The reported profit is not the operating result
Horizon Robotics reported 3.784 billion Chinese yuan of profit for the first half of 2026, compared with a 5.233 billion yuan loss a year earlier. The headline sits beside a different operating picture in the same filing: continuing-operation revenue increased 32.9 percent to 2.055 billion yuan and gross profit rose by the same rate to 1.356 billion yuan, but the operating loss widened 11.1 percent to 1.672 billion yuan. Its adjusted net loss, which removes specified non-cash and one-off items, widened 25.4 percent to 1.671 billion yuan. The company is listed in Hong Kong under stock code 9660.[1,2,3]
The reconciliation changes the meaning of the profit number. Horizon recorded a 5.241 billion yuan fair-value gain on financial liabilities and a 2.779 billion yuan gain from deconsolidating D-Robotics, while also recording a 2.074 billion yuan loss from investments accounted for using the equity method. Those items do not make the revenue gain unimportant, but they mean the reported profit cannot by itself show that the underlying assisted-driving business has reached operating profitability. The more useful test is whether revenue mix, delivery margins and cash conversion improve together.[1,3]
Licensing carried the blended margin
Horizon’s revenue mix shifted toward its higher-margin line. License and services revenue increased 52.7 percent to 1.129 billion yuan and accounted for 55 percent of continuing-operation revenue, up from 47.8 percent a year earlier. Product solutions increased 14.8 percent to 925.6 million yuan and fell to 45 percent of revenue. The filing explicitly attributes the unchanged 66 percent blended gross margin to the larger contribution from license and services despite a lower product-solution margin. Macrostream’s independent review reaches the same distinction: licensing revenue was recognized ahead of many vehicle-volume ramps.[1,3]
That distinction matters for automotive-compute economics. License and services produced a 90.4 percent gross margin, up from 89.7 percent, while product solutions produced a 36.2 percent gross margin, down from 44.2 percent. Horizon says it supplied domain controllers and other integrated devices with only nominal markup on non-core components to support early customer adoption of its full-scenario driving solution. It says the comparable product margin would have been 48.1 percent excluding that effect. The filing supports a stronger licensing model, but it does not yet prove that the hardware-and-delivery line will retain an improved margin once those customer programs scale.[1,3]
Receivables and financing remain the commercial test
The balance sheet adds a second constraint. Cash and cash equivalents fell 26.4 percent from 20.188 billion yuan at December 31 to 14.866 billion yuan at June 30. Net trade receivables increased from 1.760 billion yuan to 2.430 billion yuan, while the credit-loss allowance on current trade receivables rose from 102.1 million yuan to 177.7 million yuan. Horizon says it offered preferential credit terms to selected early-stage customers, which lengthened receivable aging and raised the allowance. That is not evidence of a default; it is evidence that top-line growth has not yet become a simple cash-collection story.[1]
After the reporting period, Horizon paid CARIAD about 398.9 million US dollars in cash and issued 1.302 billion Class B shares under an amended convertible-loan arrangement, completed on August 3. It also issued 450 million US dollars of zero-coupon convertible bonds on July 29; the filing says the accounting impact was still being assessed. The transactions reduced the potential CARIAD dilution described by Horizon, but they also make the next set of cash, debt and share-count disclosures important. A Hong Kong profit headline therefore tells only part of the capital story.[1]
What would change the conclusion
Horizon has disclosed product milestones that could test the model soon: it says Journey 6B design wins exceed 10 million units, with mass production expected as early as the third quarter of 2026, and it expects cockpit-driving integration to enter mass production in the fourth quarter. The decision delta is not that licensing growth is weak; it is that its high margin currently masks a less favorable product margin and a wider core loss. The next filing needs to show whether product margins hold without nominal-markup controller bundles, whether receivables convert into cash, and how the August financing changes the balance sheet. None of those checkpoints is a recommendation to trade the shares.[1,3]