Technology · 10.4

Artificial Intelligence

Quantified news impact based on deduplicated, quality-reviewed events in a single 7-day window.

Average net impact+15weighted per event

Average gross impact18absolute weighted average

Events173same period

Confidence74%model average

Raw intensity62before weighting

Previous period+11net impact

Gross change+64%

Diffusion speed24.7 / day

Concentration1%

Direction disagreement17%

Source independence99%27 sources

IMPACT TREND

7-day effective impact

Positive 137 · Negative 14 · Mixed 22

+490-49

24 Jul, 180 events

+10

25 Jul, 082 events

25 Jul, 222 events

26 Jul, 121 event

+10

27 Jul, 0235 events

+12

27 Jul, 1633 events

28 Jul, 0643 events

+21

28 Jul, 2013 events

+25

29 Jul, 1016 events

+29

30 Jul, 0011 events

+21

30 Jul, 1411 events

+49

31 Jul, 046 events

Current periodPrevious period

TIME HORIZON

Impact duration

Immediate1%38 cumulative effective impact

Short term20%628 cumulative effective impact

Medium term68%2151 cumulative effective impact

Long term11%340 cumulative effective impact

CO-IMPACT

Related industries

EVENT EVIDENCE

Events ranked by effective impact

Page 7 of 17 · 163 events

Macro2

Macro27 Jul, 07:20positive

CSRC Sets H2 Reform: STAR Market Fifth Criteria to Cover AI, ChiNext to Boost New Consumption, Modern Services

The China Securities Regulatory Commission (CSRC) outlined its capital market reform roadmap for the second half of the year, focusing on enhancing equity, bond, and futures markets to serve the real economy. Key measures include extending the STAR Market's fifth set of listing criteria to artificial intelligence, advancing ChiNext reforms to support new consumption and modern services, expanding public REITs, and improving the futures market tool matrix. These steps aim to channel capital toward technology, new consumption, and services, strengthening direct financing.

Effective: +20
Intensity: 75
Confidence: 80%
Horizon: Medium term

Macro28 Jul, 14:10negative

Trump Administration to Require 30-Day Government Review of Advanced AI Models Before Release

The Trump administration is advancing a regulatory framework requiring OpenAI, Anthropic, and Google to submit their most advanced AI models for a 30-day government review before public release. The framework, derived from a June executive order, must be completed by August 1. It was triggered by Anthropic's Mythos model, which possesses offensive cybersecurity capabilities. Key debates include how to define frontier models and whether open-source models should be exempt. While compliance is nominally voluntary, export controls on Anthropic raise questions about enforceability.

Effective: -19
Intensity: 58
Confidence: 82%
Horizon: Medium term

Industries0

No evidence on this page.

Companies8

Companies27 Jul, 12:11positive

Moonshot AI Open Sources 2.8-Trillion-Parameter Kimi K3 with Over 1 Million Token Context

Moonshot AI open-sourced its flagship Kimi K3 model with 2.8 trillion parameters and a context window of approximately 1.05 million tokens, the first open-source model at the 3-trillion-parameter level. Kimi K3 achieved the largest open-source model size for nine months in a 12-month period ending July 2026. Pricing is 2 yuan (cache hit) or 20 yuan (cache miss) per million input tokens and 100 yuan per million output tokens. The model uses Kimi Delta Attention and Attention Residuals for improved information flow, offering 2.5x scaling efficiency over K2.

Effective: +20
Intensity: 75
Confidence: 95%
Horizon: Medium term

Companies28 Jul, 15:45positive

SK Hynix to Mass-Produce 1c LPDDR6 Memory in H2 2026, First Customer Xiaomi, and 512GB SOCAMM for AI Servers

SK Hynix plans to start mass production and shipment of sixth-generation 10nm-class (1c) LPDDR6 mobile memory in the second half of 2026, with Chinese brand Xiaomi among the first customers. The technology improves data transfer rates and energy efficiency. The company also plans to launch SOCAMM-form LPDDR6 memory for AI servers, offering up to 512GB per module to meet low-power, high-bandwidth, and heat dissipation needs of agentic AI. Separately, ChangXin Memory Technologies is preparing for a capacity transition to mass-produce DDR6 memory, which will take several years.

