文 | 时间线Timelines
2016 年 8 月 9 日，英伟达 CEO 黄仁勋来到 OpenAI 位于旧金山的办公室，向这家当时成立不久的人工智能公司交付了世界上第一台 DGX-1 超级计算系统。彼时，恐怕很少有人能够意识到，这次交付会成为全球 AI 算力产业爆发的一个预言式的历史切片。
同样是在 2016 年 8 月，在大洋彼岸的中国，脱胎于中科院的寒武纪完成了来自元禾原点、科大讯飞等机构的 Pre-A 轮融资。那时候，作为一家创业公司的寒武纪，几乎与英伟达没有产生任何勾连。
此后十年间，情况发生了巨大的变化。
凭借着 AI 的大爆发，英伟达从一家以游戏显卡闻名的芯片公司，一跃成长为全球 AI 时代最重要的基础设施供应商，市值突破 5 万亿美元，并在很长时间位居全球第一。但与此同时，英伟达在中国市场的缺位，也让寒武纪也由此成为国产算力崛起最具代表性的受益者之一。
甚至寒武纪在资本市场因为视为「英伟达平替」，并由此获得资本追捧，并一度被称为「寒王」。
到 2026 年 10 月 23 日，情况再次出现波动。
当天晚间，寒武纪发布 2026 年三季度财报，单季营收和净利润双双创下历史新高，前三季度归母净利润达到 37.15 亿元。而大洋彼岸的美股市场，英伟达市值在小幅度下跌中，也被上涨的苹果超越，位居第二。
这个时候，在寒武纪交出上市以来最好成绩单的同时，其股价却已经较高点回撤超过 40%，市值从一度突破 1 万亿元回落至约 5862 亿元。
这仿佛意味着，十年之后，寒武纪终于在业绩上迎来了最好的时代；但资本市场曾经用来给它定价的「英伟达平替」叙事，却似乎开始褪去了。
最佳财报背后的一场冒险
从财务数字来看，寒武纪确实交出了一份自成立以来的最佳季度成绩。
财报显示，寒武纪在 2026 年 Q3 实现营业收入 35.31 亿元，同比增长 104.47%；归母净利润 14.04 亿元，同比增长 147.74%；扣非归母净利润 13.58 亿元，同比增长 168.13%。
今年前三个季度，寒武纪累计实现营业收入 95.26 亿元，同比增长 106.76%；归母净利润 37.15 亿元，同比增长 131.49%；扣非归母净利润 35.23 亿元，同比增长 148.30%。
从这个整体面的数据来看，无论是营收，还是净利润，寒武纪的业务支撑力都还是非常强的。
尤其值得注意的是，寒武纪 Q3 的扣非净利润增速高于归母净利润增速。这意味着，寒武纪 Q3 的利润增长，主要还是由主营业务驱动，而不是过度依赖政府补助、投资收益等非经常性损益。
从环比角度来看，寒武纪 Q2 营收相比 Q2 增长了 13.50%，归母净利润也环比增长约 8%。同时，今年第一至第三季度，公司归母净利润分别为 10.13 亿元、12.98 亿元和 14.04 亿元，呈现出逐季度增长的态势。
再来看盈利能力。今年前三季度，寒武纪的营业成本为 41.41 亿元，据此计算，公司的毛利率约为 56.53%，相比上年同期，甚至提升了约 1.24 个百分点；同期归母净利率约为 39%——其中，寒武纪 Q3 的归母净利率更是接近 40%。
对于一家仍在快速扩张的芯片设计公司来说，这样的利润率已经相当可观。其核心逻辑是，随着收入规模迅速扩大，销售、管理等费用并没有同步翻倍，公司的经营杠杆由此得到释放。
需要说明的是，在快速发展的 AI 算力行业，寒武纪也没有因为利润增长而减少研发投入。财报显示，三季度，寒武纪研发投入为 5.23 亿元，同比增长 73.66%；前三季度研发投入为 12.25 亿元，同比增长 45.35%。
不过，如果继续观察资产负债表，这份财报也并非没有瑕疵。
其中，截至三季度末，寒武纪的存货达到 116.65 亿元，较 2025 年末的 49.44 亿元增长 136%；预付款项达到 30.44 亿元，较上年末增长约 309%。两项数据合计达到 147.09 亿元，已经明显超过寒武纪前三季度的营业收入。
与此同时，寒武纪的应付票据和应付账款分别升至 49.66 亿元和 18.27 亿元，货币资金则降至 5.55 亿元。
那么，为什么寒武纪要进行如此大规模的备货和预付？
从公司业务模式来看，这种做法有一定的合理性。毕竟，寒武纪采用 Fabless 模式，晶圆制造、封装测试等环节需要依赖外部供应商。在国产算力需求快速增长、上游原材料价格上涨的情况下，公司需要提前锁定产能和关键原材料，为后续交付做准备。
