Circular capital: Inside the closed-loop ecosystem propelling (and distorting) the AI boom

Circular capital: Inside the closed-loop ecosystem propelling (and distorting) the AI boom

The artificial intelligence sector is experiencing an unprecedented surge, driven by what many observers describe as an arms race among tech giants and startups alike. Major players like Microsoft, Amazon, Nvidia, and Oracle are pouring billions into promising AI ventures such as OpenAI, Anthropic, and Scale AI, creating intricate funding ecosystems that blur the lines between investment and self-serving commerce.

These startups, in turn, funnel much of that capital back into the investors’ own products, including cloud computing services, specialised chips, and data infrastructure. This circular flow of money strengthens the positions of a handful of dominant companies while raising serious questions about competition and the efficient use of resources in a field still in its early stages.

Circular capital loops

This setup resembles a high-stakes poker game where the house always wins, potentially stifling innovation from smaller players and inflating valuations beyond sustainable levels. The industry appears to operate on the belief that AI could evolve into a winner-take-all market, justifying these closed loops as a necessary hedge against being outpaced.

Recent reports indicate OpenAI’s valuation has climbed to around 324 billion dollars, with Anthropic not far behind at 178 billion dollars, figures that underscore the rapid escalation in private market enthusiasm. Scale AI, meanwhile, maintains a valuation near 29 billion dollars, often tied more to projected spending on infrastructure than to immediate revenue streams.

Regulatory scrutiny mounts

Regulatory scrutiny is intensifying as these dynamics unfold, with authorities expressing growing alarm over market concentration and potential antitrust issues. Nvidia, commanding over 80 per cent of the AI chip market, faces investigations from the US Department of Justice regarding its acquisition of Run:ai, a move that could further entrench its dominance.

The Financial Stability Board has issued warnings about the systemic risks posed by AI’s heavy reliance on a limited number of infrastructure providers, highlighting vulnerabilities in areas like cybersecurity and model governance that could cascade through the financial system. In my view, these concerns are well-founded, as the concentration of power in a few hands echoes past tech bubbles where over-dependence on key suppliers led to widespread disruptions.

Capital allocation risks

The circular capital loops exacerbate this, as seen in deals where OpenAI commits to massive spending on Oracle’s cloud services following investments from similar tech behemoths. While analysts remain optimistic about AI’s transformative potential in the long term, they caution against short-term returns hampered by regulatory hurdles and inefficient capital allocation.

The risk of overvaluation looms large, with private AI firms’ worth often predicated on future infrastructure expenditures rather than proven profitability, a pattern that could precipitate corrections if growth expectations falter.

Macro market backdrop

Shifting to broader economic indicators, global risk sentiment stays subdued as markets await new developments amid worries ranging from labor market slowdowns to persistent inflation. Investors are closely monitoring upcoming US initial jobless claims data, with estimates around 233,000 following last week’s 231,000, a figure that could sway perceptions of the Federal Reserve’s policy direction.

The Swiss National Bank recently held its policy rate at 0.00 per cent, aligning with expectations and reflecting a cautious approach to monetary easing in the face of stable inflation. Wall Street closed lower on Wednesday, with the Dow Jones Industrial Average down 0.37 per cent at 46,121, the S&P 500 off 0.28 per cent at 6,638, and the Nasdaq declining 0.34 per cent to 22,498, driven by retreats in technology stocks amid valuation concerns.

Wall Street and commodities

Treasury yields edged higher, with the 10-year note at 4.147 per cent and the 2-year at 3.604 per cent, signalling mixed expectations for interest rate paths. The US dollar index strengthened by 0.6 per cent to 97.873, while gold prices dipped 0.7 per cent to 3,736 dollars per ounce, pulling back from recent highs as the dollar gained ground. Brent crude rose 2.5 per cent to settle at 69.31 dollars per barrel, buoyed by supply concerns from ongoing geopolitical tensions in Ukraine impacting Russian oil facilities.

Asian equities showed mixed performance, with Chinese markets buoyed by AI and tech optimism, though early trading today indicated continued variability. US equity futures point to a higher open, suggesting some rebound potential. In my opinion, this muted sentiment reflects a market grappling with uncertainty, where AI hype provides sporadic lifts but broader economic signals like job data and yields temper enthusiasm, potentially setting the stage for volatility if inflation proves stickier than anticipated.

