Silicon Valley Giants Suddenly Call for a Slower Pace? Don’t Be Deceived by the Surface—It’s a Well-Crafted Game of Territory Dispute
Hello everyone, I’m your financial journalist.
Recently, something quite interesting happened in the AI community: the tech giants who used to compete fiercely to be the first, often to the point of burning out their servers and互相 criticizing (Anthropic, OpenAI, xAI, Google), suddenly all acted in unison and said, “Oh, AI is developing too fast; it’s a bit dangerous. We need to slow down.”
At first glance, it seems like these tech leaders have realized their responsibilities and are acting for the safety of humanity. But if you only see that, you’re being too naive. In the business world, there is no kindness without a motive; everything is driven by profit.
Today, I’ll use simple language to peel back the facade of this so-called “safety concern” and break down the real motivations behind their actions: Why do they suddenly want to slow down? What are their individual plans? And how did China end up being involved in this situation?
---
1. The Surface Reason: “Safety Concerns,” but the Real Reason: “Cash Strains”: Who’s Calling for a Pause?
First, let’s clarify that the companies calling for a slowdown were once the biggest proponents of accelerating AI development.
- Anthropic: Once mocked as a “doomsday prophet,” it has now become a “safety champion.”
- OpenAI: Once the leader in the competitive race, it suddenly announced it won’t go public in 2026 and will focus on safety.
- xAI (owned by Elon Musk): Previously at odds with OpenAI, it now agrees to slow down.
- Google DeepMind (led by Demis Hassabis): Also supports stricter safety standards.
Why this sudden unity? Because they can no longer afford to keep up.
It’s like gamblers in a casino who realize the table is about to collapse or their money is running out. They’re calling for a slowdown not because they fear AI will destroy the world, but because the cost of this “arms race” is becoming unbearable.
1. The Speed of Spending Outpacing Income
The current AI competition isn’t about who’s smarter; it’s about who has the most money.
- OpenAI has a “Stargate” project that requires $500 billion to build data centers.
- Anthropic has a $100 billion contract for computing power from Amazon.
- NVIDIA provided $105 billion in guarantees to OpenAI.
These numbers are enormous. Even established companies like Oracle are spending $28.5 billion in a single quarter just to compete in the AI cloud market, with negative free cash flows.
The key point is this: The models haven’t even been fully monetized, yet the money has already been spent. If this cycle of “you release a new model, I need to build a new data center” continues, everyone’s balance sheets will be in trouble. So, the real purpose of slowing down is to buy time for capital expenditures.
2. A Collective Game of Prisoners’ Dilemma
This creates a typical prisoners’ dilemma:
- Everyone wants the other players to slow down while they speed up.
- But they fear that if they stop first and the others don’t, they’ll lose.
- Thus, they’ve reached a tacit agreement to slow down together.
The direct benefits are:
- Existing products can sell for a few more months: The next generation of models isn’t out yet, so current models can continue to generate revenue.
- Next investments can be postponed: There’s no rush to pour more money into new data centers.
So, don’t believe the talk about “safety for humanity”; believe it’s about cash flow safety.
---
2. Hidden Motives: Each Giant’s “Safety” Agenda Actually Serves Their Own Business Interests
Although they all talk about safety, their real intentions are clear. Their definition of safety serves their own business interests:
- Anthropic: Wants to be the “referee” and turn safety into a barrier to entry.
- Their strategy: By promoting strict safety standards, they aim to make it difficult for smaller companies and startups to compete.
- Benefits: Only giants like Anthropic and OpenAI can afford these standards. Anthropic can use this to strengthen its position and even potentially restrict competitors with regulatory arguments.
- Subtext: “I’m not only better in product but also more compliant. Either follow my costly compliance requirements or be out.”
- OpenAI: Wants to take a break from the pressure of going public.
- Their strategy: By slowing down, OpenAI can delay its public offering. The public market will question its losses and massive investments. If everyone slows down, OpenAI can maintain its lead in private markets.
- Subtext: “Don’t rush me to make a profit before considering the next round of spending.”
- Elon Musk: Both a player in AI and a business owner (SpaceX/Tesla):
- His strategy: Supports slowing down the advancement of AI models while still promoting AI applications.