因此，存货和预付款项的增长，可以理解为寒武纪对于未来订单和收入增长的一种提前投入。
但这并不意味着存货越高越好。今年前三季度，寒武纪的资产减值损失达到 5.02 亿元，上年同期为 0.87 亿元。公司表示，这主要是因为前期战略备货的原材料库龄增长，相应计提了存货跌价准备。这说明，大规模备货本身就是一项需要承担成本的经营决策。
也就是说，如果后续产品交付、客户需求或者产品迭代不及预期，今天的战略备货就可能变成明天的减值压力。
另外值得关注的，是寒武纪的现金流。今年前三季度，寒武纪经营活动现金流净额为 2.28 亿元，相比去年同期明显改善，但与同期 37.15 亿元的归母净利润相比，依然存在较大差距；三季度单季经营活动现金流净额为 -0.84 亿元。
所以，对于如今的寒武纪来说，核心矛盾已经不在于能否实现盈利，而在于如何把高速增长转化为更加稳定的经营质量。
国产 AI 算力格局，又变了
与寒武纪 Q3 业绩形成鲜明对比的，是寒武纪股价在三季度的明显下滑。
实际上，就在三个多月前，也就是 2026 年 6 月底，寒武纪的市值在一路暴涨中一度突破 1 万亿元。而从彼时高点算起，到 Q3 财报发布，寒武纪的市值已经回撤超过 40%。也就是说，市场开始重新给它定价。
当然，这个重新定价的过程，某种程度上是有其内在的合理性的。
实际上，自 2020 年上市之后的很长一段时间里，寒武纪在漫长的亏损过程中，虽然经历了股价起伏，却并没有受到真正的重视。直到 2022 年下半年，伴随着地缘政治导致的国产 AI 算力替代潮，国产 AI 算力替代已经成为寒武纪最核心的资本叙事。
寒武纪作为「英伟达平替」的角色被资本疯狂追捧。
于是，从 2023 年，寒武纪的股价在波折中一路向上。2024 年全年，在全年财报亏损 4.43 亿元的情况下，寒武纪的 A 股股价实现了 387% 的全年涨幅——在这样股价变动中，寒武纪被冠以「股王」的称号。
2025 全年，在 AI 芯片业务的持续落地中，寒武纪的业绩整体爆发，其全年营业收入 64.97 亿元，同比增长 453.2%；归母净利润 20.59 亿元，首次实现年度盈利。这样的业务表现，加上市场对于寒武纪未来发展的高预期，寒武纪的股价在 2025 年的持续上涨中一度超越茅台，寒武纪也被股民称为「寒王」。
2026 年，随着一季度和二季度业绩继续增长，寒武纪股价再次走高，市值最终在 6 月 30 日突破 1 万亿元。
从这个过程来看，寒武纪的万亿市值并不是单纯由概念推动的。它确实经历了从长期亏损到规模盈利的基本面反转——但问题在于，市场给它定价时，也不只是在计算已经实现的利润，而是在提前支付「中国英伟达」的想象。
然而，过去一年，英伟达和华为的一系列动态，正在改变这套叙事的基础。
先来看。在一系列地缘博弈中，英伟达依旧在 2026 年的中国 AI 算力市场占据重要地位。根据 IDC 数据，英伟达在 2026 上半年的中国 AI 加速计算市场虽然下降了 44%，但依旧占有 34% 的份额。
再看华为。在寒武纪成立之初，华为曾与之在终端 AI 芯片层面有过合作，这也让当时成立不久的寒武纪迅速声名鹊起。不过，多年之后，在国产 AI 算力的主战场上，两家公司已经从早期的合作伙伴，变成了同一赛道上的重要竞争者。
其中在 2025 年 9 月，华为公布了未来三年的昇腾芯片路线图：2026 年推出昇腾 950 系列，2027 年推出昇腾 960，2028 年推出昇腾 970，目标是几乎保持一年一代、算力持续翻倍的迭代节奏。不仅如此，华为也在软硬件等方面协同发力，在落地方面也动作很快，俨然成为 AI 国产算力的代表。
而最新的外媒消息显示，DeepSeek 的工程师甚至已经在参与华为下一代昇腾芯片的设计，DeepSeek 这边，最快 2026 年底开始用昇腾 950DT 做预训练。
在这种竞争格局下，寒武纪依然是国产 AI 算力的重要供应商，但很难再被视为国产算力唯一的答案。
更何况，阿里平头哥、百度昆仑芯、海光信息、摩尔线程、沐曦等公司也在加速推进产品和商业化，互联网大客户同时还在发展自研芯片。随着更多国产芯片企业进入资本市场，寒武纪过去独享的稀缺性溢价，也会被逐渐分流。
千亿营收目标下的估值重构
有意思的是，在产业竞争发生变化的同时，寒武纪给出了非常积极的经营目标。
就在 Q3 财报发布的同时，寒武纪也特意发布了一份关于公司近期生产经营情况的公告。公告明确表示，当前人工智能算力市场需求持续旺盛，能够支撑业务稳步增长；随着市场持续拓展与收入规模扩大，公司业绩有望保持增长。
