Futures
Access hundreds of perpetual contracts
TradFi
Gold
One platform for global traditional assets
Options
Hot
Trade European-style vanilla options
Unified Account
Maximize your capital efficiency
Demo Trading
Introduction to Futures Trading
Learn the basics of futures trading
Futures Events
Join events to earn rewards
Demo Trading
Use virtual funds to practice risk-free trading
Launch
CandyDrop
Collect candies to earn airdrops
Launchpool
Quick staking, earn potential new tokens
HODLer Airdrop
Hold GT and get massive airdrops for free
Launchpad
Be early to the next big token project
Alpha Points
Trade on-chain assets and earn airdrops
Futures Points
Earn futures points and claim airdrop rewards
How to view today's rebound? Some positive market directions recently
The index showed a reversal today compared to yesterday’s trend. Yesterday opened high and declined, while today tested the bottom and rebounded. By the close, the Shanghai Composite rose 0.32%, the Shenzhen Component increased 1.05%, and the ChiNext Index gained 2.02%. [Taoguba]
Both markets had more advancing stocks than declining ones, with over 3,500 stocks rising; total daily turnover was 2.03 trillion yuan, down 173.6 billion yuan from the previous day, with net outflows of 4.837 billion yuan from domestic main funds. The median of gains and losses was 0.8%.
In the 13 trading days this month, the market had 8 bearish and 5 bullish days. Overall, the Shanghai Index has fallen 2.4%. However, the Shenzhen and ChiNext indices continue to perform strongly. The Shenzhen Index’s pattern remains intact, and today, after rising, the ChiNext Index also moved above all moving averages. Based on the trend, it still suggests focusing more on stocks in the Shenzhen and ChiNext markets.
However, today’s rebound was characterized by significantly reduced volume, indicating a short-term downtrend persists. Coupled with external uncertainties like the Iran situation, the outlook remains cautious. Today’s rebound should be viewed as a temporary correction.
In the short term, the index needs to stabilize above the 60-day moving average, followed by a strong volume bullish candle to break through multiple resistance lines and confirm the trend. Until then, it’s prudent to prepare for both sides—mainly observing and reducing unnecessary trades. For trading, maintaining light positions is advisable, especially for those with larger losses this month. Staying on the sidelines is the safest approach.
Yesterday’s broad decline repeatedly reminded us that in uncertain conditions with low win rates, frequent trading and chasing highs or selling lows are highly risky. When stocks held are falling, chasing strong stocks that then decline further can lead to significant losses.
Today, the market continued to rebound, with three stocks in the portfolio rising nearly 5%.
The core focus remains on the GTC conference led by Jensen Huang.
Today, Alibaba and Baidu joined tech giants like Google and Tencent in raising prices to accelerate monetization of AI services.
Alibaba Cloud announced that starting April 18, due to exploding global AI demand and rising supply chain costs, prices for AI computing power and storage products will increase, with a maximum rise of 34%. This includes a 5%-34% increase in computing cards like the PingTouGe ZhenWu 810E, and a 30% increase in file storage products (CPFS, Intelligent Computing Edition).
Baidu Smart Cloud will also raise prices from April 18, with AI computing power products increasing by about 5%-30%, and parallel file storage up approximately 30%. The reasons are similar to Alibaba Cloud’s—rising demand for computing power and significant increases in core hardware and infrastructure costs.
Last week, Tencent announced that the price of its “Hun Yuan” foundational model on its AI platform increased over four times; it also started charging for third-party models hosted on its cloud, such as Zhipu AI and Moonshade, ending free trials. Additionally, Google recently announced plans to raise related service prices.
Recently, tokens have been the most discussed topic—driven by AI applications and token logic. The full potential has yet to be realized; we are only at the beginning.
Alibaba CEO Wu Yongming internally stated that AI Agents are highly dependent on tokens, and a “demand explosion” is imminent.
Jensen Huang pointed out that the correlation between tokens and AI enterprise revenue is becoming clearer—more computing power means more tokens generated, which boosts income and makes AI smarter.
He even suggested that in future Silicon Valley job markets, tokens might become part of engineers’ annual compensation packages, such as “employment with token quotas,” becoming a new recruitment topic.
At GTC, Huang predicted that the Blackwell and Rubin platforms will have procurement orders totaling $500 billion before 2026. He also set a new target for next year, expecting this figure to double to over $1 trillion.
To support this trillion-dollar growth, Huang introduced Nvidia’s new narrative—shifting from data centers to token factories. He believes that in the future, every AI company and cloud service provider should focus on token factory efficiency as a core metric.
Tokens will also become a new “commodity,” with re-pricing based on throughput and interaction speed. At GTC, Nvidia launched Vera Rubin, consisting of 7 new chips, 5 rack systems, and 1 supercomputer.
72 GPUs interconnected via NVLink will accelerate prefill calculations and ensure fast response times during token generation (KV Cache). The new Vera CPU acts as a “scheduler,” handling control tasks to free up GPUs. A storage network optimized for AI data flow was rebuilt, featuring the CPO Spectrum-X switch, which integrates optical engines and switch chips—marking the world’s first optical encapsulated Ethernet switch, replacing traditional pluggable modules.
Vera Rubin is expected to ship in the second half of 2026. Its advantages include inference speeds five times faster than the previous Blackwell Ultra and a tenfold reduction in token costs, using only a quarter of the GPUs required by MoE models.
Besides Vera Rubin, Nvidia also previewed the next-generation GPU architecture, Vera Rubin Ultra (launching in 2028), and Feynman architecture. Overall, future Nvidia platforms will revolve around 3D stacking, LPU integration, heterogeneous storage, CPO (co-packaged optics), and copper interconnects.
This also benefits domestic AI computing directions and high-growth sectors in Q1, including storage, PCB, AI computing power, device materials, and packaging/testing.
Additionally, energy storage and computing-electrical synergy are part of future plans for green data centers, where energy storage may become the main power source. The AI industry is shifting toward infrastructure that combines computing and power, with networked computing and energy security becoming key investment focuses.
Next-generation copper-clad laminate (CCL) materials like M10 will drive a new upgrade cycle for AI server PCB materials.
Oil prices are also rising. Historically, sharp increases in oil prices in 1979, 2004, 2010, and 2022 significantly impacted inflation, especially affecting our PPI and US CPI. The positive effects mainly benefit exports and commodities.
Other important updates include:
However, when asked about the developer, Hunter Alpha refused to disclose details. Its profile describes it as a model with 1 trillion parameters and a context window up to 1 million tokens, aligning with Chinese media expectations for DeepSeek’s next V4 model, possibly launching as early as April. Last week, China’s weekly AI model call volume reached 4.69 trillion tokens, up 11.83% from the previous week; US weekly call volume was 3.294 trillion tokens, down 9.33%. China’s AI model call volume has surpassed the US for two consecutive weeks.
On the international stage, crises and opportunities coexist. Short-term shocks from geopolitical conflicts and chaos disrupt established development rhythms, but these upheavals also create new possibilities, offering opportunities to break through bottlenecks and rebuild advantages.