Article by Zuihua FunTalk, Author: He Yiran, Editor: Liu Yuxiang
Now, Anthropic is not only the favorite of developers but also of tech leaders.
Because it helps big investors boost the stock price.
On May 6 local time, Musk announced that his artificial intelligence company, xAI, would be formally dissolved as an independent entity and fully integrated into SpaceX as an AI product named SpaceXAI. Simultaneously, SpaceXAI announced a global top-tier computing partnership with Anthropic: the entire computing power of the Colossus 1 supercomputing cluster will be exclusively leased to Anthropic for Claude series model inference services.
Colossus, located in Memphis, Tennessee, USA, is one of the world’s largest and fastest-deployed artificial intelligence supercomputing clusters. The Colossus 1 cluster features over 220,000 NVIDIA GPUs, including densely deployed H100, H200, and next-generation GB200 accelerators.
In return, Anthropic stated that it has "expressed interest in collaborating with SpaceX to develop gigawatts of orbital AI computing power."
It’s not just Musk—Google has long supported Anthropic. On April 25, Google further increased its commitment, planning to invest up to $40 billion in Anthropic and provide 5 gigawatts of computing power. Google has pledged an immediate cash investment of $10 billion at a $350 billion valuation, matching the valuation from its funding round in February. Upon achieving performance milestones, Google will invest an additional $30 billion and significantly expand Anthropic’s computing capacity.
Google’s own Gemini model is among the so-called "Big Three," yet Google has also made significant investments in Anthropic. Public information shows that Google’s parent company, Alphabet, holds approximately 14% of Anthropic. Amazon holds a slightly larger stake of about 15%–16%, making it the largest shareholder, with cumulative investments exceeding $8 billion and providing substantial computing power and cloud services support.
Just before SpaceX’s upcoming IPO, Musk abandoned xAI, which is also easy to understand from a business perspective. xAI has been pushed out of the top tier of AI companies, with a peak valuation of just $250 billion; it makes more sense to lease the Colossus 1 supercomputing cluster to Anthropic in exchange for Anthropic committing to future space computing orders from SpaceX, allowing SpaceX’s business narrative to continue.
After all, Google and Amazon, which invested in Anthropic early, have reaped substantial returns. Alphabet’s first-quarter profit surged 81% to $62.6 billion, while Amazon’s net profit jumped 77% to $30.3 billion, with $28.7 billion and $16.8 billion of those profits respectively stemming from adjustments to Anthropic’s equity value.
In addition to the unrealized valuation of Anthropic, there are also tangible financial commitments: Anthropic has pledged to pay Google $200 billion over the next five years to use its cloud servers and chips, accounting for more than 40% of the future revenue commitments disclosed in Google’s latest earnings call. Amazon Web Services has also accumulated significant compute demand orders from Anthropic.
Google, Amazon, and Tesla (Musk) have all seen strong recent stock performance due to their heavy investments in Anthropic, while Microsoft, OpenAI’s ally and also a minor shareholder in Anthropic’s Series G round, has seen relatively weaker stock performance.
The AI "cold war" may have already erupted, with the two camps being Anthropic-Google-Musk, abbreviated as the A camp; and OpenAI-Microsoft-SoftBank, abbreviated as the O camp. Interestingly, tensions are already emerging within the OpenAI camp.
This is precisely when A is launching a major offensive.
01
On April 28 local time, the trial for Musk's lawsuit against OpenAI officially began at the federal court in Oakland, California.
Elon Musk testified for three consecutive days, accusing himself of having invested substantial funds based on his belief in OpenAI’s nonprofit promises. He is seeking a court order to remove Altman and Brockman from their management roles, restore OpenAI’s status as a purely nonprofit research organization, recover the technological and commercial value equivalent to his $38 million donation, and redirect it to charitable organizations. Most importantly, OpenAI must open-source all AI technologies and disclose the full details of its training data sources.
This lawsuit goes far beyond the personal feud between Musk and Altman—it is a targeted strike against OpenAI. The suit was deliberately filed at a critical juncture as OpenAI prepares for its IPO, clearly aiming to disrupt the commercial progress of this AI industry leader and lay the groundwork for the Anthropic-Musk alliance.
U.S. courts do not allow photography, so the media cannot depict courtroom scenes; however, photos of Musk and Altman in various poses are widely circulated on social media platforms.
These incredibly realistic images were all generated by OpenAI’s ChatGPT Images 2.0, which launched unexpectedly in April. This thought-capable image model achieves “unprecedented specificity and fidelity.” In Altman’s words, it’s like “a leap from cave paintings to the Renaissance.”
