The Information Machine

Chinese AI Models and Products Gain Structural Ground on US Rivals · history

Version 19

2026-07-08 18:19 UTC · 323 items

What

Chinese AI providers have gained structural ground on US rivals across pricing, model performance, enterprise adoption, open-source download share, and hardware. Chinese models account for 1.15B tracked open-source downloads vs 723M for US models [1], with pricing as low as $0.18 per million tokens [3]. At the hardware layer, CXMT has entered the global RAM market [7], secured a reported $3B memory deal with Tencent [8], and Huawei is marketing Ascend AI accelerators across South Korea, the Middle East, and Southeast Asia [11][12]. The most significant new development: DeepSeek has reportedly been working on its own AI inference chip for approximately one year, hiring engineers and meeting with hardware partners, driven by US export controls [15][16].

Why it matters

DeepSeek's chip development effort, if it succeeds, removes a key point of leverage in US export policy at the same moment Chinese models are winning the cost-performance tier where most enterprise work runs. The combination of software gains (download share, enterprise migration, pricing) and hardware moves (CXMT global RAM, Huawei's international Ascend sales, DeepSeek's chip ambitions) reflects vertical integration across the AI stack rather than isolated wins at a single layer.

Open questions

  • DeepSeek has reportedly spent approximately one year on AI inference chip development, hiring engineers and meeting with hardware partners [15][16]; how close is it to a functional design, and can a model company replicate the organizational depth required for silicon?

  • Huawei is seeking Ascend AI chip customers in the Middle East and Southeast Asia — many of them US security partners — in addition to planned South Korea sales in Q4 2026 [11][12]; what is the US response to allied purchases of Chinese AI accelerators?

  • Apple is seeking US government approval to source memory from blacklisted CXMT for China-market devices [9][10], while CXMT has simultaneously secured a $3B deal with Tencent [8]; will the US grant Apple's request, and does CXMT's domestic scale change the calculus?

  • Eric Schmidt, ITIF, Jensen Huang, and Aravind Srinivas all argue export controls have not contained China's AI development [18][21][19][20]; does DeepSeek's move into chip design add concrete weight to that argument, and what policy revision follows?

Narrative

Chinese AI providers have taken positions across the model, pricing, enterprise adoption, and hardware layers of the AI stack over roughly 18 months. Academic research confirms that China surpassed the US in open-source model downloads in summer 2025; by March 2026, Chinese models had accumulated 1.15 billion tracked downloads against 723 million for US models [1]. The primary driver is Qwen, which dominates through a broad range of useful sizes — especially small, cheap-to-run models — rather than a single flagship. Tencent's Hy3 (295B total parameters, 21B active, MIT license) adds another entry, claiming to rival open-source models with two to five times more parameters [2]. At the pricing layer, Chinese models cost as low as $0.18 per million tokens against a $4 frontier average [3], and 60% of enterprises monitoring AI budgets have migrated to cheaper or open-source Chinese models [4]. Coinbase CEO Brian Armstrong disclosed his company is experimenting with Chinese open-weight models as defaults for execution tasks through its internal LLM gateway [5].

At the hardware layer, China reclaimed the world's top-ranked supercomputer in June 2026, built without US GPUs [6]. CXMT, approaching China's largest semiconductor IPO, entered the global RAM market at $138 module pricing [7] and secured a reported $3 billion memory supply deal with Tencent [8]. Apple is simultaneously in talks to source memory from CXMT and YMTC for China-market devices while seeking US government approval to buy from blacklisted CXMT [9][10]. Huawei is marketing Ascend AI accelerators internationally: it plans South Korea sales in Q4 2026 and is separately seeking customers across the Middle East and Southeast Asia [11][12]. Meituan's LongCat-2.0 (1.6 trillion parameters) became the first large Chinese model trained from scratch on domestic Huawei hardware [13], and DeepSeek V4 has been evaluated on Huawei Ascend NPUs [14], indicating the Ascend software ecosystem is gaining traction for frontier inference.