- Benefits: This gives xAI more time to develop, and SpaceX can continue selling computing power and chips.
- Resource allocation: Musk doesn’t want all capital to go into training larger models; he wants it to flow into his other businesses.
- Subtext: “Stop focusing on model parameters and share some resources with me.”
- Venture Capital (e.g., Sax): Fears that giants will monopolize the market.
- Their concern: If safety standards are too strict, startups won’t be able to enter.
- Logic: VC makes money by investing in new companies. If OpenAI and Anthropic set high barriers, VC will lose investment opportunities.
- Subtext: “You giants can talk about safety, but don’t block our entry.”
3. The White House’s Dilemma: AI’s Rise Affects Ordinary People
This debate in Silicon Valley has quickly spread to Washington, as the costs of AI are being borne by ordinary Americans:
- Rising electricity costs: Data centers consume a lot of energy, driving up electricity prices.
- Cost shifting: Tech companies’ data center construction requires grid expansions, which are passed on to consumers.
- Political implications: Electricity prices became a hot topic in the 2026 midterms, with Democrats criticizing Republicans for sacrificing consumer interests for tech giants and Republicans trying to legislate to control these costs.
- Trump’s “China Card”:
- Trump used China as a scapegoat to argue that slowing down would allow China to overtake the U.S.
- Logic: The focus shifted from “safety vs speed” to “safety/slowing down vs China’s leadership.”
- Tactic: By blaming China, Trump unites the public against those who advocate slowing down, emphasizing the need to keep up with China’s progress.
This move is clever:
1. It frames opponents as “helping China.”
2. It provides a justification for continuing rapid development, even if it means higher costs.
3. It unites the public, regardless of party or stance, on the issue of “not losing to China.”
---
4. The “Hongmen Banquet” for China: The U.S. Wants to Slow Down, but Only If China Does Too
The most critical part of this story is that the U.S. giants’ call for a slowdown is actually a strategy to include China in this slowdown.
- Analogy with nuclear arms control: Anthropic suggests coordinating AI development like the SALT negotiations between the U.S. and the Soviet Union.
- Core logic: AI is a strategic tool that can change power balances. The U.S. wants to slow down for safety but doesn’t want China to catch up.
- Subtext: “If I slow down, you have to slow down too; otherwise, I’ll gain the advantage.”
- The Trap of Rule-Making: If China and the U.S. sit down to negotiate a slowdown, who will control the verification of safety standards?
- The U.S.’s advantage: It has the most mature evaluation systems, advanced chips, and strong cloud infrastructure.
- Risk: If the U.S. sets the standards, it can deny others access to technology.
This is a strategic maneuver: If China agrees to slow down, the U.S. can use safety as a pretext to limit its technological progress.
- European Concerns: Germany and others are worried about being marginalized.
- Their concern: They don’t want to be excluded from AI development and fear being controlled by U.S.-China agreements.
---
5. Conclusion: The Future of AI Shouldn’t Be a Zero-Sum Game Between China and the U.S.
This “collective slowdown” is about more than just AI safety; it’s a complex interplay of industry interests, capital cycles, and geopolitics:
- For the giants: It’s a way to reduce capital pressure and consolidate market positions.
- For the U.S. government: It’s a political tool to balance public opinion and maintain its advantage over China.
- For China: It’s a strategic challenge that requires caution.
What should we do?
- Don’t naively think a slowdown is good: For followers, a leader’s slowdown often means setting obstacles.
- Be wary of rule-making hegemony: The real risk isn’t about the speed of AI but who defines what’s safe and cutting-edge. If the U.S. sets the rules, it can use safety as a tool to restrict China.
- Promote global governance: AI affects everyone; decisions shouldn’t be made by just two countries.
- High-risk technologies (like autonomous weapons) need global regulation.
- Developmental applications (like healthcare, education, industry) should be promoted for global benefit.
- Rule-making should involve multiple countries to ensure a fair and inclusive global AI landscape.
In conclusion, AI is a transformative force that shouldn’t be a tool for a few powers. Its impact should benefit all people, not just a bargaining chip in the hands of a few giants.