Since the emergence of the DeepSeek-R1 open-source AI large model, which combines the core attributes of 'low cost' and 'high performance', a massive wave of deploying or locally accessing DeepSeek, a major generative AI application, has swept across the world.

In the financial market, investors are flocking to Software stocks, betting that software companies will have performance data many times stronger than the current figures under the AI boom, and those companies with strong performance growth are expected to replicate the epic 1000% surge in stock prices since October 2022.$NVIDIA(NVDA.US)$The 'style frenzy curve'.

Currently,$Amazon (AMZN.US)$NVIDIA and$Microsoft (MSFT.US)$The DeepSeek-R1 has been launched and deployed on the Cloud Computing power platform, supporting user access to the DeepSeek-R1 large model.$Alibaba (BABA.US)$、 $TENCENT (00700.HK)$、$JD.com (JD.US)$、Kingsoft Cloud (03896.HK).、$Youdao (DAO.US)$with$Zhihu(ZH.US)$Other$China National Software & Service (600536.SH)$Leaders, as well as the three major telecommunications carriers in the country —$CHINA TELECOM(00728.HK)$、$ China United Network Communications (00762.HK)$with$CHINA MOBILE(00941.HK)$, many have announced their integration with the DeepSeek open-source large model.

In recent times, the software stocks at the core of the 'DeepSeek wave' have significantly outperformed both the overall market benchmark and the leading stocks in the market since 2023, namely the 'Semiconductors'.

Looking ahead to the AI application trend, with the new paradigm of 'ultra-low-cost AI training/inference' led by DeepSeek greatly streamlining the deployment workload of large AI models, killer generative AI applications covering both B-end and C-end industries are likely to experience explosive growth, along with 'AI agents' that could significantly boost human productivity. This is also why global funds have recently flowed into software stocks, especially the Chinese software giants with significant valuation advantages.

DeepSeek-R1 has emerged, making a significant announcement to the global open-source large model field as AI training/inference enters a very low-cost paradigm of 'extreme compression + efficient enhanced training + significantly simplified AI inference computing architecture.'

In addition to the recent intensive disclosures of Earnings Reports from technology companies, $Meta Platforms (META.US)$ 、$Palantir(PLTR.US)$、$SAP SE(SAP.US)$And$Applovin(APP.US)$As software companies announce exceptionally strong AI revenue and performance outlooks, institutions such as Goldman Sachs, Bank of America, and UBS Group, along with individual investors, are increasingly convinced of the grand "long-term bull market narrative logic" for software stocks, consequently driving up the stock prices of global software giants.

These bullish funds are betting that there will be more heavyweight "generative AI applications" that surpass the comprehensive performance of DeepSeek-R1 at a lower cost, and that the scale of "AI agent" applications catalyzed by DeepSeek-R1 will see explosive growth.

AI application software is undoubtedly a groundbreaking core vehicle for the penetration of large-scale AI models into human life, capable of achieving transformative improvements in enterprise operational efficiency, as well as in personal work and study efficiencies.

The field of AI application software includes all software products that use AI technology, especially those that leverage generative AI to solve practical problems, featuring capabilities such as natural language processing, predictive analytics, efficiently addressing user inquiries, or automating tedious tasks, thereby directly providing real value to enterprises or individuals, such as improving efficiency and optimizing cumbersome workflows.

From the current technological trajectory, the development direction of AI application software is centered around "generative AI application software" (such as DeepSeek, ChatGPT, Sora, and Claude launched by Anthropic and other globally popular applications), and based on generative AI, the AI functionalities are shifting from a chat-box style Q&A to "AI agents capable of autonomously executing various tedious and complex tasks," such as$Salesforce (CRM.US)$"Agentforce 2.0," the AI agent embedded in Microsoft's Dynamics 365 series.

These two major directions are not only a sign of the progress of AI technology but also represent the main line of the comprehensive revolution of AI in the future.

With the substantial decrease in AI training costs led by DeepSeek, as well as the drastic reduction in Token costs at the inference end, AI agents and generative AI software are expected to accelerate penetration into various industries, which is why Wall Street remains extremely bullish on software stocks even during the sharp decline.$NASDAQ 100 Index(.NDX.US)$Among them, AI agents are very likely to be a major trend in AI applications before 2030. The emergence of AI agents signifies that artificial intelligence is evolving from an information-assisted tool into a highly intelligent productivity tool.