不过，外界更加关心的，则是此前股权激励计划提出的「千亿目标」。
所谓「千亿」，指的是寒武纪 2026 年至 2028 年三年累计营业收入不低于 1000 亿元。具体目标为：2026 年营业收入不低于 135 亿元，2026 年至 2027 年累计营业收入不低于 405 亿元，2026 年至 2028 年累计营业收入不低于 1000 亿元。
公司也明确强调，这些考核指标不构成对投资者的业绩预测和实质承诺。
拆解来看，以 2026 年前三季度 95.26 亿元的收入计算，寒武纪要完成 2026 年的目标，就需要在 Q4 实现约 39.74 亿元的收入，相比三季度再增长约 12.5%。考虑到公司目前的收入趋势和备货规模，这个目标并非遥不可及。
真正的挑战在 2027 年和 2028 年。按照目标值简单拆分，寒武纪 2027 年需要实现约 270 亿元收入，相比 2026 年目标值增长 100%；2028 年则需要实现约 595 亿元收入，相比 2027 年再增长约 120%。
也就是说，在完成今年收入增长之后，寒武纪还需要连续两年实现翻倍式扩张。
这个目标实现的条件，不仅需要国内算力需求继续增长，还需要寒武纪在上游产能、关键原材料、产品迭代、客户拓展和软件生态等方面同时跟上，并获得客户认可。任何一个环节出现明显短板，都可能影响最终的收入兑现。
而从目前的估值来看，市场已经提前计入了相当乐观的预期。以 2025 年 Q4 和 2026 年前三季度的归母净利润合计计算，寒武纪目前的滚动市盈率仍然约为 141 倍——即使将今年前三季度的利润简单年化，其对应的市盈率也接近 118 倍。
不仅如此，据 iFinD 在 2026 年 10 月 24 日汇总的机构一致预期显示，寒武纪 2026 年至 2028 年的归母净利润预测均值分别约为 56.47 亿元、120.25 亿元和 208.81 亿元。以当前的最新市值计算，对应市盈率分别约为 104 倍、49 倍和 28 倍。
换言之，即使寒武纪未来两年的利润按照当前机构预期快速增长，现在的股价也已经提前计入了相当一部分增长。过去，市场可以用「英伟达平替」来解释这种估值；现在，这套故事已经很难继续讲下去。
毕竟，国产算力市场仍在扩大，但寒武纪面对的已经不是一个「只要有产品就能获得稀缺溢价」的市场。所以，寒武纪的「英伟达平替」叙事，实际上已经被资本市场淡化。
当然，从业务的角度，寒武纪未来依然有不小的发展空间。如果公司能够完成三年累计千亿元的收入目标，将大规模备货转化为实际交付，并在客户、产品和软件生态上建立更加稳固的优势，它仍然有机会成为中国 AI 芯片产业的重要一极。
但是，接近 6000 亿元的市值，已经预支了未来数年的高增长；而在全新的国产 AI 算力供给环境下，这种增长已经开始面临不确定性。
好消息是，寒武纪依旧有足够长的时间证明自己。
十年前，当黄仁勋把 DGX-1 交给 OpenAI 时，寒武纪还只是一家刚刚完成早期融资的创业公司；十年后，寒武纪已经用业绩证明，自己能够在全球 AI 算力格局的裂缝中迅速成长。
只不过，从一家抓住时代机会的公司，走向一家能够穿越竞争周期的公司，只会是另外一场更加艰难的跋涉——但这是寒武纪必须走的一条蜕变之路，在资本的光环褪去之后，这条路还很漫长。
本文由 AI图鉴自动抓取并翻译，版权归原作者所有；如需查看原文请点下方链接。
Timelines
On August 9, 2016, Young Weidar CEO Wong In-hoon arrived at Open AI ' s office in San Francisco and delivered the world ' s first DGX-1 super-calculation system to a newly established artificial intelligence company. On the other hand, I'm afraid that very few will realize that this delivery will become a predictive historical slice of the eruption of the global AI computing industry.