The most concrete new development is DeepSeek's reported move into chip design. Three sources cited by Reuters say DeepSeek has been working on AI inference chip development for approximately one year, meeting with hardware and silicon partners and actively hiring engineers, with US export controls as the primary stated driver [15][16]. This mirrors a similar move by OpenAI and, if successful, would reduce a key point of leverage in US export policy. In parallel, China pitched open-source AI as a development tool at the UN AI for Good summit, positioning itself as the developing world's preferred AI partner [16].

The debate over US export controls has accumulated substantial evidence running against the policy's stated aims. A research study found Chinese developer activity on open LLM projects grew significantly more than US activity after major export controls [17]. Eric Schmidt — whose SCSP think tank helped design the export control architecture — publicly acknowledged the controls have not achieved their intended effect [18]. Nvidia CEO Jensen Huang argues controls function as industrial stimulus for Huawei [19]; Perplexity CEO Aravind Srinivas argues controls compressed the frontier gap to approximately 12 months while forcing China to build superior domestic infrastructure [20]; ITIF argues overly stringent controls are eroding US AI leadership [21]. Anthropic CEO Dario Amodei holds the opposing view, calling restriction clearly in the US national security interest and dismissing counterarguments as 'fishy' [22]. DeepSeek's chip development effort, if confirmed at scale, extends the practical argument that export controls are accelerating Chinese vertical integration rather than preventing it.

Timeline

  • 2026-06-16: DeepSeek raises $7.4B at a $50B valuation; The Information later reports a preview of Anthropic's Mythos model prompted the fundraise to remain competitive. [34][30]
  • 2026-06-16: Z.ai releases GLM-5.2 (753B parameters, MIT license), topping the Artificial Analysis Intelligence Index v4.1. [35]
  • 2026-06-23: SemiAnalysis reports CXMT is approaching China's largest semiconductor IPO with DRAM technology scaled toward the 10nm class. [27][28]
  • 2026-06-26: UBS survey finds 60% of enterprises monitoring AI budgets migrating to cheaper or open-source Chinese models. [4]
  • 2026-06-27: J.P. Morgan report: Chinese models up to 50x cheaper per token; Chinese firms held over 45% of OpenRouter traffic by April 2026, up from under 2% in late 2024. [36]
  • 2026-06-28: China reclaims the world's top-ranked supercomputer for the first time since 2018, built without US GPUs. [6][37]
  • 2026-06-29: Apple in talks to source memory from CXMT and YMTC for China-market devices; separately seeks US government approval to purchase from blacklisted CXMT. [9][10]
  • 2026-06-29: CXMT secures a reported $3 billion memory supply deal with Tencent. [8]
  • 2026-06-30: Meituan's LongCat-2.0 (1.6T parameters) confirmed trained from scratch on 50,000 Huawei Atlas-950 chips — the first large Chinese model to use domestic hardware for both pre-training and inference. [13]
  • 2026-06-30: Citibank Research: Chinese models as cheap as $0.18 per million tokens vs $4 frontier average; OpenRouter open-source processing share grew from 34% in January to 65% in June 2026. [3]
  • 2026-06-30: Eric Schmidt publicly acknowledges the chip export controls his SCSP think tank helped design have not achieved their intended effect. [18]
  • 2026-06-30: Coinbase CEO Brian Armstrong discloses experimenting with Chinese open-weight models as defaults in its internal LLM gateway for execution tasks. [5]
  • 2026-07-01: Research study finds US chip restrictions accelerated China's open AI ecosystem: Chinese developer activity on open LLM projects grew more than US activity after major export controls. [17]
  • 2026-07-01: CXMT enters the global RAM market with module pricing around $138, competing with Samsung and SK Hynix. [7]
  • 2026-07-03: Huawei preparing to sell Ascend AI accelerator chips in South Korea in Q4 2026 — its first international AI hardware sales effort. [11]
  • 2026-07-04: Academic research confirms China surpassed the US in open model downloads in summer 2025; by March 2026, Chinese models had 1.15B tracked downloads vs 723M for US models, with Qwen dominant across model sizes. [1]
  • 2026-07-06: Tencent releases Hy3 (295B parameters, 21B active, MIT license), claiming to rival open-source models with 2-5x more parameters; available free on OpenRouter until July 21. [2]
  • 2026-07-06: Bloomberg reports Huawei is seeking Ascend AI chip customers across the Middle East and Southeast Asia, expanding beyond its previously reported South Korea effort. [12]
  • 2026-07-07: Reuters reports DeepSeek has been developing its own AI inference chip for approximately one year, hiring engineers and meeting with hardware partners, driven by US export controls. [15][16]

