DeepSeek

Today’s Key News Recap

09992

KALU

HSAI

00700

80700

TCTZF

TCEHY

02525

01347

688347

DXYZ

ORCL

ORCL-D

DPSK

QCOM

300274

HPPD

OpenAI

HTCD

0427 | Dolphin Research Focus: 🐬 Macro/Industry1) Last week, DeepSeek surprise-launched its flagship foundation model DeepSeek-V4 and fully open-sourced it, a key market catalyst. With AI advances as a tailwind, semis extended gains today: the Hong Kong Semiconductor Index jumped 5%+, led by SMIC and Hua Hong Semiconductor.The new model is tuned for domestic compute and pursues a non-CUDA stack, expanding edge-AI use cases and sustaining demand for local chips. Flows continue to pile into themes of self-reliant compute and domestic substitution...

POP MART

Kaiser Aluminum

Hesai

TENCENT

TENCENT-R

Tencent Holdings Limited

Tencent

HESAI-W

HUA HONG SEMI

Hua Hong

Destiny Tech100

Oracle

ORACLE CORP DEPOSITARY SH REP 1/2000TH PFD SER D

MediaTek

Qualcomm

Sungrow Power Supply

POP MART HK SDR 20to1

Coca Cola

Tencent HK SDR 10to1

SMIC

SMIC HK SDR 5to1

DeepSeek-V4 open-sourced, igniting the AI compute supply chain; Tencent's QClaw rolls out a major update | Daily News Recap

TECH

03750

300750

NVDA

HCCD

07709

SSNGY

00941

80941

HCMD

600941

NVDL

07788

07388

NVDY

NVDD

NVDX

NVDQ

Apr 20 ｜ Dolphin Research Focus: 🐬 Macro/Industry. 1) Ibrahim Azizi, chair of Iran's Islamic Consultative Assembly Committee on National Security and Foreign Policy, told Al Jazeera that Iran will continue talks with the U.S., but stressed it is 'not at any cost' and will not accept arbitrary U.S. demands.The current U.S.-Iran temporary ceasefire expires on Apr 22. The two sides held a first round in Oman earlier, and the region sits at an inflection point between war and peace.The restart signals a pragmatic stance by Tehran. Yet core differences remain, leaving limited room for maneuver...

Bio-Techne

CATL

NVIDIA

Micron Tech

CATL HK SDR 30to1

XL2CSOPHYNIX

Samsung Electronics 

CHINA MOBILE

CHINA MOBILE-R

CN Mobile HK SDR 5to1

China Mobile

GraniteShares 2x Long NVDA Daily ETF

XL2CSOPNVDA

XI2CSOPNVDA

YieldMax NVDA Option Income Strategy ETF

Direxion Daily NVDA Bear 1X ETF

T-Rex 2X Long NVIDIA Daily Target ETF

T-Rex 2X Inverse NVIDIA Daily Target ETF

MU debuts 24Gb GDDR7 memory; DeepSeek's first external round sparks AI capital frenzy | Daily News Recap

MiniMax

Deep Research

GOOG

GOOGL

02390

00100

In the business-model analysis 'Deep Dive into MiniMax and Zhipu: Are Foundation Models, Compute Intensity and Financing Stamina the Decider?', Dolphin Research argued that amid high costs and rapid iteration, the foundation-model industry has turned into a brutal test of financing endurance. But the true core of that money magnetism rests on two fundamentals for model companies: 1) model capability; 2) product and commercial execution. This piece focuses on how $MINIMAX-W(00100.HK) performs on those two fronts...

Alphabet - C

Alphabet

ZHIHU-W

Zhihu

MINIMAX-W

MiniMax’s sky‑high valuation: bubble or the future?

Alibaba

Baidu

09988

89988

HBBD

BABA

BYTED

BIDU

89888

09888

HBUD

$Alibaba(BABA.US) $ByteDance(BYTED.NA) $TENCENT(00700.HK) $Baidu(BIDU.US) In our previous discussions of the AI value chain, Dolphin Research focused chiefly on North America. We examined the tangled, interdependent bargaining among upstream players.By contrast, China’s AI ecosystem only truly broke into the mainstream last year, shaped by a consumer-driven base. After DeepSeek broke the compute-cost stalemate, the industry has tacitly converged on an application-first path as the locus of competition...