Also in August 2016, China, on the other side of the ocean, completed the Pre-A round of financing from institutions such as Yuanhua, Koda Z66Z, from the Cold War of the Chinese Academy of Sciences. At that time, as a business start-up company, the cold martial arts had almost nothing to do with Britain.
The situation has changed dramatically in the decade since then.
With the explosion of AI, the value of Young Wida, a chip company known as the GPU game, jumped into the world's most important infrastructure provider in the AI era, surpassing $5 trillion and ranking first in the world for a long time. At the same time, however, the lack of a position in the Chinese market has also made the cold war one of the most representative beneficiaries of the rise in national production.
Even in the capital market, the Cold War was seen as the "Everda Pendulum" and thus gained capital attraction, once known as the "Choose King".
By October 23, 2026, the situation had fluctuated again.
In the evening of the same day, the financial paper publish 2026, the third quarter of which saw both single-season revenue and net profits, reached 37.15 billion yuan in net profits to the mother in the first three quarters. The market value of the United States stock market on the other side of the ocean has fallen slightly and has been surpassed by rising apples, second in rank.
At this point, while Cold Wu has handed over the best score card since it was put on the market, its stock price has receded by more than 40 per cent, and its market value has fallen from a once-in-a-lifetime breakthrough of $1 trillion to about $586.2 billion.
As if it meant that, 10 years later, the era of the cold war had finally come to its best performance; but the narrative of the "Inverda Ping" that the capital market had used to price it seemed to have faded away.
An adventure behind the best financials.
