TMTPOST -- The Hong Kong-listed AI and robotics company reported adjusted net loss RMB 60 million, narrowing by 32.1% year on year.

Its revenue was RMB1.28 billion, up 25.3% year on year; newly signed orders reached RMB 2.38 billion, up 35.5%; and gross margin rose from 35.1% to 35.8%, according to the Interim Financial Report released in late August.

Losses on the books widened, with net loss expanding from RMB 50 million in the same period last year to RMB 180 million.

In July 2025, Geek+ listed on the Hong Kong Stock Exchange as the “world’s first AMR (Autonomous Mobile Robot, mainly used for handling and picking in warehouses) pure-play,” pricing its IPO at HK$16.8. Six months later, the share price briefly surged to HK$33.4—nearly double the IPO price. It then trended downward, and by the end of August this year it was down more than 39% from the IPO price, almost a 70% pullback from its peak earlier in the year. Its market capitalization correspondingly fell from about HK$43 billion at the peak to HK$13.4 billion.

Over the past year, Geek+ also went looking for new growth engines. It successively launched an unmanned picking workstation, the general-purpose warehouse humanoid robot Gino 1, and the Gravity embodied-intelligence framework, and it has tried to fold AMRs, robotic arms, and humanoid robots into a single end-to-end unmanned-warehouse solution.

But for now, the business that still underpins revenue is the warehouse robotics segment it has been building for a decade. Although embodied intelligence has already won orders, the company has not disclosed specific figures for contract value, shipment volume, or revenue contribution—meaning it still has some way to go before it can become a true second growth curve.

The core business is growing, the new story is being told—so why can’t the share price get moving?

The Core Business Didn’t Crack

Looking back over the past year, Geek+ has sent out a series of signals.

In February, it unveiled the humanoid robot Gino 1, and the stock jumped more than 11% into the close that day; in the same month, it was included in the Stock Connect program. At the end of March, it released its 2025 annual report, with adjusted net profit turning positive for the first time on a full-year basis. In June, Geek+ also announced a share buyback plan of up to HK$2 billion.

But rallies driven by these signals didn’t last long. As of August 28, Geek+ closed at HK$10.2, already below its IPO price of HK$16.8. Compared with the peak of HK$33.4 in January this year, the stock had pulled back by nearly 70%.

Judging solely by the stock chart, it’s easy to suspect something has gone wrong with Geek+’s business. But based on its interim report, its core business was still growing, though profitability wasn’t as steady as the pace of revenue growth might suggest.

In the first half of 2026, Geek+ posted revenue of RMB 1.28 billion, up 25.3% year over year; gross profit was RMB 460 million, up 27.8% year over year—slightly faster than revenue growth—lifting gross margin from 35.1% to 35.8%.

Geek+’s revenue mainly breaks down into two segments: warehouse fulfillment and industrial handling.

Warehouse fulfillment is the core pillar. This segment primarily serves customers in retail, e-commerce, and logistics, using AMRs to automate warehouse operations such as moving goods and order picking. The financial report shows that in the first half of the year, revenue from this business reached RMB 1.20 billion, up 24.8% year over year, accounting for 93.6% of total revenue.

Industrial handling is much smaller in scale, mainly serving manufacturing customers in sectors such as automotive, new energy, and electronics. It generated RMB 80 million in revenue in the first half, representing 6.3% of the total and growing 34.4% year over year. Notably, for full-year 2025, this business had declined 34.6% year over year, and it returned to growth in the first half of this year.

It’s clear that Geek+ still relies heavily on warehouse AMRs, and its revenue mix hasn’t changed materially.The largest core segment kept growing, and the previously contracting industrial handling business returned to growth, though it remains small for now.

By region, overseas markets not only contributed the majority of revenue but also delivered higher gross margins.

In the first half, revenue from outside mainland China exceeded 75%. Overseas sales gross margin reached 46.2%, more than 10 percentage points higher than the overall gross margin of 35.8%. Overseas operations not only generated most of the revenue, but also served as a key pillar supporting the company’s current gross-margin level.

The customer base also remained fairly stable. As of the end of June, Geek+ had served more than 1,000 end customers, maintained a repeat-purchase rate above 80%, and delivered a cumulative total of more than 81,000 robots.

Orders are a better indicator of follow-on demand. In the first half of the year, Geek+ secured RMB 2.38 billion in new orders, up 35.5% year on year—nearly 1.9 times its revenue over the same period. Among them, orders for pallet-to-person solutions rose by more than 200% year on year, while orders from production and manufacturing scenarios surged by more than 600%.

When orders are growing faster than revenue, it suggests that front-end demand is still expanding, providing some support for revenue growth down the road.

With both revenue and orders rising, why did it still lose money in the first half of the year?

The main drag on reported profit was the exchange rate.In the first half of 2026, Geek+ recorded RMB 100 million in foreign-exchange losses, whereas it posted RMB 80 million in foreign-exchange gains over the same period last year—a swing of more than RMB 180 million. Excluding FX gains/losses and share-based payments, adjusted net loss came in at RMB 60 million, narrowing by 32.1% year on year.