BABA-W

BABA-WR

Alibaba HK SDR 5to1

ByteDance

BIDU-SWR

BIDU-SW

Baidu HK SDR 10to1

AI Internet: 2026 ByteDance vs. BABA vs. Tencent — Three-Way Showdown

NOVAQ

NOVA

01696

NVMI

AAPL

603020

AMZN

1202 | Dolphin Research Key Focus: 🐬 Macro/Industry 1. DeepSeek simultaneously launched two official versions of AI models, DeepSeek-V3.2 and DeepSeek-V3.2-Speciale. The former's inference capability is close to GPT-5, with tool invocation reaching the top level of open-source models; the latter won four international competition gold medals, specializing in complex tasks. The models introduce the DSA sparse attention mechanism, significantly reducing inference costs. The dual-model strategy precisely covers different needs, and technological breakthroughs narrow the gap between open-source and closed-source models...

SUNNOVA ENERGY INTERNATIONAL INC

Sunnova Energy

SISRAM MED

Nova

Apple

Apple Group

Amazon

Global Tech Blitz: Nvidia Invests in Synopsys, Apple AI Shakeup, Alibaba Updates AI Model | Today's Important News Recap

Google

TSLA

DOCS

GGLS

GGLL

TSDD

TSLL

TSLQ

09366

07766

07366

TSLR

FATE

GEMI

OpenAI has been incredibly busy lately, dabbling in e-commerce, social media, and now launching a browser. Emmmm... it seems they've made enemies with half of the 'Seven Sisters.' Dolphin Research can hardly keep up with the trending topics every three days. Before the earnings report of $Alphabet(GOOGL.US), let's catch up on OpenAI's release of Atlas last week, an AI browser that claims to kick Chrome's butt. In fact, OpenAI's plan to launch an AI browser was leaked by Reuters back in July: OpenAI plans to launch an AI browser in the future...

Tesla

Doximity

Direxion Daily Googl Bear 1x Shares

Direxion Daily GOOGL Bull 2X Shares

GraniteShares 2x Short TSLA Daily ETF

Direxion Daily TSLA Bull 2X Shares

AXS TSLA Bear Daily ETF

XI2CSOPTSLA-U

XL2CSOPTSLA

XI2CSOPTSLA

GraniteShares 2x Long TSLA Daily ETF

Fate Therap

Park Hotels & Resort

Gemini Space Station

Google: My Fate is Determined by Myself, Not by OpenAI

Baidu 1Q25 Quick Interpretation: The first-quarter performance slightly exceeded expectations overall (cloud, user numbers). Simply put, DeepSeek was a big help!1. The main highlight was the Smart Cloud. After DeepSeek gained popularity at the beginning of the year, Baidu Cloud quickly integrated it, offering not only the lowest computing power prices on the market but also a free trial for a week. This drove Smart Cloud's growth rate to 42% in Q1, far exceeding the institutional expectation of 25%.2. Advertising revenue continued to face pressure, but given that the macroeconomic situation in Q1 was not as bad as initially feared, some institutions later became slightly more optimistic about ad revenue (above Bloomberg consensus). However, the actual performance was still not great.3. Thanks to the full version of DeepSeek's launch, the user base (monthly active users) of Baidu Mobile saw a significant boost, with a net increase of 45 million quarter-on-quarter—a much-needed relief. However, these returning users are likely more engaged with the AI search page, making it difficult for traditional ad models to directly benefit.4. Gross margin declined by 7 percentage points due to changes in the business structure. However, in terms of operating expenses, while sales expenses increased year-on-year due to promotional needs, R&amp;D expenses accelerated their decline. This is likely due to AI's efficiency improvements in internal R&amp;D, effectively replacing some basic R&amp;D staff. As a result, the decline in operating profit margin was less severe than the gross margin drop. From a business perspective, we also estimated through simple breakdowns that Smart Cloud, smart hardware, and other businesses like Robo-taxi showed improved loss reduction.A detailed review will be published later in the community. Key points to focus on during the earnings call: 1) DeepSeek's contribution to cloud growth and its sustainability, as well as the profitability of the cloud business; 2) How AI search commercialization will gradually roll out and the company's phased goals. $Baidu(BIDU.US) $BIDU-SW(09888.HK)