Although Image2.0 made a splash in social circles, the capital market has not shown sufficient excitement. The reason is straightforward: Image2.0 has limited availability, and after generating three images, regular users must upgrade to the paid ChatGPT Plus plan. For users seeking novelty, once they’ve gotten their fill, they move on—making it increasingly difficult for OpenAI to generate substantial commercial returns from its large user base.
In terms of traffic share from large models, ChatGPT’s traffic share dropped to approximately 56.7% in March 2026, while Google’s Gemini rose to 25.5% and Anthropic’s Claude increased to 6%. ChatGPT’s previously dominant share has clearly diminished.
More concerning is OpenAI’s revenue-generating ability, as its prolonged losses have gradually eroded investor patience. Third-party statistics show that in the first quarter of 2026, Anthropic captured a 31.4% revenue share of the global large model market, surpassing OpenAI’s 29%.
Keep in mind that Anthropic has only 134 million monthly active users, yet generates an average monthly revenue per user of $16.20, compared to OpenAI’s $2.20. The stark difference in revenue performance between the two companies is not about whether consumer or business models are superior, but rather about whether their products can convince customers to adopt their business models. Anthropic has focused on high-certainty payment scenarios and users with strong willingness to pay, creating a virtuous business cycle.
Another factor affecting revenue is the joint announcement in April by Microsoft and OpenAI that they are officially ending their seven-year exclusive partnership; Microsoft will no longer pay OpenAI a revenue share, while OpenAI’s revenue share payments to Microsoft will continue until 2030.
To improve revenue, OpenAI has updated its advertising business alongside the release of GPT-5.5, launching its advertising platform out of beta and making it fully available to U.S. businesses. Advertisers can now self-register and set budgets and bidding strategies, purchasing ad space on a cost-per-click (CPC) or cost-per-thousand-impressions (CPM) basis.
The AI answer stream will establish a clear two-track system, with ads appearing as sponsored modules after the AI response, intelligently displayed based on the conversation context. OpenAI emphasizes that the advertising system is independent of answer generation; advertisers cannot influence the AI's output and do not have access to users' personal data.
If 2023 was the spotlight year for the AI industry, then by 2026, everyone will be doing the math.
Over the past few years, any startup associated with the "AI" concept could attract capital, and internet giants could endlessly expand their imagination. But by 2026, the myth began to fade, as most AI unicorns turned out to be "bubbles," and the inflated balloons of tech giants were brought back down to earth.
The rules of the era of momentum are gradually coming to an end; the core challenge facing the industry is no longer who can build the most impressive model, but who can convince enterprise customers to willingly pay.
Especially after the release and open-sourcing of DeepSeek-V4, a ruthless "kill line" was established: any proprietary model with performance inferior to DeepSeek-V4 will have zero commercial value.
This has directly and indirectly driven various parties to accelerate their concentration toward the top players like Anthropic and OpenAI, also compelling OpenAI to accelerate its commercialization efforts.
02
Those who haven’t decided which side to support may be the most anxious.
The most typical example is perhaps Meta.
According to the Q1 financial report, Meta's total revenue increased by approximately 33% year-over-year. While this figure appears positive, it conceals a significant crisis—daily active users decreased by about 20 million quarter-over-quarter, marking the first quarterly decline since this metric was disclosed. Meta’s return on AI investment is entirely dependent on improved ad conversion rates and increased user engagement on its social platforms; the decline in daily active users is a highly alarming signal.
Meta has raised its full-year capital expenditure forecast from approximately $115 billion to $135 billion to approximately $125 billion to $145 billion, primarily due to rising component prices and increased costs for new data centers. This means that Meta’s cost increases are not driven by an intentional expansion of strategic advantages, but rather by the necessity to absorb supply chain price hikes.
An even more critical blow came from China’s halt of Meta’s $2 billion acquisition of Manus. Under regulatory requirements, Meta must cancel the acquisition, refund the paid purchase price, delete related products and data, and return ownership of Manus’s technology.
At a critical turning point where the AI industry is shifting from "telling stories" to "doing the math," unquantifiable uncertainty has become the capital market's most intolerable weakness. After the earnings report, Meta's stock price dropped by approximately 7%, erasing nearly all of its gains since the beginning of the year.
In comparison, Google was more favored by the capital markets in its first-quarter earnings report.
Although Gemini's progress has been the slowest among the "Big Three," Google's rich ecosystem and diverse use cases have deeply integrated the Gemini model into its search, cloud, and subscription systems, turning it into an accelerator for existing business growth—monthly active paying users for Gemini Enterprise increased by approximately 40% month-over-month, and enterprise AI solutions are becoming a primary driver of cloud revenue, with search advertising continuing to grow strongly under AI's influence.
More importantly, it has also invested in Anthropic and secured a $100 billion order from Anthropic, which the market has responded to positively.