Perspectives

Jensen Huang (Nvidia)

Chip export controls don't prevent China from developing AI, function as industrial stimulus for Huawei, and cede operating-layer standards to China; the long-term risk is a world where US technology is absent from systems America most wants to influence.

Evolution: Consistent.

[19]

Aravind Srinivas (Perplexity)

Export controls are the primary reason a roughly 12-month frontier gap exists, but by forcing China to build domestic infrastructure they are converting China into a more capable long-run competitor at the physical layer.

Evolution: Consistent.

[23][24][20]

Dario Amodei (Anthropic)

Restricting China's AI access is clearly in the US national security interest; counterarguments are 'fishy'; Anthropic has actively lobbied for AI chip export controls.

Evolution: Consistent.

[22][20]

Brian Armstrong (Coinbase)

Coinbase is experimenting with Chinese open-weight models as defaults for execution tasks, routing frontier models only to planning; frontier models are overkill for routine AI workloads at current pricing.

Evolution: Consistent.

[5]

Nathan Lambert (Interconnects)

GLM-5.2 is the first open-weight model to perform credibly as a general coding agent, comparable in significance to DeepSeek R1; it puts economic pressure on Anthropic's Claude Code revenue.

Evolution: Consistent.

[25]

Zvi Mowshowitz

GLM-5.2 is the strongest open-weight model available but is very likely heavily distilled from Claude Opus, trails frontier closed models substantially, and occupies an awkward commercial niche — not cheap enough for bulk tasks, not strong enough for the hardest ones.

Evolution: Consistent.

[26]

SemiAnalysis

Covers Chinese AI and semiconductor infrastructure with technical grounding: CXMT's DRAM traces to Qimonda documentation and is approaching a major IPO; LongCat-2.0's domestic chip pre-training is the first for a large Chinese model; DeepSeek V4 has been evaluated on Huawei Ascend NPUs.

Evolution: Consistent.

[27][28][14]

ITIF (Information Technology and Innovation Foundation)

Overly stringent US export controls are eroding American AI leadership rather than protecting it, adding a policy-research institutional voice to the Huang/Srinivas/Schmidt position.

Evolution: Consistent.

[21]

