NVIDIA CEO Reveals GPT-6 Astra Trained Using 100,000 Blackwell GPUs
NVIDIA CEO Jensen Huang disclosed that the training of GPT-6 Astra utilized over 100,000 Grace Blackwell GPUs, forming a high-speed interconnected cluster via NVLink72. The next batch will feature 400,000 GPUs, although the specific models were not specified. Leading model companies in China still show a significant gap in computing power, with ByteDance currently the closest, connecting approximately 36,000 B200 GPUs. ByteDance's domestic clusters primarily use the Hopper architecture's H20 and H800. Kimi obtained around 20,000 Hopper GPUs through Alibaba, specifically the H200 model, but Alibaba denied this claim. The H200 belongs to the previous generation of Hopper, while the B200 has been upgraded to Blackwell. The GB200 combines Grace CPU and Blackwell GPU, with NVL72 capable of placing 72 GPUs in the same high-speed interconnect domain. DeepSeek has not disclosed the complete training hardware for V4, but leaked information indicates the company had about 20,000 H-equivalent computing power in May, with Huawei providing approximately 16,000 units of the 950 capacity. Liang Wenfeng stated that this batch of domestic cards is equivalent to 4,000 Nvidia B-series cards, sufficient for training the current generation of models. The Blackwell used by Astra is not NVIDIA's latest generation; Vera Rubin has entered full-scale production, and NVIDIA estimates that the number of GPUs required for training large MoE models with Rubin can be reduced to a quarter of that needed for Blackwell. Currently, no Chinese company has disclosed a case of using 100,000 advanced GPUs of the same generation for a single model training, highlighting the gap between Chinese and American AI companies in terms of both quantity and generation of cards.
-- Price
This content is provided for general informational purposes only and doesn't constitute financial, investment, legal, or tax advice. Any events, rewards, online promotions, or related information mentioned herein should not be considered a recommendation, solicitation, or invitation to purchase, sell, trade, or otherwise deal in any crypto assets. Crypto assets are highly volatile and may result in loss. The availability of WEEX services, products, and related events may vary by region. You are responsible for ensuring that your participation is in accordance with applicable local laws and regulations.
You may also like

Stablecoin remittances show 84% cost gap, but payout reliability varies

ASML Develops Hybrid Bonding Equipment

The correlation between Bitcoin and US Treasuries exceeds 0.6

Capital B reports a loss of €69.5 million due to BTC devaluation in semiannual results
AMD Stock Ahead of Q3 2026 Earnings: Revenue, AI Demand, and Key Risks to Watch
AMD reports Q3 2026 results on November 3. See the data-center, AI demand, margin, and guidance signals that could shape AMD stock.

Goldman Sachs Analysis: Interest Rate Cuts Expected in Second Half of 2027, How Long Will High Rates Last?
![[Exclusive] Coinbase Transaction Data Shows 100 Million Trades Misrepresented... Data Analysis Firm Corrects Errors](/public-static/20_9098579959.png?format=avif)
[Exclusive] Coinbase Transaction Data Shows 100 Million Trades Misrepresented... Data Analysis Firm Corrects Errors

DBS Bank: Core CPI Expected to Slow to 0.2%

Where Are the Discrepancies in This Round of CPI? Three Judgments from Morgan Stanley, Nomura, and Citigroup

Major Banks' Q3 Reports Show Resilience in U.S. Consumer Spending and Active AI Financing Demand

Strive buys 3,311 Bitcoin as BTC holdings reach 32,773

Spy Module Confirmed in Ledger Flex Purchased at Singapore Airport

BigTime|How Much is GSR Worth: Public Financial Clues, Valuation Multiples, and Its Boundaries

Goldman Sachs Interpretation: The Higher U.S. Treasury Yields, the More Reasons to Go Long?

IMF: What Conditions Are Needed for Tokenization to Realize Value?

Mr&强 Analyzes US Stocks and Crypto Market, BTC and ETH Strategy Recommendations

Saylor: AI Will Amplify Bitcoin's Value

CITIC Securities: High Earnings Growth in US Stocks No Longer Concentrated in Technology Companies

Rune Points Out BBO Price Manipulation Risk in PaperTrade

BTCPay Server Docker users must select Tor for onion access

Severe Criticism from the Senate of the U.S. Treasury Buyback Policy; An Intervention That Backfired

OJK Prepares New Crypto Trading System in Indonesia, Will the Role of Exchanges Change? - Fintech World

What still works in crypto marketing in 2026 (and what doesn’t)

MedCred: a clandestine publication claims exposure of data from 274,534 users

BigTime|Behind the Market Maker's Entry: The Transparency Gap in Market Making Contracts

Citi Predicts Dovish Surprise from the Fed, Core PCE Near 2% May Affect Rate Hike Path

Mr&强 Analyzes BTC and ETH Opening Strategies

Fed Minutes Show Cooling Rate Hike Expectations, Samsung Q3 Operating Profit Soars 783%

September FOMC Meeting Minutes Released; Possibility of Further Rate Hike Exists

Contagion Effect: Which Latin American Economies Are Most Exposed to Rising Rates in the US
Stablecoin remittances show 84% cost gap, but payout reliability varies
ASML Develops Hybrid Bonding Equipment
The correlation between Bitcoin and US Treasuries exceeds 0.6
Capital B reports a loss of €69.5 million due to BTC devaluation in semiannual results
AMD Stock Ahead of Q3 2026 Earnings: Revenue, AI Demand, and Key Risks to Watch
AMD reports Q3 2026 results on November 3. See the data-center, AI demand, margin, and guidance signals that could shape AMD stock.






