Moore Threads Hong Kong Listing: Domestic GPU Liquidity Arbitrage and Dollar Fault Lines

01 Trigger Event

Bloomberg reported on August 9 that Moore Threads Technology Co., just one year after its Shanghai STAR Market debut and a cumulative 420% share price gain, has announced plans for a Hong Kong secondary listing at an "appropriate time." This is not an isolated case—over the past 12 months, Cambricon, Hygon Information, and SMIC have all seen their A-share valuation anchors rewritten. Moore Threads joining the Hong Kong queue signals that this narrative has officially entered its second phase.

02 What This Really Means

On the surface, it looks like "another funding round." But what is actually happening is that the capital structures of domestic AI chip vendors are beginning to physically decouple from the dollar system.

Moore Threads' IPO valuation on the STAR Market in 2024 was in the neighborhood of $8 billion (exact figures not verified line by line; refer to official disclosures), and a 420% gain in one year has pushed the A-share multiple to extremes. Such valuations are unthinkable on Nasdaq—peers AMD and Marvell are still anchored by forward earnings. But A-share liquidity, policy premia, and the discount rate applied to domestic substitution have produced an asset that dollar investors cannot hedge.

So the question is not "why is Moore Threads going to Hong Kong," but "why must it tap another market at all." There are three layers to the answer:

First, RMB funds can no longer absorb the entire industrial chain. Between Moore Threads, Cambricon, Hygon, Enflame, and Biren, any single company seeking to expand HBM capacity or advanced-node production requires capital in the multi-billion-dollar range. The STAR Market's carrying capacity has hit its ceiling.

Second, dollar funds can still participate. Hong Kong is a gray zone where RMB-denominated pricing coexists with US capital access. Moore Threads choosing Hong Kong over a Shanghai secondary offering actively opens a channel for global capital—something the STAR Market cannot provide.

Third, export controls have already rewritten the customer structure of domestic chip companies. Demand that might have gone to hyperscalers or overseas inference markets has been severed, redirecting toward domestic substitution procurement orders from ByteDance, Alibaba, Tencent, and telecom operators. Revenue visibility has improved, and capital markets are willing to assign higher premia. This in turn supports the Hong Kong valuation.

03 Historical Analogy / Structural Comparison

The closest inflection point is the 2014 Hong Kong standoff preceding Alibaba's US listing. Alibaba was rejected by HKEX over its partnership structure and went to NYSE, creating the largest IPO at the time. That episode was about "Chinese internet companies needing a dollar valuation anchor."

Today the direction is reversed: Chinese hard-tech companies need a de-dollarized valuation anchor, and Hong Kong is the default option. The difference: Alibaba's round was dollar capital chasing Chinese growth assets; this round is financial decoupling forced by US-China technological bifurcation.

A cooler parallel is Japan's semiconductor industry in the 1980s. After the 1985 Plaza Accord, Japanese DRAM vendors faced US anti-dumping duties and tariffs. Japan's banking system (the main bank system) became the primary funding source, and Tokyo capital markets absorbed virtually all expansion capital. The result: Japanese semiconductors' global share peaked in 1986-1990—but because they were entirely detached from the dollar system, they collapsed together when the bubble burst in the 1990s.

I don't know whether Moore Threads will follow this path, but the structural resemblance is clear: when capital markets are bound to domestic currency and technology customers are bound to the domestic market, valuations open a premium window detached from global comparables. That window is both opportunity and a source of fragility.

04 What This Means for AI Builders

If your product depends on domestic GPU inference (Cambricon, Moore Threads, Hygon DCU), the implications of this capital move are:

• Domestic GPU pricing leverage will be further strengthened. With Moore Threads' Hong Kong valuation rising, its chip ASPs will not decrease, because gross margins must support the valuation narrative. This means the inference cost per FLOP on domestic GPUs will not be significantly cheaper than H100/B200 in the near term—unless you sign long-term contracts or commit to capacity.
• A byproduct of the Hong Kong listing is more transparent disclosure. HKEX inquiries on related-party transactions and customer concentration are far stricter than the STAR Market's. Over the next 12 months, expect to see Moore Threads—and potentially Cambricon's Hong Kong entity—disclose their top five customers, with procurement shares from ByteDance/Alibaba/Tencent coming to light. This is hard data for AI infra investors, more useful than any whitepaper.
• Domestic multi-GPU inference paths will accelerate. With capital secured, Moore Threads and Cambricon's next-generation products will likely focus on "H200/B200-comparable multi-node interconnect" rather than single-card performance. If your product assumes NVLink-only topology, you may need to reassess.
• The arbitrage window at the routing layer is tightening. Token gateways like opcx.ai derive core value from cross-vendor scheduling—as domestic GPU price-performance improves, scheduling space actually expands, but vendor concentration risk also rises (domestic side may shrink to only 3-4 players). This point I may be underestimating—

05 The Bear Case / Risks

I may be overstating the inflection-point significance. Three reasons:

First, the 420% rally may itself be a bubble. The 2024-2025 domestic chip rally on the STAR Market is underwritten by policy expectations and retail money, not institutional pricing. I have not seen any credible independent due diligence on Moore Threads' real gross margins, customer structure, or capacity utilization. If the Hong Kong prospectus fails to support fundamentals, this "valuation anchor" could collapse in reverse—similar to Didi's Hong Kong dual listing in 2021, which halved in value.

Second, export controls are a double-edged sword. My read is "controls force domestic substitution, and capital markets pay a premium." But the reverse: if controls tighten further to sub-7nm equipment, materials, and all EUV alternative paths get sealed off, domestic chip companies will be locked into trailing-edge nodes. At that point Hong Kong valuations would instantly flip from "domestic premium" to "residual liquidation." Cambricon already demonstrated this cycle in 2022-2023.

Third, Hong Kong's own political variable. I called Hong Kong a "gray zone," but that gray is fading. The 2024 National Security Law implementation rules and the 2025 listing rule revisions are both compressing the operational space for international capital. If the US SEC further adds Chinese-concept chip companies to the pre-delisting list (recall China Mobile and CNOOC in 2023), Hong Kong's valuation anchor could be externally severed as well. This is a macro risk I haven't stress-tested internally, but one I must hedge against.

So the final word: the real test of this wave is not at the moment of listing, but in the first annual report after the Hong Kong listing—customer concentration, gross margins, overseas business share. These three data points will determine whether this is the prelude to a "Chinese Nvidia" or the start of a "Japanese DRAM Season Two".