The year 2025 marks a pivotal moment in the realm of large models, filled with debate and transformationOver the past year, these models have woven themselves intricately into daily life and work, showcasing their power and potentialOpenAI, a leading player in this field, has achieved remarkable growth with its monthly active users reaching between 500 to 600 million, surpassing platforms like TikTok and making it the fastest-growing product in the history of global technologyThis rapid adoption signifies a growing trend toward the integration of artificial intelligence in various sectors.
However, despite the recognition and excitement surrounding large models, their practical implementation has not quite met industry expectationsCompanies within the sector are now consolidating and re-evaluating their strategiesFor instance, among the “Six Little Tigers” of the large model industry, certain firms have decided to abandon pre-training initiatives while grappling with ongoing rumors of layoffs and restructuring
As 2025 dawns, OneZero and others find themselves amidst turmoil and controversy, indicating an uncertain road ahead.
The financing landscape for large models has seen pronounced fluctuationsIn mid-2024, many of the “Six Little Tigers” sought valuations exceeding 20 billion, indicating some initial success; however, the market stagnated after this peakIt wasn't until the final month of 2024, following a series of shake-ups, that a renewed interest in funding returnedInternationally and locally, significant investments became newsworthy, with xAI securing $6 billion, Perplexity AI completing $500 million in funding, and Zhipu AI landing another $4.5 billion roundThis underlines the volatility and potential of the investment landscape surrounding large models.
In this environment of alternating enthusiasm and skepticism, several startups have navigated the complexities of the previous year, identifying promising pathways forward
- Europe: Recovery and Investment Prospects
- Fund Issuance Revives, Bond Funds Take Center Stage
- Plunging Again! Oil Prices Hit New Lows
- UK Property Profits Hit Decade Low
- Sense of Recession Grips US Despite Data
The pressing question remains: will large models prove to be a viable business in the coming year? Is there still room for growth, especially for burgeoning companies in the sector? Despite a heated funding marketplace, forecasts indicate that by 2024, at least ten unicorns focused on large models may emerge.
According to recent data, there were over 168 instances of financing exceeding 100 million dollars within the global large model industry in the past year, accumulating a total investment of over 400 billion yuanAs noted by Ni Zhengdong, founder of Qianbei Group, at the annual China Equity Investment Conference, "the hottest track in venture capital circles in 2024 is undoubtedly AI." In the global arena, leading companies like xAI, OpenAI, and Anthropic have continued their dominance, securing immense funding rounds and pushing their valuations to staggering heights, with OpenAI breaking the $100 billion mark.
In China, the first quarter of 2024 saw significant investments, such as Dark Side of the Moon securing over $1 billion in Series A+ funding, while MiniMax raised $600 million in Series B funding
These instances highlight a shift in the funding environment, allowing previously stagnant figures to quadruple, setting a new standard that other companies must now contend withAs these companies scramble to prevent falling behind in the quest for Artificial General Intelligence (AGI), they are urgently adjusting their strategies to reach the critical valuation milestone.
While six prominent firms chase innovation, others like Zhongguancun Kejin, Xiaomi's Xiaoice, Yuanxiang, and Sibichai also have valuations exceeding 10 billion, demonstrating that the competitive landscape is both diverse and rapidly evolvingThis landscape is epitomized by the latest ‘2024 Hurun China AI Company 50 Strong,’ where these firms were prominently featured, each carving out a niche in sectors that bridge AI with user needs.
Yet, under this veil of optimism lies a stark reality—the large model industry is not without its controversies
Reports from October suggested that at least two of the "Six Little Tigers" were gradually retracting from pre-training efforts, with corresponding fallout affecting stockholder relationsThe turbulence persists into 2025 with OneZero reallocating personnel and possibly splitting ventures to adaptThe ongoing competition among open-source models further complicates these dynamics, leading most companies to reassess the need to invest heavily in generic large model pre-training.
As observed, the American landscape has narrowed to five key players: OpenAI, Anthropic, Google, Meta, and xAI, while China's large model race has similarly tightened its contendersNotably, many firms are collectively confronting a critical hurdle—the business model of large models remains to be effectively defined and operationalizedIn the consumer market, a "must-have" application has yet to emerge, while in business-to-business contexts, an abundance of projects is met with skepticism regarding the return on investment.
Furthermore, the space for emerging companies becomes increasingly cramped as larger corporations exert market pressures
Rapid advancements have illustrated the capabilities of established firms in outpacing newcomers, emphasizing a harsh competitive realityThe unfolding situation raises crucial questions about whether the market holds sustainable opportunities for new entrants.
Despite these predicaments, some experts remain optimisticChen Yu, a partner at Yunqi Capital, asserts that "there certainly is room for growth." However, he warns that the competition will intensify, likely leaving just a couple of firms standing amidst the frayLi Guangmi, CEO of Shixiang Technology, advises startups to focus on niche markets, indicating that great companies often emerge from less saturated segmentsEach major corporation has inherent weaknesses—notably, Google’s hesitance to directly integrate AI search capabilities in a manner that could disrupt its core services.
As for enterprise-level markets, conditions become even more nuanced and present both challenges and opportunities
Corporate leaders suggest that the transition toward digitalization requires comprehensive large model frameworks, akin to establishing a power grid for the revolutionary electricity market—enabling targeted integration of technologies into operational processesLarge enterprises, often hampered by high overhead, may leave ample space for smaller players who can deliver customized solutionsThe demand for tailored applications persists across diverse sectors, fostering a landscape where collaboration may breed success across business boundaries.
Amidst these developments, further examination reveals the complexity of clients’ needsThe archetype of a successful large model application hinges on the ability to adapt to distinctive consumer requirements, particularly in areas such as intelligent customer service where traditional standards may have fallen shortA significant increase in demand exists for personalized solutions that can scale without the overwhelming expenses typically associated with generic approaches.
Looking ahead, the prominent question is whether businesses can transition from having a foundational large model to developing sustainable commercial franchises that withstand rigorous market scrutiny
As 2025 approaches, the overarching objective shifts toward proving resilience by delivering measurable returns and sustainable growthThe quest for breakthroughs in profitability remains a common priority among industry heads, as many navigate the nuanced landscape of revenue expansion, cost management, and strategic positioning in anticipation of a rapidly evolving paradigm.
Despite emerging challenges, the large model market continues to unveil latent value and revenue generation opportunitiesA recent Sullivan report underscored the accelerating market potential in China, projecting growth from approximately 10.5 billion yuan in 2023 to 16.5 billion yuan in 2024, with a stunning compound annual growth rate of 40% predicted through 2028. Companies worldwide are thus actively pursuing avenues for establishing competitive advantages.
On the international front, Anthropic has cemented itself as a leader in coding applications through advances in its Claude-3.5-Sonnet model, igniting interest amongst software developers and other specialized users
Similarly, companies like Cursor and Bolt.new have recorded substantial annual revenue increases, reflecting a global trend toward monetizing coding capabilities driven by AIIn the rapidly evolving landscape, the emphasis remains on focused innovation and carving distinct paths that resonate with user demands.
The technological landscape is witnessing significant investments concentrated on cognitive applications, particularly as companies harness the burgeoning potential of AIThis trend mirrors broader patterns within high-growth sectors, with firms like Zhipu and Zhongguancun Kejin securing projects that highlight the effective deployment of AI across diverse industriesTheir strategies, primarily targeting enterprise-level markets, underscore a commitment to cultivating meaningful client relationships and creating bespoke solutions that stand to enhance business operations.
While the path for C-end commercialization remains opaque, strategic investment in B-end applications offers a clearer trajectory for success