But FX wasn’t the whole story. In the first half, Geek+ posted gross profit of RMB 460 million, selling and marketing expenses of RMB 277 million, and R&D expenses of RMB 188 million. Those two expense lines alone totaled RMB 465 million—already exceeding total gross profit for the same period; add nearly RMB 100 million in administrative expenses, and the core business is still some distance away from consistently turning a profit.

In 2025, Geek+ had just achieved adjusted net profit of RMB 40 million, delivering full-year adjusted profitability for the first time, only to slip back into an adjusted loss in the first half of this year.

Cash flow is another signal worth watching. In the first half, Geek+ recorded net cash outflow from operating activities of RMB 290 million, compared with an outflow of RMB 110 million a year earlier—an obvious expansion in cash burn. Over the same period, accounts receivable and notes increased from RMB 960 million at the end of 2025 to RMB 1.11 billion, inventories rose from RMB 800 million to RMB 870 million, and prepayments and other receivables climbed from RMB 240 million to RMB 390 million.

This is tied to the nature of project-based business: the more orders you take, the more cash gets tied up in projects and working capital. With order growth in the first half, receivables, inventories, and prepayments all increased, ultimately showing up as pressure on operating cash flow. Going forward, in addition to whether orders can keep growing, it will also be crucial to see whether delivery and collections can keep pace.

Overall, this interim report isn’t bad. Revenue, orders, and gross margin all improved, and the adjusted loss continued to narrow; the pressure is concentrated on earnings stability and cash flow.

Judging by the fact that the stock has fallen below its issue price, the capital market is no longer looking only at whether the AMR business can keep growing, but is demanding that Geek+ answer two more specific questions: Can the existing business deliver profits consistently, and when will new businesses start contributing revenue?

Humanoid Robots Not Yet in Sizable Deliveries

Geek+ is betting its second growth curve on embodied intelligence.

It moved very quickly along this path. In July 2025, it set up an embodied intelligence subsidiary; a month later, it released the Geek+ Brain embodied foundation model for warehouse scenarios and a general-purpose robotic-arm operation solution; in October it launched an unmanned picking workstation, and within three months passed POC acceptance by a Fortune Global 500 customer. By February this year, it introduced the general-purpose humanoid robot Gino 1, and in July it released the Gravity framework and the Gravity 4D model.

Geek+ wants to connect the mobile, picking, and handling equipment that used to be scattered across warehouses: AMRs handle mobility; the unmanned picking workstation handles picking at fixed stations; Gino 1 covers more steps that still require human operation. All of it is then orchestrated by the same embodied intelligence system—piecing together an end-to-end lights-out warehouse.

The biggest advantage of this route is that Geek+ doesn’t lack real-world deployment scenarios: new products can plug directly into its existing customer base.

The interim report disclosed that the unmanned picking workstation and Gino 1 have already reached partnerships with multiple Fortune Global 500 companies.

The question is: how far has Geek+ actually come so far? Judging from the interim report, there are still relatively few quantifiable commercial results. The financial statements only disclosed that embodied intelligence “achieved a breakthrough in order wins,” but did not release specific figures for order value, shipment volume, or revenue contribution from the related products.

The spending has shown up in the financials first. In the first half of 2026, Geek+’s R&D expenses were RMB 190 million, of which RMB 40 million went directly to embodied intelligence—nearly a quarter of total R&D spend.

Geek+ even did the math in the interim report: excluding this embodied intelligence R&D investment, the adjusted loss would have been only RMB 20 million, an 81.9% improvement year on year.

These numbers say a lot about where Geek+ is right now. Its existing business is already capable of making money, but selling, R&D, and other expenses are still high. On top of that, embodied intelligence is still in the investment phase, further increasing cost pressure. This creates an awkward valuation situation: it can’t yet command a humanoid-robot valuation, while the profitability of its AMR business is being diluted by investment in the new business.

In a media interview in July, Geek+ co-founder Chen Xi said that Gino 1 was planned to enter small-batch production and delivery in the third quarter of this year, and to begin POC programs with customers in China and overseas. In other words, Geek+’s humanoid robot is still at the stage of moving from product validation to commercial delivery.

After more than two years of the humanoid-robot boom, capital markets have become reluctant to pay simply for product launches. Investors care more about whether launches can translate into volume orders, whether robots can be delivered reliably, and ultimately how much revenue they can generate.

The commercialization of embodied intelligence has yet to materialize, which can only explain part of why Geek+’s valuation has come under pressure. The share-price pullback over the past year was compounded by shareholder sell-downs and a shift in industry sentiment. In January this year, cornerstone investors came off lock-up; in July, Warburg Pincus—Geek+’s largest institutional shareholder for nine years—sold 88.61 million shares, cutting its H-share stake from 13.4% to 4.8% and ceasing to be a major shareholder. With such a large block of shares hitting the market, the stock also faced additional near-term supply pressure.

Turning back to Geek+ itself, the company has said it expects cumulative shipments of more than 10,000 embodied-intelligence units over the next three years. What comes next is whether existing orders can be converted into scaled deliveries—and ultimately into steady revenue.