Horizon Robotics earnings: Don't focus on net profit yet

As someone who has long tracked the autonomous driving supply chain and automotive electronics financial reports, I tend to break down Horizon Robotics' 2026 H1 report into three questions: where does the money come from, where does the profit come from, and is the autonomous driving business actually making money? This report is worth twenty minutes of careful reading, focusing on separating book numbers from operational results. Book profitability means the last line of the report is positive; operational profitability means earning back money from selling chips, creating solutions, and delivering services themselves. For autonomous driving suppliers, the latter determines whether there are resources left for R&D, capacity expansion, and after-sales support.

To read this report, grab three numbers first: net profit of 3.784 billion yuan, fair value change of CARIAD convertible bonds of 5.241 billion yuan, and comprehensive gross margin of 66%. These three numbers cannot be read together indiscriminately. Net profit is the final book result; fair value change is accounting income brought by asset price changes, not necessarily from daily operations; gross margin is the ratio after deducting direct costs from product or solution revenue, reflecting whether the business itself has gross profit space, but it doesn't equal cash flow, nor does it equal net profit.

These days, I've been testing AI assistants for summarizing financial reports. They can list the numbers, but they easily mix up accounting income and operating income. Key footnotes still require manual verification. Open the interim report or financial news page, locate net profit first, then search for CARIAD convertible bond fair value changes, and finally check the comprehensive gross margin. If opening a PDF, don't flip through the main text; use search to jump near the corresponding tables to confirm amounts, units, and standards. This step requires judging whether this income belongs to recurring operating profits.

I prefer compressing the judgment into three lines. Net profit is 3.784 billion yuan. Among this, there is a 5.241 billion yuan gain from CARIAD convertible bond fair value changes. Then look at whether the autonomous driving chip and solution businesses have covered this gain. If the third line lacks sufficient evidence, this report looks more like a first-time book turnaround; whether the main business has turned profitable needs separate examination.

This is easy to misread. Seeing a two-year turnaround target leads people to assume it's already achieved. Targets must be verified against revenue, gross profit, expenses, impairments, and one-off gains to see the realization path. Treating the three major businesses as a safety cushion also requires seeing whether they contribute revenue, gross profit, or capital gains. With Horizon's numbers laid out, the fair value change gain exceeds net profit, so operational pressure still needs attention. A comprehensive gross margin of 66% indicates that the products or solutions themselves have a decent gross profit foundation. To convert gross profit into sustained operating profit, autonomous driving suppliers face R&D investment, customer project delivery, after-sales support, and supply chain costs eating into the book gross profit.

Horizon Robotics' 2026 H1 report lays bare the profit structure of autonomous driving suppliers. Book profits can come from asset revaluation, but operating profits must come from the continuous delivery of chips, algorithms, solutions, and services. For readers, the next step isn't to rush to judge who wins, but to find an autonomous driving company and re-check net profit, fair value changes, and gross margins using the same standards, clarifying first which money is being earned.