From the financial figures, it is true that the cold war brought in the best quarterly achievements since its inception.
According to the financial statements, in 2026, Q3 realized business income of 3,531 million yuan, an increase of 104.47 per cent over the same period; net profits of 1.404 billion yuan for mothers, an increase of 147.74 per cent over the same period; and net profits of 1,358 million yuan for non-mothers, an increase of 168.13 per cent over the same period.
In the first three quarters of the year, the cold war accumulated operating income of 95.26 billion yuan, an increase of 106.76 per cent over the same period; net earnings of 3,7.15 billion yuan for mothers, an increase of 131.49 per cent over the same period; and net profits of 35.23 billion yuan for non-mothers, an increase of 148.30 per cent over the same period.
From this overall data, both revenue collection and net profits, the operational support of the Cold War is still very strong.
Of particular note is the higher rate of increase in net profits from cold martial law Q3 than net profits from returning mothers. This means that the profit growth of the Cold War Q3 has been driven largely by the dominant business, rather than by excessive reliance on non-recurrent gains and losses such as government subsidies, investment returns, etc.
From a ring comparison point of view, the Q2 harvest in the cold martial arts increased by 13.50 per cent compared to Q2 and the net profit to the mother increased by about 8 per cent. At the same time, in the first to third quarters of this year, net profits from the return of companies to their mothers were 1,013 million yuan, 1,298 million yuan and 14.04 billion yuan, respectively, reflecting quarterly growth.
Then look at profitability. In the first three quarters of this year, the operating cost of the cold war was RMB 41.41 billion, according to which the company's Māori rate was about 56.53 per cent, an increase of about 1.24 percentage points compared with the same period of the previous year; the net rate of return to the mother was about 39 per cent - of which the net return rate for the cold war Q3 was close to 40 per cent.
For a chip design company that is still expanding rapidly, the profitability is already considerable. The core logic is that, as revenues expand rapidly, the costs of sales, management and so on are not doubled simultaneously, thus releasing the leverage of the company.
It should be noted that in the fast-growing AI computing industry, Cold War did not reduce R & D inputs due to increased profits. According to the press, investment in research and development for the cold war was $523 million in the third quarter, an increase of 73.66 per cent over the previous three quarters, and 12.25 billion in research and development, an increase of 45.35 per cent over the previous three quarters.
However, if the balance sheet were to continue to be observed, the financial statements were not without defects.
Of these, by the end of the third quarter, cold martial stock amounted to $166.5 billion, an increase of 136 per cent from $4,944 million at the end of 2025; advances amounted to $5,044 million, an increase of approximately 309 per cent over the end of the previous year. The combined figures amount to RMB 147.09 billion and have significantly exceeded the operating income of the first three quarters of the cold war.
At the same time, the bills payable and accounts payable in the cold war rose to 4966 million yuan and 1.827 billion yuan respectively, while monetary funds fell to 550 million yuan.
Why, then, is it such a large supply and advance?
This approach is justified by the business model of the company. After all, the Fables model is used in the cold war, and the crystal circle manufacturing, containment testing, etc. are dependent on external suppliers. In the context of rapidly growing domestic capacity demand and rising upstream raw materials prices, companies need to pre-position capacity and critical raw materials in preparation for subsequent deliveries.
Therefore, the increase in inventories and advances can be understood as an early investment in future order and income growth in the cold war.
This does not mean that the stock is as high as possible. In the first three quarters of the year, the impairment of the assets of the cold martial arts amounted to 502 million yuan, compared to 87 million yuan during the same period of the previous year. According to the company, this was mainly due to the age of the raw materials stock for strategic forward delivery, which was accordingly prepared for stock price reductions. This suggests that large-scale stock preparation is in itself a costed operational decision.
In other words, if subsequent product delivery, customer demand or product overlays fall short of expectations, today ' s strategic readiness may become tomorrow ' s impairment pressure.