AppLovin demonstrates to investors just how explosive the demand for AI application software is.

AppLovin, known as the 'AI super bull stock,' reported much better-than-expected Q4 2024 results this week. The Advertising Marketing and mobile application technology development company AppLovin saw its stock price increase by an astonishing 700% for the entire year of 2024, with a rise of 46% since 2025, leading the entire US stock market, significantly outpacing AI Chip leader NVIDIA. After announcing its Q4 Earnings Reports, AppLovin's stock rose by 24% by the close of US markets on Thursday, with intraday gains approaching 40%.

As of the time of writing, AppLovin has risen over 5%.

The stock performance of AppLovin since 2024 has been globally shocking, while the core logic behind the surge in stock price lies in the explosive growth in demand for AppLovin's online advertising marketing software solutions based on generative AI. AppLovin can efficiently and highly compatibly match supply and demand amidst a vast number of advertising auctions.

The company has launched the AI advertising engine model AXON 2.0 and developed online advertising software solutions based on generative AI, while providing 'Generative AI + Advertising Marketing' services to advertisers and application developers.

With the gradual formation of AI computing power resources covering AI hardware infrastructure and deployment on the AI large model and application software ends, AppLovin has begun to provide global enterprises with 'Generative AI + Advertising Marketing' based on AI technology, encompassing a one-stop online advertising marketing ecosystem platform that includes user reach, traffic monetization, advertising creative design, and marketing data detection.

Performance data shows that AppLovin's Q4 revenue reached 1.37 billion USD, a year-on-year increase of 44%, with a net income reaching 0.5992 billion USD, a staggering year-on-year increase of 248%, equivalent to an EPS of 1.73 USD, far exceeding market expectations of 1.26 USD.

According to the Earnings Reports, the company’s advertising-related business revenue in the fourth quarter soared 73%, reaching nearly 1 billion USD. AppLovin provides AI-driven advertising software, and management emphasizes that improvements to its advertising AI large model are still in the early stages, with plans to achieve a more personalized advertising delivery model in the future.

The performance expectations set by AppLovin's management are also strong, with revenue expected to be between 1.355 billion USD and 1.385 billion USD in the first quarter of 2025, higher than the market expectation of 1.32 billion USD, among which over 1 billion USD is expected to come from its advertising business sector.

The social media giant Facebook and Instagram's parent company, which has a user base of 3 billion.$Meta Platforms (META.US)$, has recently seen its stock price rise for a record 18 consecutive days, also based on the strong growth prospects of AI revenue brought by 'Generative AI + Advertising Marketing.'

Since the emergence of DeepSeek-R1, the logic behind the rise of the 'seven giants' in the US stock market has fundamentally changed, as investors have begun to strongly question the rationality of the US tech giants' seemingly frenzied AI spending plans. Except for Meta, the stock price performance of other giants has significantly lagged.$S&P 500 Index(.SPX.US)$It has become the core negative catalyst that drags down the entire US stock market's rise.

In the field of digital advertising that Meta relies on, the powerful open-source AI model launched by Meta, which has 3 billion users, along with various generative AI-assisted software tools, drives advertisers to reach a larger potential user base, bringing a brand-new AI-based advertising recommendation experience for Meta's advertisers and users. This is also an important logic behind Wall Street's expectation that Meta's stock price will continue to rise.

Digital advertising is the core revenue engine for Meta, and 3 billion users can be regarded as its cornerstone. Meta's AI advertising assistance tools and Meta AI have helped Meta's advertising business revenue exceed expectations for several consecutive quarters. Survey data shows that most advertisers plan to increase or maintain their advertising spending on Meta's digital advertising platform. Therefore, Meta's exclusive AI ecosystem may help it continue to embark on a positive growth trajectory in 2025 and beyond.

Focusing on 'AI + Data Analysis', the American data software giant.$Palantir(PLTR.US)$After announcing its earnings report in early February, the stock price has risen over 40%. In 2024, the company's stock price soared by about 340%. The stock price of this software giant has indeed ridden the wave of investors' immense enthusiasm for AI, and more and more business and government clients are starting to use Palantir's 'AI + Data Analysis' software ecosystem.

