Manycore, the first of the Hangzhou ‘Little Dragons’ to go public, pushes ‘spatial intelligence’ as the next wave of AI development

Manycore Tech began trading in Hong Kong after seeking up to HK$1.02 billion ($130 million), becoming the first of Hangzhou’s six “Little Dragons” to reach public markets. The design AI startup reported 2024 revenue of 820 million yuan ($120 million), up 8.6% year over year, with 18.6 million yuan of operating profit but a 428 million yuan net loss. The listing highlights continued Hong Kong demand for AI IPOs and investor interest in China’s emerging “physical AI” and spatial intelligence theme despite weakness in China’s real estate market.

Analysis

The immediate beneficiaries are the infrastructure layer, not the IPO itself: GPU vendors, cloud/HPC suppliers, and the ecosystem that sells training capacity into China’s physical-AI buildout. If Manycore’s pitch resonates, it reinforces a broader capex cycle in which scarce compute becomes the gating item, which is structurally constructive for NVDA and adjacent semiconductor supply-chain names over the next 6-18 months. The more interesting second-order effect is on design software: if 3D generation and spatial tooling migrate into model-native workflows, incumbent creative suites face a slower but more durable margin squeeze as AI features become table stakes rather than monetizable add-ons.

For ADBE, the risk is less immediate share-loss and more pricing power erosion over 2-4 quarters. The market is still underestimating how quickly enterprise buyers will re-bundle design spend around lower-cost AI-native tools, especially when the core workflow can be open-sourced while the proprietary moat shifts to datasets and distribution. That combination is toxic for software incumbents with high gross margins: they can defend share with product breadth, but not necessarily defend ARPU.

The Hong Kong IPO backdrop matters because it offers Chinese AI firms a financing route that is politically cleaner than the U.S. path and potentially cheaper than private capital. That supports more listings, more secondary supply, and more cross-border investor attention on China tech names with international revenue footprints, which is modestly positive for BABA/HSBC as passive vehicles to the theme. The contrarian point is that the market may be overpricing the monetization speed of physical AI: robotics and autonomous systems are capex-heavy, slow to deploy, and still bottlenecked by integration, so today’s enthusiasm can outrun revenue conversion by several years.

Key Decisions for Investors

• Short ADBE on a 1-3 month horizon via put spreads or a modest cash short; thesis is margin compression from AI-native design workflows and slower monetization of its own AI stack. Risk/reward improves on strength, especially if management leans into bundled AI pricing.
• Long NVDA as the cleaner expression of the physical-AI capex cycle; use 6-12 month horizon and pair against a basket of software names if you want to isolate compute demand from application-layer execution risk.
• Pair trade: long BABA / short ADBE over 3-6 months. BABA offers leveraged exposure to China tech re-rating and optionality on AI infrastructure spend, while ADBE faces direct product displacement risk from the same AI wave.
• Tactical long HSBC on Hong Kong IPO flow momentum for the next 1-2 quarters; the trade is less about underwriting economics and more about custody, capital-markets fees, and regional deal flow.
• Watch for weakness in non-NVDA semis only if there is evidence the IPO pipeline is converting into actual training spend; otherwise avoid chasing the broad China AI equity basket until monetization is visible.