Amazon has leveraged its cloud business advantages to adopt a risk-hedging strategy in the AI infrastructure era. While investing in OpenAI and making it AWS’s exclusive third-party cloud distribution partner, Amazon has also maintained its multi-billion-dollar compute agreement with Anthropic. In Andy Jassy’s view, he does not bet on which model will ultimately prevail; instead, he ensures that all potential winners must operate through AWS.
The value of AI must ultimately be validated through commercial implementation. OpenAI’s original vision of AGI was based on the logic of global monopolization, but this approach has largely been proven unsuccessful in practice. The construction of AI infrastructure and compliance issues require a decentralized, region- and domain-specific global strategy.
Anthropic’s rapid rise is clearly reshaping Silicon Valley’s power structure. Over the past year, Anthropic’s valuation has increased tenfold, reaching a peak valuation of $1.2 trillion, making it the hottest super unicorn. Revenue in the first quarter of 2026 surged 80-fold year-over-year, far exceeding its original target of a 10-fold increase, with full-year 2026 revenue projected at $18 billion.
In 2023, OpenAI surged ahead; in 2024, global competition intensified; in 2025, the "Big Three" emerged, and this year a duopoly is taking shape.
It's quite possible that Anthropic will lead next year, and it's not impossible.
03
The anchor point of AI narratives is shifting from NVIDIA to Anthropic.
Over the past two years, NVIDIA’s GPUs were the starting point for every large model story—those with the most compute held the power. But now, compute itself is becoming commoditized; what’s truly scarce is no longer chips, but leading application-layer companies that can turn compute into sustained revenue. Anthropic is the rightful heir to this new narrative.
The most subtle shift has occurred on the capital side. Today, Google, Amazon, and SpaceX have all made "investing in Anthropic" a central narrative in their earnings reports, tied to investment gains and compute orders. This deep integration goes beyond ordinary financial investment—Anthropic’s valuation changes directly impact the quarterly profits of these three giants, as well as future expectations.
Data speaks volumes. In Alphabet's first-quarter net profit, more than half came from unrealized gains on equity investments, with Anthropic and SpaceX contributing the largest shares. Amazon's first-quarter net profit surged 77% to $30.255 billion, with $16.8 billion in pre-tax gains attributable to Anthropic's valuation adjustment. Although SpaceX is not publicly traded, its upcoming IPO roadshow highlights annual cloud computing revenue of $3 to $4 billion from Anthropic and promised space computing orders as key narratives supporting its estimated valuation of $1.75 trillion to $2 trillion.
Once this structure becomes entrenched, Anthropic will no longer be merely an AI company, but a "systemic financial node." Any negative news—whether technical bottlenecks, security incidents, or key talent departures—will no longer be solely its own concern, but will simultaneously impact the technology sector through three transmission channels: Google’s equity investment valuation, Amazon’s AWS computing order fulfillment expectations, and SpaceX’s IPO pricing logic.
In contrast, Microsoft, an ally of OpenAI, has seen relatively weak stock performance precisely because it lacks this "equity-order-profit" feedback loop. Microsoft and OpenAI ended their exclusive partnership in April and no longer share in OpenAI’s revenue streams, retaining only OpenAI’s ongoing payment obligations to Microsoft. No matter how high OpenAI’s valuation rises, Microsoft’s balance sheet cannot reflect any corresponding investment gains.
When the entire market’s capital builds a positive feedback loop around Anthropic, Microsoft’s position becomes notably passive.
However, when the entire market bets on the same winner, the risk is simultaneously amplified.
Anthropic’s valuation is predicated on maintaining technological leadership and revenue growth, supported by commitments for compute orders and equity stakes from the three major players. This is a self-reinforcing cycle, yet also a structurally fragile one. If any link weakens—such as the next-generation Claude underperforming expectations, or open-source models significantly closing the performance gap with Anthropic’s models—it could trigger a chain reaction, spreading from Anthropic’s own valuation correction to the financial results and stock performance of the three major players.
The AI industry is transitioning from a fragmented competitive landscape to one dominated by a single leader with several strong contenders, but the capital market seems unprepared to accept a critical reality: when all eggs are placed in one basket, that basket itself becomes the greatest risk.
DeepSeek's open-source "kill line" hangs like a sword of Damocles over Anthropic—so long as the open-source community maintains performance close to or equal to Anthropic's, the commercial value anchor of Anthropic could shift at any moment.
Of course, Anthropic is well aware of this, as are its allies and the Trump administration’s covert allies; therefore, continuing to suppress open-source models like DeepSeek and China’s domestic computing and storage capabilities will be part of their long-term strategy.
This is not just about the U.S. stock market's AI narrative—it's a war over technological supremacy.