Also of concern is the cold-blooded cash flow. In the first three quarters of the year, net cash flows from cold-blooded operations amounted to 228 million yuan, a marked improvement over the same period last year, but there was still a large gap over the same period, when net returns from mothers amounted to 3,715 million yuan; net cash flows from single-season operations in the three quarters amounted to -84 million yuan.
Thus, for the modern era of combat, the core contradiction is not about making profits, but about turning high growth into a more stable quality of business.
National production, AI. The algorithm has changed again.
In stark contrast to Cold Wu Q3 performance, cold Wu stock prices declined markedly in three quarters.
In fact, just over three months ago, at the end of June 2026, the market value of the cold war was one trillion yuan in a sudden hike. From time to time, to Q3 financial paper publish, the market value of the Cold War has been withdrawn by more than 40 per cent. In other words, the market began to re-pricing it.
Of course, this process of re-pricing is inherently justified to some extent.
In fact, for a long time since it was put on the market in 2020, the cold war has not received real attention in the long period of loss, despite stock price fluctuations. Until the second half of 2022, when geopolitics led to a wave of domestic power substitutions, it was the core of capital narratives of the Cold War.
The role of Cold Wu as "in Wei Da Ping" has been admired by capital.
So from 2023, Cold Wu stock prices went up in the middle. Throughout 2024, with a loss of $443 million in financial returns for the whole year, Cold War A stock price achieved 387 per cent of the annual increase - in such a change of stock prices, it was crowned as the "equity king".
Throughout the year, in the continuing fall of the AI chip business, the overall performance of the cold war broke out, with an annual operating income of $6,497 million, or 453.2 per cent over the same period; net profits to mothers of $2,059 million were realized for the first time. This business performance, combined with the high expectations of the market for the future of the Cold War, led to an increase in the price of the Cold War stock, which, in 2025, was once overstretched, and also known by the shareholders as the "The Cold King."
In 2026, as performance continued to grow in the first and second quarters, the value of the cold war stock rose again, and the market value eventually exceeded 1 trillion yuan on 30 June.
From this process, the trillions of the market value of the cold war is not purely conceptually driven. Indeed, it has experienced a fundamental reversal from long-term losses to large-scale profits – but the problem is that the market priced it, not just to calculate the profits that have been realized, but to pay ahead of time the imagination of “China’s Young Wida.”
Over the past year, however, a series of developments, led by the United Kingdom of Great Britain and Northern Ireland, are changing the basis of this narrative.
Look first. In a series of geographies, Ying Weidar continues to occupy an important position in China's al-Calculation Market in 2026. According to the IDC, the ACAI market in China in the first half of 2026 fell by 44 per cent, but still accounts for 34 per cent.
And look at Wigand. At the beginning of the Cold War era, China had worked with it at the end of the AI chip, which had quickly given rise to the reputation of the new era. However, many years later, in the main battlefield of the national production AI, two companies have become important competitors on the same track, from early partners.
In September 2025, China published a three-year roll-out chip road map: the 950 series was launched in 2026 and the 960 and 970 in 2028 in 2027, with the goal of maintaining an almost one-year, multi-fold, iterative rhythm. Not only that, but China is also working together in the areas of software and hardware, and is fast on the ground, becoming the representative of AI's national capacity.
And the latest foreign sources show that DeepSeek's engineers are even involved in the design of the next generation's chip, DeepSeek, the fastest part of the world to start pre-training with 950DT by the end of 2026.
In this pattern of competition, the cold war remains an important supplier of national production in AI, but it is difficult to see it as the only answer.
Moreover, companies such as Ali Pinghead, Baidu Quinlan core, Sealight Information, Moor Lines, Mouse, etc. are also accelerating product advancement and commercialization, while large Internet customers are developing self-study chips. As more domestic chip companies enter capital markets, the scarce premiums that were the exclusive preserve of the Cold War will be gradually diverted.