As of the time of writing, Palantir has risen over 1%, reaching a historical high.

Palantir's generative AI platform 'AIP' is fully integrated with Palantir's existing data analysis software ecosystem, allowing clients to access Palantir's core modules and functions through a simple Q&A method, enabling organizations to effectively apply generative AI to data analysis, enhancing insights and operational efficiency.

Palantir, with its AI-driven data analytics platform, not only dominates in the field of National Security but has recently shown an incredibly strong growth potential on the demand side in the American commercial market. Palantir's AI software ecosystem covers almost all military departments in the USA and has expanded to American allies. With the explosive growth in demand for commercial AI software, Palantir is leading a revolution in AI software, rather than being fully dominated by high-performance AI Chips.

According to the Earnings Reports, the company's sales in the fourth quarter surged by 36%, reaching $0.8275 billion, with commercial sales in the American market increasing significantly by 64%, totaling $0.214 billion. Regarding the performance outlook for the commercial business, Palantir expects that sales in the US commercial sector will grow by approximately 54% by 2025, reaching $1.08 billion.

A well-established software giant in the USA.$Salesforce (CRM.US)$Recently stated that due to the strong demand from customers for the new AI agent product Agentforce 2.0, the company plans to hire more professional AI product personnel to sell this new AI agent product.

It is reported that Salesforce, while announcing layoffs, is also hiring professional staff to sell its latest AI products, highlighting the accelerated expansion of user scale for Salesforce AI application software, especially the newly launched AI agent product, which has prompted the company to allocate more resources to the AI product department.

Salesforce's Executive Vice President Mike Spencer stated in an interview last December that Salesforce has signed "a substantial number" of enterprise deals related to the company's AI agent product - Agentforce. The analysis team from Citigroup recently commented in a report on Salesforce's growth prospects, noting that "Salesforce's AI agent tools are starting to replace the traditional narrative logic of Customer Relationship Management software (CRM)."

The German software giant.$SAP SE(SAP.US)$The Q4 cloud computing business revenue announced at the end of January increased by 27% year-on-year, reaching 4.71 billion euros, highlighting the strong demand for AI applications, AI software development, and cloud AI inference computing power in the global layout of AI. Many enterprise customers are starting to configure SAP AI applications and flocking to the SAP cloud platform to develop AI application software and deploy inference computing resources. SAP's AI strategy helped the software company leap to become the most valuable technology company in Europe last year, significantly surpassing the Dutch lithography giant.$ASML Holding (ASML.US)$。

With the blessing of DeepSeek, the wave of AI applications will sweep the globe! Software giants are entering a golden era.

DeepSeek's low cost + ultra-efficient + performance comparable to OpenAI's model o1 is attributed to the application of 'extreme engineering' and 'fine-tuning' to every part of the large model training process, along with the introduction of pure reinforcement learning instead of traditional supervised learning (SFT) based on a large number of examples, significantly reducing the training/inference cost of large models.

The efficient training and data compression strategy oriented towards extreme engineering utilizes multi-layer attention (MLA) — especially performing low-rank transformation on the Query side to reduce activation memory burden during training. It also includes FP8 mixed precision training, DualPipe parallel communication, MoE load balancing, etc., allowing DeepSeek to maximize hardware resource utilization during the training phase, reducing 'unnecessary computing power waste', along with innovative AI training initiatives such as 'reinforcement learning (i.e., RL) + distillation + specialized data optimization', creating a low-cost AI computing power new paradigm of 'extreme compression + efficient reinforcement training + significantly simplified AI inference computing architecture'.

With an extremely low investment cost of less than 6 million dollars and under the performance conditions of H800 chips that are far inferior to H100 and Blackwell, the DeepSeek team has built an open-source AI model with performance comparable to OpenAI's o1. In contrast, Anthropic and OpenAI's training costs reach up to 1 billion dollars. DeepSeek's inference input and output token pricing compared to OpenAI's pricing can be described as 'deeply discounted'. DeepSeek charges only 2.19 dollars per million output tokens, while OpenAI's GPT-4 prices reach up to 60 dollars.