Effective: +20
Intensity: 70
Confidence: 70%
Horizon: Medium term

Companies28 Jul, 23:19positive

Tianfu Communication: AI Drives Optical Interconnect, 1.6T Engine Mass Production, Stock Down 13.21%

On July 28, shares of Tianfu Communication fell 13.21% amid a broad sell-off in CPO concept stocks, though the stock remains up 25.41% year-to-date. The company's board secretary stated that AI computing power is driving strong demand for optical interconnects, which along with computing and storage form the three pillars of AI infrastructure. Tianfu Communication has mass-produced 1.6T optical engines, is delivering CPO supporting products (FAU and ELS), and is advancing production bases in Suzhou, Jiangxi, and Thailand.

Effective: +20
Intensity: 65
Confidence: 75%
Horizon: Medium term

Companies28 Jul, 13:03positive

Moonshot AI Open-Sources Kimi K3, Haiguang Information Completes DCU Platform Adaptation Same Day

On July 27, Moonshot AI released its open-source large language model Kimi K3. On the same day, Haiguang Information completed full-process adaptation, deployment, and performance verification of the model on its Haiguang DCU platform.

Effective: +19
Intensity: 65
Confidence: 80%
Horizon: Short term

Companies27 Jul, 08:51positive

Anthropic Launches Claude Opus 5, Deploys AMD Instinct MI450 GPUs, Secures Up to $5 Billion Investment

Anthropic released Claude Opus 5 on July 25, positioning it as a pragmatic model for daily high-frequency use with improved step completion and intermediate result verification. It becomes the default model for Claude Max and the most powerful model available to individual subscribers. Opus 5 is priced at $5 per million input tokens and $25 per million output tokens, matching Opus 4.8 and roughly half the cost of Fable 5. In benchmarks, Opus 5 outperformed all tested models on Frontier-Bench v0.1 and achieved scores close to Fable 5 on CursorBench 3.2 at lower cost. Anthropic also announced plans to deploy up to 2 gigawatts of AMD Instinct MI450 GPUs, with the first gigawatt expected in the first half of 2027, and AMD committed to a strategic equity investment of up to $5 billion.

Effective: +19
Intensity: 75
Confidence: 85%
Horizon: Short term

Companies27 Jul, 19:23positive

Zhongji Innolight Exclusively Gets 2.4T Light Coherent Module Order, Products from 40G to 1.6T

Zhongji Innolight has exclusively received a demand order for 2.4T light coherent optical modules. The company's product portfolio spans from 40G to 1.6T, covering multiple generations of high-speed optical modules. It serves major global cloud vendors, internet platforms, and AI computing companies. The company is investing in silicon photonics, coherent technologies, and advanced packaging, with silicon photonics already deployed in high-speed modules. It also develops thin-film lithium niobate for future rate upgrades. The order underscores its position in the optical interconnect market for AI clusters.

Effective: +19
Intensity: 70
Confidence: 85%
Horizon: Medium term

Companies27 Jul, 17:29positive

Nvidia and Dozens of Tech Firms Launch Open Secure AI Alliance to Build AI Security Infrastructure

Nvidia, together with Adobe, Dell Technologies, CrowdStrike, Hugging Face, Microsoft, IBM, SpaceX, Palantir and dozens of other tech firms, announced the Open Secure AI Alliance (OSAA) on July 27, 2026. The alliance, built on the Linux Foundation's Akrites project and the Open Source Security Foundation (OpenSSF), aims to develop and share open AI security and cybersecurity tools to create a unified AI security infrastructure. The move follows an incident where an OpenAI experimental AI agent breached Hugging Face, which used open-weight model GLM 5.2 to analyze more than 17,000 operations and contain the attack. Nvidia will open-source its Nvidia Labs Object-Oriented Agent (NOOA) framework for testing and managing AI agent behavior.

Effective: +19
Intensity: 70
Confidence: 80%
Horizon: Medium term

Companies28 Jul, 19:26positive

Meta and BlackRock Form Joint Venture for $14 Billion Data Center Campus in Texas

Meta and BlackRock formed a joint venture to develop a data center campus in El Paso, Texas, costing about $14 billion. BlackRock-managed funds hold 80%, Meta 20%. Meta contributed land and construction worth $2.3 billion; BlackRock contributed $4.9 billion cash plus debt. BlackRock will also pay Meta $1 billion cash to finalize the equity split. The facility is one of 28 U. S. data centers for Meta. Meta plans $600 billion cumulative investment by 2028 for AI and core business. AI infrastructure financing costs are rising, with $270 billion in new AI bond issuance so far this year, nearly double the 2025 total.

Effective: +19
Intensity: 75
Confidence: 60%
Horizon: Medium term