Tensions

  • Jensen Huang, Aravind Srinivas, Eric Schmidt, and ITIF argue US chip export controls have not contained China's AI development and are harming US competitiveness [19][20][18][17][21]; Dario Amodei argues restriction is clearly in the US national security interest and counterarguments are 'fishy' [22]; DeepSeek's move into chip design adds a concrete illustration of the controls-accelerate-vertical-integration argument [15]. [19][20][18][17][21][22][15]
  • Nathan Lambert argues GLM-5.2 is the first open-weight model to match closed frontier performance in coding agent harnesses [25]; Zvi Mowshowitz argues it is very likely heavily distilled from Claude Opus and trails frontier closed models substantially [26]. [25][26]
  • Enterprise migration data — UBS 60% migrating, Coinbase default routing, Citibank pricing, academic download-share research — treats Chinese model adoption as structural [4][5][3][1]; developer-community commentators argue the gains reflect spot-market price routing conditional on investor tolerance for losses [29]. [4][5][3][1][29]
  • The Information's report that Anthropic's Mythos preview prompted DeepSeek's $7.4B raise implies the US frontier still holds a lead that Chinese players are spending to close [30]; the same period's OpenRouter share data and Citibank pricing show Chinese models are winning the cost-performance tier where most enterprise work runs [3]. [30][3]
  • CXMT is entering the global RAM market at $138 module pricing, signaling competitive intent against Samsung and SK Hynix [7]; an investment analyst argues the Apple-CXMT approval headline is 'almost pure FUD' and changes nothing for memory incumbents [31]. [7][31]
  • The US government believes a banned EUV lithography tool reached China; ASML categorically denies having ever shipped one, saying it tracks all approximately 314-340 EUV units worldwide [32][33]. [32][33]