Crypto under pressure

Turning to cryptocurrencies, contrary to chatter among some circles that altcoins are outperforming Bitcoin, the data paints a different picture of weakening momentum for alternatives. The CoinMarketCap Altcoin Season Index stands at 68 out of 100, still in altcoin territory but down 4.23 per cent over the past 24 hours from last week’s 77, indicating a cooling trend.

Bitcoin’s dominance has risen to 57.97 per cent, up 0.25 points in the last day, as capital shifts toward the flagship cryptocurrency amid altcoin retreats. Ethereum, a bellwether for the sector, has fallen 11.6 per cent weekly, with Chainlink down 11.2 per cent and Cardano dropping 12.0 per cent, underscoring broader underperformance.

Derivatives markets reinforce this caution, with altcoin funding rates turning negative at -0.00035835 per cent and open interest declining 4.1 per cent in 24 hours, compared to Bitcoin’s more resilient metrics.

Investor takeaway

From my standpoint, this shift signals a risk-off environment in crypto, where Bitcoin’s perceived safety draws inflows during uncertainty, much like gold in traditional markets. Historically, Altcoin Season Index readings dipping below 70 often herald Bitcoin dominance rebounds, and current social discussions around Ethereum’s high fees and upcoming upgrades like Pectra in Q4 2025 add to the drag.

Traders unwinding leveraged positions faster in altcoins than in Bitcoin further erodes confidence in near-term rallies for alternatives, suggesting investors should prioritise Bitcoin amid this rotation.

Overall, the interplay between AI’s frenetic funding cycles, emerging regulatory pressures, subdued macro conditions, and crypto’s Bitcoin-centric tilt illustrates a financial landscape fraught with opportunity and peril.

I believe the AI arms race, while fuelling innovation, risks over-investment that could echo the dot-com era’s excesses if not tempered by competition and oversight. Investors would do well to diversify beyond concentrated bets, monitoring systemic risks and market signals closely to navigate what may prove a pivotal juncture for technology-driven growth.

 

Source: https://e27.co/circular-capital-inside-the-closed-loop-ecosystem-propelling-and-distorting-the-ai-boom-20250925/

Anndy Lian is an early blockchain adopter and experienced serial entrepreneur who is known for his work in the government sector. He is a best selling book author- “NFT: From Zero to Hero” and “Blockchain Revolution 2030”.

Currently, he is appointed as the Chief Digital Advisor at Mongolia Productivity Organization, championing national digitization. Prior to his current appointments, he was the Chairman of BigONE Exchange, a global top 30 ranked crypto spot exchange and was also the Advisory Board Member for Hyundai DAC, the blockchain arm of South Korea’s largest car manufacturer Hyundai Motor Group. Lian played a pivotal role as the Blockchain Advisor for Asian Productivity Organisation (APO), an intergovernmental organization committed to improving productivity in the Asia-Pacific region.

An avid supporter of incubating start-ups, Anndy has also been a private investor for the past eight years. With a growth investment mindset, Anndy strategically demonstrates this in the companies he chooses to be involved with. He believes that what he is doing through blockchain technology currently will revolutionise and redefine traditional businesses. He also believes that the blockchain industry has to be “redecentralised”.

j j j

Web4 Explained: A Vision for Practical, AI-Integrated Blockchain- Anndy Lian

Web4 Explained: A Vision for Practical, AI-Integrated Blockchain- Anndy Lian

Rejecting calls for ideological purity in Web3, Singapore-based fund manager and intergovernmental advisor Anndy Lian has unveiled a vision for “Web4,” a practical, AI-integrated internet. Speaking at Taipei Blockchain Week 2025, Lian argued that security, user-centric design, and financial incentives—not abstract ideals—are the keys to driving mainstream adoption.

Key Points

  • Introducing Web4: AI-native, decentralized framework focused on usability and practical application over ideology.
  • Zero Data AI Architecture: AI operates without storing raw user data, with blockchain verifying privacy.
  • Incentives Over Ideals: Financial rewards, such as AI-powered trading agents, are the most effective driver for mainstream adoption.

Cutting Through the Hype

Web3 has long promised a decentralized internet, but Lian argues that many platforms only offer an illusion of decentralization, replicating centralized power structures behind a blockchain veneer. Speaking on the “Infra Wars” panel, he presented a grounded approach, prioritizing functionality and security over rigid adherence to decentralization for its own sake.

