DeepSeek 新一轮融资估值约710亿美元，资金将用于自建数据中心和AI芯片

DeepSeek 已经再次开始融资。中国的 AI 实验室刚刚完成了首轮融资，需要资金用于自建数据中心和芯片，以维持其激进的定价策略。

据《金融时报》报道：https://archive.ph/AnZMc，DeepSeek 在五月底完成了约 70 亿美元的首轮融资：https://the-decoder.com/deepseek-takes-outside-money-for-the-first-time-at-a-50-billion-valuation/，公司估值为 520 亿美元。现在，总部位于杭州的这家初创公司正在与新的投资者进行早期谈判，拟以约 710 亿美元的投前估值进行新一轮融资。资金将用于建设自有数据中心和购买 AI 芯片。与此同时，DeepSeek 正在研发自己的推理芯片：https://the-decoder.com/deepseek-is-designing-its-own-ai-chip/，以减少对 Nvidia 和华为的依赖。创始人梁文锋个人投入约 30 亿美元，成为最大投资者。其他投资者包括宁德时代、腾讯、京东、网易以及中国国家支持的 AI 基金。

低价策略需要雄厚资金支持 广告

资金需求直接与 DeepSeek 的激进扩张相关。公司最近推出了 V4-Pro 和 V4-Flash：https://the-decoder.com/as-agentic-ai-pushes-rivals-to-raise-prices-and-cap-usage-deepseek-ships-a-good-enough-model-for-almost-nothing/，这是拥有多达 1.6 万亿参数的最大开源权重模型。V4-Pro 的超低价格已永久生效：https://the-decoder.com/deepseek-makes-its-75-percent-discount-permanent-pricing-output-tokens-at-least-34x-below-gpt-5-5/，输入成本大约比 GPT-5.5 便宜 11 倍。广告 DEC_D_Incontent-1

这一策略正在奏效。根据美国金融服务公司 Ramp 的数据：https://the-decoder.com/deepseek-topped-ramps-trending-software-vendors-in-june-2026-as-us-companies-chase-cheaper-ai/，该公司跟踪 5 万多家企业的实际支出，DeepSeek 是 6 月份美国企业中增长最快的软件供应商之一。然而，Ramp 也指出了安全风险，因为企业的数据显示是直接通过 DeepSeek 的平台传输的。

与此同时，与西方领导者的性能差距正在扩大。OpenAI 的 GPT-5.6 Sol：https://the-decoder.com/openais-claude-mythos-competitor-gpt-5-6-sol-launches-under-government-controlled-access-it-calls-unsustainable/ 和 Anthropic 的 Claude Mythos：https://the-decoder.com/new-claude-mythos-becomes-the-first-ai-model-to-clear-all-cyberattack-simulations-from-britains-ai-safety-agency/ 已经达到了 DeepSeek 目前无法匹敌的新水平。尽管如此，性能差距仍远小于价格差距。广告

中国国内的竞争正在迅速升温。智谱 AI 发布了 GLM-5.2：https://the-decoder.com/zhipu-ais-glm-5-2-closes-in-on-closed-source-leaders-in-coding-marathons/，这是一款开源模型，在长时间编码任务中仅落后 Anthropic 的 Opus 模型几个百分点，并且正在企业中获得关注：https://the-decoder.com/databricks-makes-chinese-open-source-model-glm-5-2-its-default-coding-engine-after-it-matched-opus-at-lower-cost/。据报道，MiniMax 正在研发一款 2.7 万亿参数模型：https://the-decoder.com/chinese-ai-startup-minimax-plans-to-open-source-a-2-7-trillion-parameter-model-later-this-year/，可能在第三季度就能发布。而 Kimi 背后的公司 Moonshot AI 正寻求新资本，估值最高达到 300 亿美元：https://the-decoder.com/moonshot-ai-targets-a-30-billion-valuation-more-than-six-times-its-late-2025-worth/。中国 AI 实验室之间的竞赛从未如此激烈。

保持对 AI 的关注。信息清晰、有用，无冗余。

关注 The Decoder 获取 AI 新闻、背景故事和专家分析。

The Decoder：https://the-decoder.com/

DeepSeek is already raising again. The Chinese AI lab just closed its first funding round and needs capital for its own data centers and chips to keep its aggressive pricing strategy going.

