In the past decade, the alliance between Silicon Valley and Wall Street has formed a classic dual oligopoly narrative, with Silicon Valley crafting stories through technological breakthroughs and Wall Street amplifying expectations via financial leverageThis narrative framework must continually evolve, requiring a steady infusion of new material and fresh plotsIn the last two years, artificial intelligence (AI) has unmistakably taken center stage in this evolving tale.

However, come 2025, this well-established narrative may face significant shifts.

While tech giants like Google and Meta plan to invest billions into building massive computing clusters, a company named DeepSeek has demonstrated that equivalent performance can be achieved using only one-third of the computational resourcesThis indicates that merely piling on processing power is no longer the only pathway to intelligent evolution; the efficiency of algorithms is emerging as a new valuation benchmark.

The rise of OpenAI serves as a prime example of this narrative shift.

Founded in December 2015, OpenAI was valued at only $1 billion in 2019. However, following the launch of ChatGPT at the end of 2022, its valuation skyrocketed to $20 billion

Advertisements

By October 2024, after completing a fresh round of financing totaling $6.6 billion, OpenAI's valuation soared to an impressive $157 billion, despite still being far from profitability.

Other leading AI firms are similarly maneuvering for substantial funding at extremely high valuationsFor instance, Elon Musk's xAI, established in 2023, received $6 billion in a Series C funding round in December 2024, with a valuation exceeding $40 billionAnthropic is also on the hunt for an additional $2 billion in funding at a staggering $60 billion valuation.

This model is underpinned by the notion that technological breakthroughs require vast amounts of funding to support computational power and talent acquisition, while capital market valuations depend on the imagined potential of a "technological moat."

A report published by PitchBook, a U.Sventure capital research firm, revealed that in January of this year, U.S

Advertisements

AI startups garnered approximately $97 billion in venture capital financing in 2024, accounting for nearly half of the $209 billion raised by all startups in that year, with most of these AI firms emerging from Silicon Valley.

The generosity of venture capital institutions stems from their belief that these Silicon Valley AI companies, backed by a robust capital market, would yield substantial returnsIn fact, this belief has proven to be valid.

Even amidst recent market adjustments over the past couple of weeks, companies like Tesla and Nvidia maintain exceptionally high valuationsCurrently, Tesla's price-to-earnings (P/E) ratio stands at an astonishing 170 times, as investors now view it not merely as an electric vehicle manufacturer, but as Musk repeatedly asserts, as an AI companySimilarly, Nvidia boasts a P/E ratio of 45 and Microsoft 33.

The seven major U.Stech juggernauts require new material and narratives to continue captivating investors with their compelling stories

Advertisements

Over the past couple of years, AI has acted as the stabilizing anchor for their inflated valuationsConsequently, the “tech-capital” paradigm has evolved into an “AI-capital” model.

Yet, the fragility of this AI-capital loop lies in its complete reliance on the positive feedback cycle of "technological leadership" and "capital returns." When DeepSeek achieved model performance comparable to OpenAI's multimillion-dollar investment with only $5.5 million in training costs, the AI narrative mythos crafted by the collaboration of Silicon Valley and Wall Street came under intense scrutiny.

More critically, DeepSeek's open-source strategy directly dismantled the scarcity premium associated with closed-source models, which represents a core variable in the AI-capital valuation model.

The paradigm shift instigated by low-cost AI is notable.

DeepSeek's disruptive innovation doesn’t stem from generational technological leaps, but from an extreme optimization of existing resources

• Corporate Bond ETF vs. Benchmark Treasury ETF
• Navigating Uncertainty: The Fed's Challenge
• Major Launch of Credit Bond ETFs!
• Bank of Japan Signals Interest Rate Hike
• Nvidia's Stock Surge Highlights Its Dominant Position

Its V3 model enhances unit performance output by over twofold through heterogeneous computing architecture and dynamic load balancing algorithms, with training costs only a fraction of the industry averageThis propels a new technological route of low-cost, high-efficiency.

DeepSeek's "efficiency revolution" swiftly exposed the flaws of Silicon Valley's "compute power and data pileup" trajectory: preliminary costs for training GPT-5, in development by OpenAI, have reportedly exceeded $1 billion, while DeepSeek's model, the DeepSeek-R1, achieves equivalent performance at a cost of merely $600,000, thereby lowering the industry's valuation anchor.

This efficiency revolution has triggered a chain reaction in the capital marketsNvidia's stock price plummeted over 20% within a weekAlthough this decline is partly due to short-term fluctuations resulting from high valuations prompting capital flight, it reflects a burgeoning trust crisis concerning the logic of "burning cash for growth." This marks a crack within the Silicon Valley and Wall Street AI narrative framework that DeepSeek's new technological avenue has blown open.

The AI-capital narrative fundamentally embodies a conspiracy between "technological idealism" and "capital expansion needs." Microsoft’s plan to invest approximately $80 billion to build AI data centers in the 2025 fiscal year, along with OpenAI and Nvidia's collaboration with SoftBank to launch the $500 billion "Star Gate," reflects this collusion.

DeepSeek's historic contribution lies in its open-source strategy, which demonstrates the feasibility of high-cost-performance AI products through engineering practice

This change permits numerous small and medium-sized enterprises and developers to enter the arena, thereby shifting the AI application ecosystem from an oligopoly to a more inclusive participation.

As a result, the AI narrative spun by Silicon Valley and Wall Street is no longer as melodious.

In the face of this upheaval, both Silicon Valley and Wall Street are attempting to cling to their original narrative fabric.

For instance, OpenAI and Meta are engaging in "debunking" efforts to uphold their technological authority, yet they find themselves the subject of collective ridicule; Anthropic executives are appealing to U.Sauthorities for tighter chip export laws in an attempt to employ geopolitical measures to sustain technological superiority.

However, such resistance is unlikely to alter the fundamental trendThe DeepSeek phenomenon uncovers not only a new choice in technological pathways but also an industrial logic paradigm shift: open-source is becoming an unstoppable new wave in AI modeling

The decentralized ecosystem it fosters is set to continually challenge the closed-source hegemony in Silicon ValleyEven OpenAI’s CEO, Sam Altman, has conceded that the company's closed-source strategy stands on the "wrong side of history."

IBM’s CEO Arvind Krishna recently remarked that AI has long been viewed as a scale-driven game — the larger the model, the better its capabilitiesHowever, DeepSeek teaches us that the best engineering design should optimize for both performance and costOnly when a technology becomes economically viable and easily accessible can it truly effect transformative change.

A more essential question arises: Is the ultimate goal of the AI revolution to serve capital valuations, or to enhance human productivity? The value of DeepSeek is not merely in outpacing OpenAI but in proving that the true measure of technological progress is not the amount of capital raised or the number of GPUs deployed, but the reduction in costs and the greater degree of accessibility.

As the fog of capital accumulation dissipates in the AI sphere, only those enterprises that translate algorithmic precision into societal efficiency will emerge as winners in this new era