Sources

  1. [1] A paper on open model adoption and finds Chinese models, led by Qwen, now dominate. — Rohan Paul Twitter (2026-07-04)
  2. [2] tencent/Hy3 — Simon Willison (2026-07-06)
  3. [3] Reuters: Chinese models charge as little as 18 cents per million tokens versus $4 average for top models, says CitiBank … — Rohan Paul Twitter (2026-06-30)
  4. [4] UBS says 60% of companies now watching AI budgets are moving to cheaper models and open-source Chinese models — Rohan Paul Twitter (2026-06-26)
  5. [5] Coinbase CEO Brian Armstrong said Coinbase is experimenting with defaulting to Chinese open-weight models such as GLM 5.… — Rohan Paul Twitter (2026-06-30)
  6. [6] China just reclaimed the world's #1 supercomputer for the first time since 2018 — and it did it without a single NVIDIA ... — reactive:chinese-ai-competitive-rise (2026-06-28)
  7. [7] 7.2K views · 178 reactions | CHINA’S CXMT ENTERS GLOBAL RAM MARKET Chinese semiconductor firm ChangXin Memory Technologies (CXMT) has entered the global RAM market, reportedly offering memory modules priced around $138 amid higher global pricing. As China seeks semiconductor independence, CXMT’s expansion signals intensified competition in DRAM manufacturing, traditionally dominated by companies like Samsung and SK Hynix. Important Details: • CXMT entering global DRAM market • Competitive $138 pricing benchmark • Aims to challenge dominant suppliers • Supports China’s chip self-sufficiency • Expands semiconductor supply chain • Impacts global memory pricing • Strategic tech competition factor Memory chips are critical for AI systems, data centers, and consumer electronics. The semiconductor battleground is no longer just about processors — memory now shapes global tech power. #Unboxfactory #SemiconductorRace #ChinaTech #GlobalMarkets #RAMIndustry | Unbox Factory — reactive:chinese-ai-competitive-rise
  8. [8] China's CXMT wins $3 billion memory supply deal with Tencent ... — reactive:chinese-ai-competitive-rise
  9. [9] Apple is reportedly seeking U.S. approval to buy memory chips from China's blacklisted CXMT as the AI boom sends DRAM pr... — reactive:chinese-ai-competitive-rise (2026-06-29)
  10. [10] Apple is in talks to source memory chips from China’s CXMT and YMTC for China-market devices, putting $AAPL at the cente... — reactive:chinese-ai-competitive-rise (2026-07-02)
  11. [11] Huawei is preparing to sell its Ascend AI accelerator chips in South Korea in Q4 2026 — the first time it has attempted ... — reactive:chinese-ai-competitive-rise (2026-07-03)
  12. [12] Jukan @ ICML on X: "Report: Huawei Seeks AI Chip Customers in Middle East, Southeast Asia- BBG Huawei Technologies is endeavoring to establish a foothold in the AI chip markets of the Middle East and Southeast Asia, areas currently dominated by Nvidia Corp. These efforts are ongoing despite https://t.co/0IrpdJH1K9" / X — reactive:chinese-ai-competitive-rise
  13. [13] 🇨🇳China claims a new milestone in locally trained AI, as Meituan rolls out LongCat-2.0. — Rohan Paul Twitter (2026-06-30)
  14. [14] This week the InferenceX team discusses what it took to get DeepSeek V4 on InferenceX, changes in the model architecture… — SemiAnalysis Twitter (2026-07-01)
  15. [15] Facing US export controls, China's DeepSeek plans to make its own chips — Ars Technica AI (2026-07-07)
  16. [16] 🟡 The sobering of AI — Semafor Technology (2026-07-08)
  17. [17] U.S. chip restrictions helped push China to build and spread open AI models. — Rohan Paul Twitter (2026-07-01)
  18. [18] Eric Schmidt — whose SCSP think tank helped design the US chip export control architecture — just publicly admitted at h... — reactive:chinese-ai-competitive-rise (2026-06-27)
  19. [19] Jensen Huang explains how blocking China from Nvidia does not mean blocking China from AI. — Rohan Paul Twitter (2026-06-29)
  20. [20] Aravind Srinivas just explained why China’s open-source AI may become more powerful than ever. — Rohan Paul Twitter (2026-06-30)
  21. [21] Overly Stringent Export Controls Chip Away at American AI Leadership | Blogs | May 5, 2025 | ITIF — reactive:chinese-ai-competitive-rise
  22. [22] Dario Amodei has a really hardline view that China shouldn’t have strong AI. — Rohan Paul Twitter (2026-06-29)
  23. [23] Opinion from a former Meta PM. — Rohan Paul Twitter (2026-06-30)
  24. [24] AI at scale is constrained by physical inputs, and China has more slack in electricity plus dominant control over severa… — Rohan Paul Twitter (2026-06-30)
  25. [25] GLM-5.2 is the step change for open agents — Interconnects (2026-06-22)
  26. [26] GLM-5.2 Is The New Best Open Model — Zvi's AI Roundups (2026-06-22)
  27. [27] China’s CXMT Is Set to Challenge DRAM Incumbents — SemiAnalysis Twitter (2026-06-23)
  28. [28] CXMT (ChangXin Memory Technologies), China’s top domestic DRAM maker, is preparing for a major IPO on Shanghai’s STAR Ma... — reactive:chinese-ai-competitive-rise (2026-06-23)
  29. [29] The AI model market is turning into an inference spot market. Developers are routing to whatever model gives the best co... — reactive:chinese-ai-competitive-rise (2026-06-08)
  30. [30] The Information reports that Anthropic’s Mythos preview spooked DeepSeek into fundraising. — Rohan Paul Twitter (2026-06-26)
  31. [31] Memory investors, don't panic the Apple-CXMT headline is almost pure FUD and here is why it changes nothing for memory s... — reactive:chinese-ai-competitive-rise (2026-07-01)
  32. [32] ASML just became the center of a US-China chip fight after Washington said it fears a banned EUV lithography tool may ha… — Rohan Paul Twitter (2026-06-19)
  33. [33] ASML denies US government report that its EUV chipmaking tool ... — reactive:chinese-ai-competitive-rise
  34. [34] DeepSeek takes the crown as China’s most valuable AI startup after a massive $7.4B raise at a $50B valuation. — Rohan Paul Twitter (2026-06-16)
  35. [35] GLM-5.2 is probably the most powerful text-only open weights LLM — Simon Willison (2026-06-17)
  36. [36] 🇨🇳🇺🇸Chinese AI models are up to 50 times cheaper than their American counterparts on a per-token basis. — Rohan Paul Twitter (2026-06-27)
  37. [37] The world's fastest supercomputer just went live in Shenzhen — and it contains zero US GPUs. — reactive:chinese-ai-competitive-rise (2026-06-28)