Sugar-Free Chocolate Hazelnut Spread Recipe

Commercial chocolate-hazelnut spread is one of the most successful desserts ever marketed as a breakfast food. Read the label and the first ingredient is usually sugar, with palm oil close behind — a typical jar runs about 56 percent sugar by weight, which means two tablespoons deliver roughly 21 grams of sugar before you have eaten anything else. The good news is that the thing people actually love about it — roasted hazelnut and cocoa — has nothing to do with the sugar. Make it at home and you can keep the flavor, lose almost all the glycemic carbohydrate, and end up with a spread whose calories come from real nuts instead of refined sweetener.

This post covers the sweetener rules that keep a homemade spread smooth, a recipe with honest per-serving macros, and how to log it so a daily spoonful does not quietly drift past your calorie target.

Which Sweeteners to Use Here

A spread has no water in it, so your sweetener has to blend into fat without grit. That rules out granular crystals unless you powder them first.

• Powdered erythritol is the default. It is glycemic-inert, dissolves into warm nut oil, and is subtracted fully when you count net carbs. Powdered, not granular — granular stays sandy.
• Monk fruit/erythritol blends work the same way and let you use less volume because they are more intensely sweet.
• Allulose gives the smoothest, most pourable result and resists the slight cooling “mouth-chill” some people notice with erythritol. It browns faster, so add it after the nuts are blended, off the heat.

Two warnings. First, the “natural” sweeteners that get suggested for spreads — jaggery, honey, maple syrup, and date paste — are still sugar. They raise blood glucose the same way table sugar does; date paste is the worst offender because it is concentrated fruit sugar marketed as a health food. Second, avoid maltitol, the sweetener in many commercial “no sugar added” chocolate spreads. Maltitol has a real glucose effect and is only partially inert, so it cannot be subtracted the way erythritol can. The difference between these sugar alcohols is exactly what our sugar alcohols gray-zone guide walks through, and it matters more here than in most recipes because a spread is eaten in small daily doses that add up.

The Recipe: Sugar-Free Chocolate Hazelnut Spread

This makes a jar of about sixteen tablespoons. The only real technique is patience: hazelnuts have to blend long enough to turn from chopped nuts into liquid butter before anything else goes in.

Makes ~16 tablespoons (8 servings of 2 tbsp).

Ingredients

• 200 g raw hazelnuts (about 1.5 cups)
• 30 g unsweetened cocoa powder (about 4 tbsp)
• 40 g powdered erythritol or monk-fruit blend
• 30 g cocoa butter or neutral oil, melted (about 2 tbsp)
• 1 tsp vanilla extract
• pinch of salt

Per serving (2 tbsp): ~185 kcal, 2 g net carbs, 4 g protein, 17 g fat.

Method

1. Roast the hazelnuts at 175°C for 10–12 minutes until fragrant and the skins darken. Tip them into a clean towel and rub off as much skin as comes loose — the skins are bitter.
2. Add the warm nuts to a food processor and run. They will go from chopped, to clumpy meal, to a stiff paste, and finally to flowing butter. This takes 5–8 minutes; scrape down the sides every minute. Do not rush it — the smooth texture depends on the nuts fully releasing their oil.
3. With the hazelnuts liquid, add the cocoa powder, powdered sweetener, melted cocoa butter, vanilla, and salt. Blend until glossy and uniform.
4. Taste and adjust sweetener. Blend in another spoon of oil if you want it more pourable.
5. Transfer to a clean jar. It firms up in the fridge; let it sit out a few minutes before spreading. Keeps about three weeks refrigerated.

A Quick Comparison

Two tablespoons of the commercial jar run about 200 kcal with 21 g sugar and 22 g total carbohydrate. The homemade version above is about 185 kcal with 2 g net carbs. The calories are nearly identical — fat is fat — but the glycemic load is in a different universe. That distinction between calories and blood-sugar impact is the whole point of counting net carbs rather than total carbs: you are not eating fewer calories, you are eating ones that do not spike glucose.

Variations

• Extra-dark: push the cocoa to 40 g and add a little more sweetener to balance the bitterness.
• Salted: double the salt and finish with flaky salt on toast.
• Runnier drizzle: use allulose and an extra tablespoon of oil for a sauce you can spoon over berries or full-fat yogurt.

How to Log This in CalEye

For a one-off taste, a photo log is fine: snap the spread on your plate and CalEye estimates calories and macros from the image. But a spread is a repeatable, weighed homemade recipe eaten in small daily portions, which is the exact case where a photo struggles — almost all the calories are hazelnut fat, and a camera cannot see how much oil is in a spoonful.

The accurate approach is to build it once in My Recipes. Enter each ingredient by weight, set the yield to 16 tablespoons (or 8 two-tablespoon servings), and CalEye divides the totals into a clean per-serving entry. From then on, logging is one tap per spoonful. CalEye applies net-carb logic automatically: erythritol is subtracted fully, while xylitol or maltitol would only be counted at about half, so the number you log reflects the real glycemic carbohydrate rather than the raw label total.

One habit worth keeping: log the spread by the measured tablespoon, not by eye. Nut spreads are calorie-dense, and the difference between a level spoon and a heaped one can be 80 calories. If you ever scale the batch up or down, our recipe scaling guide covers how the per-serving numbers move so your logs stay honest.

References

1. U.S. Department of Agriculture, Agricultural Research Service. FoodData Central. Washington, DC: USDA, 2024.
2. American Diabetes Association. “Facts About Sugar and Sugar Substitutes.” Diabetes Care / ADA Consumer Resources, 2023.
3. Mattes RD, Dreher ML. “Nuts and healthy body weight maintenance mechanisms.” Asia Pacific Journal of Clinical Nutrition, vol. 19, no. 1, 2010, pp. 137–141.