As the "low-cost computing power new paradigm led by DeepSeek sweeps the globe", AI training and application inference costs are increasingly decreasing, while comprehensively driving AI application software to accelerate penetration into various industries worldwide, fundamentally innovating the efficiency of various business scenarios and significantly increasing sales. The AI revenue data and profits of software giants such as Meta, SAP, Palantir, Salesforce, as well as AppLovin, Alibaba, and TENCENT may experience exponential growth, making them the biggest winners under the global popularity of DeepSeek and the "DeepSeek low computing cost shockwave".

Wall Street financial giant Bank of America has named AppLovin, the leader in "Generative AI + Advertising Marketing", as a "preferred stock" for 2025, showing great optimism for the growth model driven by generative AI in the software business. Bank of America has given AppLovin a target stock price of up to $580 for the next 12 months. In comparison, AppLovin's share price rose by 24% to $471.67 on Thursday.

SAP's management recently stated that as customers' demand for emerging technologies continues to grow, software companies are focusing more on AI agents that can complete tasks without supervision.

The analysis team from Bank of America stated that some AI agent tools have already demonstrated Ph.D.-level intelligence on certain tasks, and it is expected that by the second half of this year, far-reaching practical AI agent applications will emerge, which may help software companies in advantageous positions capture a share of the $12 trillion traditional services market.

Bank of America indicated that the DeepSeek open-source model has lowered the barriers to AI applications, enabling more software companies to quickly integrate high-performance AI large models, significantly enhancing social productivity with Saas and consequently driving substantial growth in Saas software subscription scale.

The urgent demand from enterprises to improve efficiency and reduce operational costs has greatly propelled the widespread application of AI agents recently. AI agents can automate repetitive tasks, perform big data analysis and summarization based on extremely powerful AI large models, provide real-time monitoring insights, and make appropriate decisions in very short times in response to extremely complex situations, thereby enhancing business operational efficiency.

For personal learning and work efficiency, the logic of enhancement is fundamentally similar. AI agents can also efficiently participate in large projects across various fields globally, from blueprint planning to implementation, significantly accelerating project progress.

According to the latest forecast report by the research firm MarketsandMarkets, the market scale of AI agents is expected to expand significantly from just $5.1 billion in 2024 to approximately $47.1 billion by 2030, with a compound annual growth expectation as high as 44.8% during the period.

The institution stated that by using AI agents, companies can automate various very complex processes and reduce human intervention to avoid frequent human errors. The integration of AI agents with enterprise-level internal automation tools will achieve a one-stop AI processing solution from formulating product sales plans to after-sales service, significantly enhancing business efficiency.

According to a recent report released by Bloomberg Industry Research Analysts, by 2032, the total revenue of the entire Generative AI market, covering markets such as AI Chips, Ethernet Switch Chips and Generative AI Software and AI Agents, is expected to grow from approximately $40 billion in 2022 to $1.3 trillion, indicating that this market is likely to expand 32 times over the next decade with a high compound annual growth rate of up to 43%.

In 2025, will global funds focus on Chinese software giants?

Compared to the strong performance of U.S. software stocks, which generally hover around historical highs in valuation, Wall Street investment giants appear to believe that U.S. software companies have basically validated the strong performance growth logic brought by the explosive demand for 'AI application software' driven by Generative AI Software and AI Agents. As DeepSeek integrates into various fields, the resulting 'super wave of AI applications' reinforces the upward momentum of Chinese software giants, which have lower valuations, especially in 2025.

Several top-tier research institutions on Wall Street are bullish on global software stocks, particularly maintaining a long-term bullish narrative logic for Chinese software giants (such as Alibaba, TENCENT). The new paradigm led by DeepSeek has lowered the entry barriers and inference costs for AI applications, helping companies deploy large-scale AI applications at lower costs, thus driving an innovative upgrade in sales of SaaS Software subscription services and the overall Cloud Computing market.

Wall Street giant Goldman Sachs, after being severely impacted by the 'DeepSeek shockwave' that hit U.S. stocks, reiterated its bullish enthusiasm for the Chinese stock market. Goldman Sachs predicts that under bullish market expectations, the MSCI Chinese Index will soar by 28%, emphasizing that stocks in the 'Software Technology' sector are expected to outperform the entire Large Cap market. The MSCI Chinese Index includes Alibaba, TENCENT,$Kweichow Moutai (600519.SH)$And$China Yangtze Power (600900.SH)$It has hovered around 66-70 points since February, similar to China's core assets.

