyone making stack decisions on TMAO should be scrutinizing alpha-GPC harder than carnitine and almost nobody does. GPLC proper carries a propionyl group that converts to succinyl-CoA and enters the Krebs cycle directly, making it anaplerotic, so it refills the cycle instead of only shuttling fuel into it, and it holds the most substantial vascular literature of any carnitine form, particularly in peripheral arterial disease and claudication, with nitric oxide and blood flow effects. Simple mapping: oral carnitine or LCLT with carbohydrate over months if you are loading muscle, LCLT for training recovery and damage control, injectable levocarnitine for plasma buffering and clinical repletion, ALCAR for brain and nerve, GPLC or propionyl-L-carnitine for perfusion and ischemic tissue. On testing, labs confirm rather than diagnose, so I would not order a carnitine panel to decide whether carnitine is interesting, only when the result changes what I do, and two earn their place: free carnitine plus the acylcarnitine profile including the acyl-to-free ratio, where above roughly 0.25 suggests acyl groups are outrunning the buffer and there is genuinely something for carnitine to act on, and a cystatin C-based eGFR, which is there for the TMAO question rather than the carnitine one, because if clearance is intact that conversation is largely academic. Everything past those two is confirmatory. So where I land: overhyped as sold, yes, and as a fat burner in a healthy, replete athlete it is close to inert, with the fasted injectable version among the least mechanistically supported uses of anything in this industry. Underrated in specific terrain, also yes, because in hemolytic states, renal wasting, chelation, high-output cardiac demand, incomplete beta-oxidation and clinical deficiency it stops being a supplement and becomes repletion of something the body genuinely cannot keep up with. Your NAD+ comparison is apt with one refinement: both are cases where the mechanism is airtight and the supplementation logic does not automatically follow from it, but carnitine has something NAD+ precursors mostly lack, a real and measurable deficiency state with a real and measurable readout, which makes it far easier to be honest about, because you can find out whether the person needs it instead of arguing about whether they might. Find the terrain that makes the molecule necessary, or leave the molecule on the shelf.