The valuation under the 100 billion-billion-dollar collection target has been reformulated.
Interestingly, while industrial competition has changed, the Cold War has given very positive business objectives.
At the same time as the Q3 financial paper publish, Chien Wuji also gave an announcement about the company's recent production operations. The announcement made it clear that the current strong demand for artificial intelligence computing markets can support steady business growth and that corporate performance is expected to grow as markets continue to expand and incomes expand.
However, what is more interesting to the outside world is the "hundred-billion-dollar goal" of the previous equity incentive scheme.
The term "billions" refers to the cumulative operating income of the Cold War in the three years between 2026 and 2028, not less than 100 billion yuan. The targets are: business income of not less than 13.5 billion yuan in 2026, cumulative business income of not less than 40.5 billion yuan in 2026 to 2027 and cumulative business income of not less than 100 billion yuan in 2026 to 2028.
Companies have also clearly emphasized that these performance indicators do not constitute performance projections and substantive commitments to investors.
In terms of disassembly, in the first three quarters of 2026, the target for 2026 will need to be met at Q4 by approximately 3,974 million yuan, an increase of about 12.5 per cent over the third quarter. This goal is not out of reach, given the current income trends of companies and the size of the stock.
The real challenge is in 2027 and 2028. According to the simple split of the target value, approximately 27 billion yuan in income was needed in 2027, an increase of 100 per cent over the target value of 2026, and about 59.5 billion yuan in 2028, an increase of 120 per cent over 2027.
In other words, after completing this year ' s growth in revenues, the cold war will require a doubling of the expansion for two consecutive years.
The conditions for achieving this goal require not only continued growth in domestic computing needs, but also a climate in which both upstream production capacity, key raw materials, product overlay, customer outreach and software ecology are matched and recognized by clients. Any apparent weakness in any chain could affect the eventual realization of income.
The market, on the other hand, has been pre-recorded into fairly optimistic expectations in the current valuation. The current rolling market gain of Cold War is still about 141 times - even if the profits of the first three quarters of the year are simply annualized, they are close to 118 times higher than the net profits of the mother in the first three quarters of 2025 and 2026.
Moreover, according to the consistent expectations of iFinD, which were consolidated on 24 October 2026, the average projected net profit of the mother in 2026 to 2028 was approximately $5,647 million, $12.25 billion and $20,881 million, respectively. At the current latest market value, the corresponding rates were approximately 104, 49 and 28 times, respectively.
In other words, even if the profits of the cold war over the next two years are growing at a rapid rate as expected by the current institutions, now stock prices have been factored into a considerable increase ahead of schedule. In the past, the market could explain this valuation by using the term "Inverda Pentecoste"; it is now difficult to continue with this story.
After all, the market for domestic output is still expanding, but the cold war is no longer faced with a market that “has access to scarce premiums as long as there are products”. As a result, the Cold War narrative of the "In Wei Dal Ping" has actually been diluted by capital markets.
Of course, from an operational point of view, there is still considerable room for development in the future. If the company is able to meet its three-year cumulative revenue target of hundreds of billions of dollars, translates large-scale supplies into actual deliveries, and builds stronger ecological advantages for customers, products, and software, it still has the opportunity to become an important pole in China’s AI chip industry.
However, the market value of close to $60 billion has advanced the high growth in the coming years, which is already beginning to face uncertainty in the context of a completely new national production AI.
The good news is that the cold war still has a long time to prove itself.
Ten years ago, when Huang In-hoon handed over DGX-1 to Open AI, Cold War was just a start-up company that had just completed its early financing; ten years later, Cold War proved that it was able to grow rapidly in the cracks of the global AI algorithm.
It is just that moving from a company that seizes the opportunities of the times to a company that can cross the competitive cycle is just another more difficult journey — but this is the path of change that must be followed in the cold war, which is long after the light of capital has faded.
Auto-fetched and translated by AI Atlas. All rights belong to the original author.