In a recent research report released by Bank of America, it pointed out that as AI applications fully spread to B-end and C-end, the market focus has shifted from semiconductor giants like NVIDIA to the second phase of the AI investment boom, focusing on generative AI and AI software stocks, as well as network terminal devices, electrical utilities supply equipment, and liquid cooling facilities provided by datacenter infrastructure providers.

As Alibaba and Tencent showcase their labels as 'encompassing cutting-edge AI models + powerful cloud AI computing systems + complete AI application software developer platforms', and as the scale of these labels grows with AI application software penetrating various industries in China, the future market scale is expected to rival that of Amazon AWS and Microsoft, possibly triggering a similar investment boom that attracted global capital to North American cloud computing giants in 2023-2024.

In comparison, Alibaba's current market cap is only 280 billion dollars, while Amazon and Microsoft reach as high as 2.45 trillion dollars and 3 trillion dollars respectively.

The new AI model technology path led by DeepSeek, focusing on 'low cost' and 'high energy efficiency', is expected to bring an overall decline in costs across the Industry Chain. For Chinese technology companies focusing on AI applications such as major Internet firms and Consumer Electronics companies, as AI applications accelerate their penetration into various domestic industries, leading to significant productivity changes, there will undoubtedly be more market opportunities and a much larger demand for AI application and reasoning computing power.

If killer-level AI Application software/AI agents start to emerge on a large scale from 2025, it would be a significant bullish signal for cloud giants like Alibaba, TENCENT, and JD.com, as these AI software, whether it's the early-stage AI software developer ecosystem platform or the incredibly vast cloud-based AI inference computing resources later on, rely on the powerful computing platform support provided by these cloud giants.

These cloud giants focus on building an ecosystem for B-end and C-end AI application software developers related to generative AI, aiming to comprehensively lower the technical threshold for non-IT professionals in various industries to develop AI applications.

The international major bank UBS Group released a latest research report offering a list comparing Chinese and American technology stocks, to help investors better understand the potential and value of Chinese technology companies. This list quickly became popular in various stock forums domestically.

In this list, the UBS analysis team compares Cambrian to NVIDIA, and...$XIAOMI-W(01810.HK)$compared to$Tesla (TSLA.US)$，$Semiconductor Manufacturing International Corporation (00981.HK)$compared to$Taiwan Semiconductor (TSM.US)$Tencent is benchmarked against Meta, while Alibaba is benchmarked against the world's leading Cloud Computing market share holder, Amazon. UBS Group emphasizes that software stocks in the Chinese stock market will be the core beneficiaries of the "DeepSeek integrates all things" major trend, especially in areas such as enterprise workflow automation, knowledge management, and vertical digital solutions.

The UBS Group analysis team states that the development of the AI industry typically drives a significant increase in the valuation of related stocks. In the past, during the 4G, 5G, and Cloud Computing eras, the performance of related stocks outperformed the overall market by 50% to 100%, and this rebound usually lasts for 1 to 2 years.

Referring to the peaks triggered by Cloud Computing from 2019 to 2020 and AI before 2023, UBS Group found that the valuation of Chinese software companies grew the most, with the P/S valuation increasing significantly from 4x to 14x. Currently, software stocks are only at 53% of their 2021 highs, and are also down from 38% in 2023. Even if the P/S exceeds the 2021 highs, the increase has not reached the benchmark levels of the 4G, 5G, and Cloud Computing eras, while Hardware stocks have performed close to their highs during that period.

The UBS Group report noted that since the release of the DeepSeek R1 model, Chinese AI-related stocks have risen about 15%, clearly outperforming the MSCI Chinese Index's 9%. UBS Group predicts that the market uptrend related to AI in the Chinese stock market may not have reached its peak yet, especially for software stocks like Alibaba, which have substantial future valuation enhancement potential.

Want to learn more about stocks benefiting from DeepSeek? Make good use of the 'Investment Theme' feature, go to Futubull>US Stocks>Investment Themes>DeepSeek Concept Stocks, and capture investment opportunities!

Editor/ping