小红书

Revisiting the Revaluation of Chinese Assets!Yesterday, I talked about Alibaba. Today, let's discuss the revaluation of Chinese assets.In terms of countries that can enter the competition in AI, only the United States and China are on the table. In the face of a worse global competitive environment,a. China can produce Deepseek;b. The explosive popularity of Nezha 2 and Black Myth: Wukong is not only a matter of cultural consumption breaking out but also reflects China's action production level and even the overseas expansion of Chinese games, which has demonstrated that China's animation production level is already top-tier. Based on the technical level, it adds the narrative soul of Chinese elements.c. The success of TikTok and the breakout of Xiaohongshu are the results of the combined promotion of technology and cultural consumption.Moving beyond short-term geopolitical games, the long-term competitiveness of society and enterprises constitutes the core underlying valuation.The competitiveness of Chinese private enterprises demonstrated in a harsh environment is completely comparable to that of Europe and Japan, so the core question is, why are Chinese assets continuously undervalued compared to their overseas counterparts?Therefore, in this round of revaluation of Chinese assets, especially Chinese technology assets, we first need to level the country-specific asset discount.Then, we can discuss whether the fundamentals are facing deflation as a cyclical issue.So here, Dolphin Research reiterates that by 2025, there will be a continuous revaluation of Chinese assets.In 2026, it will connect with the fundamental bottoming rebound.So next, if you look at Chinese assets with colored glasses and do not take them off, your excess returns will be at risk!https://longportapp.com/zh-CN/topics/27088036

中概：值得一场严肃的 “认知重估”！ - LongPort

STEP

DeepSeek Launches Image Understanding Feature in Gray-scale Testing

DeepSeek Plans First Ever Fundraising Round for at Least $300 Million

DeepSeek Lays Out Data Center in Ulanqab

Chinese AI models' weekly usage exceeds US for two consecutive weeks, MiniMax M2.5 tops the chart

OpenAI Warns US Legislators: Chinese AI Firm Attempts to Acquire Data

DeepSeek Launches New Model and Conducts Gray-scale Testing

Chinese Open-Source AI Models See Rapid Global Market Share Growth

DeepSeek Leads China's AI Industry in Open-Source Rise with DeepSeek-R1 Model

DeepSeek Releases New Model MODEL1 and Updates FlashMLA Code

DeepSeek AI Model Sparks a Boom in the Financial Industry

Deep Quest Proposes New AI Development Plan to Enhance Model Performance

Deep Search Proposes mHC Model Architecture

DeepSeek Launches and Open Sources AI Model, Promoting AI Technology Accessibility

Medical AI Industry Maintains High Growth, Hospitals and Doctors Embrace Large Models

Multiple Companies Release New Products and Technologies; Policies Drive AI Industry Development

Chinese Open-Source AI Models Gain 30% Global Market Share

AI Large Models Drive Software Service Index Growth

Zhejiang's AI Core Industry Revenue Up 22% in First Three Quarters, Computing Power Surpasses 150Eflops

DeepSeek Reaffirms Commitment to Developing Artificial General Intelligence

The World Internet Conference to be held from November 6-9 in Wuzhen

NVDS

HUAWEI

Ahead of the recent U.S.-China summit, Beijing achieved a significant milestone in its efforts to become technologically self-reliant.

Unusual Whales

AXS 1.5X NVDA Bear Daily ETF

GraniteShares 2x Short NVDA Daily ETF

Huawei

Direxion Daily NVDA Bull 2X Shares

NVIDIA Corporation

Chinese AI companies surpass Nvidia with DeepSeek choosing Huawei before the U.S.-China summit, marking a tech self-sufficiency milestone.

DeepSeek's recent valuation of 50 billion yuan (approximately $7.35 billion) marks a significant shift in the AI landscape, highlighting a divergence between US and Chinese approaches. While US companies like OpenAI and Anthropic focus on massive funding and closed-source models, Chinese firms like DeepSeek prioritize efficiency, open-source ecosystems, and lower costs. DeepSeek's founder plans to invest 20 billion yuan, signaling a move from a technology lab to a key player in China's national AI strategy. This evolution underscores a fundamental change in global AI competition, emphasizing differing development paths.

DeepSeek(DPSK.NA)

Chinese AI companies surpass Nvidia with DeepSeek choosing Huawei before the U.S.-China summit, marking a tech self-sufficiency milestone.

Deepseek's 50 billion rating: Chinese AI is starting to change.

Golden Web3.0 Daily Report | CZ Proposes Relaunching Binance.US

Italy closes antitrust probes into AI firms after commitments on 'hallucination' risks

DeepSeek begins limited testing of vision mode