DeepSeek closed its first round in late May at roughly $7 billion：https://the-decoder.com/deepseek-takes-outside-money-for-the-first-time-at-a-50-billion-valuation/, valuing the company at $52 billion, the Financial Times reports：https://archive.ph/AnZMc. Now the Hangzhou-based startup is in early talks with new investors for a round at a pre-money valuation of about $71 billion. The money would go toward building its own data centers and buying AI chips. At the same time, DeepSeek is developing its own inference chip：https://the-decoder.com/deepseek-is-designing-its-own-ai-chip/ to reduce its reliance on Nvidia and Huawei. Founder Liang Wenfeng put in about $3 billion himself, making him the largest backer. Other investors include CATL, Tencent, JD.com, NetEase, and China's state-backed AI fund.

Cheap pricing demands deep pockets Ad

The need for capital is tied directly to DeepSeek's aggressive expansion. The company recently released V4-Pro and V4-Flash：https://the-decoder.com/as-agentic-ai-pushes-rivals-to-raise-prices-and-cap-usage-deepseek-ships-a-good-enough-model-for-almost-nothing/, the largest open-weights models with up to 1.6 trillion parameters. The rock-bottom V4-Pro prices have been made permanent：https://the-decoder.com/deepseek-makes-its-75-percent-discount-permanent-pricing-output-tokens-at-least-34x-below-gpt-5-5/ and come in at roughly eleven times cheaper than GPT-5.5 on input. Ad DEC_D_Incontent-1

The strategy is working. According to US financial services firm Ramp：https://the-decoder.com/deepseek-topped-ramps-trending-software-vendors-in-june-2026-as-us-companies-chase-cheaper-ai/, which tracks real spending across more than 50,000 companies, DeepSeek was among the fastest-growing software vendors among US businesses in June. Ramp did flag security risks, though, since companies are sending data directly through DeepSeek's platform.

Meanwhile, the performance gap with Western leaders has widened. OpenAI's GPT-5.6 Sol：https://the-decoder.com/openais-claude-mythos-competitor-gpt-5-6-sol-launches-under-government-controlled-access-it-calls-unsustainable/ and Anthropic's Claude Mythos：https://the-decoder.com/new-claude-mythos-becomes-the-first-ai-model-to-clear-all-cyberattack-simulations-from-britains-ai-safety-agency/ have reached a new tier that DeepSeek can't match yet. Still, the gap in performance is far smaller than the gap in price. Ad

Competition inside China is heating up fast. Zhipu AI released GLM-5.2：https://the-decoder.com/zhipu-ais-glm-5-2-closes-in-on-closed-source-leaders-in-coding-marathons/, an open-source model that trails Anthropic's Opus models by only a few percentage points on hours-long coding tasks and is gaining traction with enterprises：https://the-decoder.com/databricks-makes-chinese-open-source-model-glm-5-2-its-default-coding-engine-after-it-matched-opus-at-lower-cost/. MiniMax is reportedly working on a 2.7 trillion-parameter model：https://the-decoder.com/chinese-ai-startup-minimax-plans-to-open-source-a-2-7-trillion-parameter-model-later-this-year/ that could ship as early as Q3. And Moonshot AI, the company behind Kimi, is looking for fresh capital at a valuation of up to $30 billion：https://the-decoder.com/moonshot-ai-targets-a-30-billion-valuation-more-than-six-times-its-late-2025-worth/. The race among Chinese AI labs has never been this intense.

Stay in the loop on AI. Clear, useful, no fluff.

Follow The Decoder for AI news, background stories and expert analyses.

The Decoder：https://the-decoder.com/