“Full decentralization, you know right now, remains a big challenge,” Lian said. “I just want to keep things very simple: computing part is definitely a must, storage if you can do it decentralized, I think it’s great, but we should always find ways to make sure that the security part of things in the infrastructure is well managed.”

Lian’s stance positions him as a pragmatist in an industry often dominated by idealists. Success, he argues, is measured not by how decentralized a system is, but by how effectively it operates without hacks or user losses.

Introducing Web4

Lian coined the term Web4 to describe a next-generation internet built with AI as a native infrastructure component. Autonomous AI agents could function as independent economic participants, such as AI-powered liquidity providers or community moderators that operate and earn within the system without direct human intervention.

“If there’s a chance for us to redo it again with all these AI experts, something like Web4 will be a lot better,” he said.

A cornerstone of this vision is the “zero data AI architecture.” In this model, AI operates without storing raw user data, while blockchain serves as a trust layer, cryptographically verifying that user data was not retained. This approach addresses privacy concerns while allowing AI to function efficiently—a balance between Big Tech data monopolies and the ideals of full decentralization.

Driving Adoption Through Incentives

Beyond technical design, Lian emphasized the challenge of bringing mainstream users to decentralized platforms. Economic incentives, not ideology, are the most effective driver.

“The best way for people to experience AI and blockchain is to teach them how to make money,” he said, highlighting AI-powered trading agents as a practical entry point. This focus reflects a broader industry shift toward creating tools with real-world utility, moving beyond speculative applications.

The Pragmatic Path Forward

Ultimately, Lian’s Web4 framework proposes a middle path: balancing technological ambition with human behavior and market reality. “Success isn’t about AI or decentralization alone,” he concluded. “It’s about protecting users, creating value, and making technology approachable. Web4 is my roadmap for that balance.”

 

Source: https://news.shib.io/2025/09/09/web4-explained-a-vision-for-practical-ai-integrated-blockchain/

Anndy Lian is an early blockchain adopter and experienced serial entrepreneur who is known for his work in the government sector. He is a best selling book author- “NFT: From Zero to Hero” and “Blockchain Revolution 2030”.

Currently, he is appointed as the Chief Digital Advisor at Mongolia Productivity Organization, championing national digitization. Prior to his current appointments, he was the Chairman of BigONE Exchange, a global top 30 ranked crypto spot exchange and was also the Advisory Board Member for Hyundai DAC, the blockchain arm of South Korea’s largest car manufacturer Hyundai Motor Group. Lian played a pivotal role as the Blockchain Advisor for Asian Productivity Organisation (APO), an intergovernmental organization committed to improving productivity in the Asia-Pacific region.

An avid supporter of incubating start-ups, Anndy has also been a private investor for the past eight years. With a growth investment mindset, Anndy strategically demonstrates this in the companies he chooses to be involved with. He believes that what he is doing through blockchain technology currently will revolutionise and redefine traditional businesses. He also believes that the blockchain industry has to be “redecentralised”.

j j j

The Great Infra Wars: How Web3 is Forging the Future of Decentralized AI | Taipei Blockchain Week 2025

The Great Infra Wars: How Web3 is Forging the Future of Decentralized AI | Taipei Blockchain Week 2025

Taipei, Taiwan – Sept 2025 – As artificial intelligence reshapes the digital landscape, a critical battle is unfolding beneath the surface: the fight to build the infrastructure capable of hosting truly decentralized AI. At Taipei Blockchain Week 2025, the panel “Infra Wars: The Battle to Host the AI-powered Web” cut through the hype, revealing the profound technical and philosophical challenges at the intersection of Web3 and AI. Moderated by Lee Ting Ting, Founder of FansNetwork, the session brought together infrastructure pioneers to dissect how blockchain can solve AI’s most pressing limitations, from computational bottlenecks to data sovereignty crises.

The Speed vs. Decentralization Dilemma: Rethinking Consensus

The panel opened with a fundamental tension: AI demands blistering speed, while blockchain prioritizes decentralization, often at the cost of performance. “We all know AI models have immense complexity, and users care about speed,” noted moderator Lee Ting Ting, framing the core conflict. “How are emerging consensus mechanisms being redesigned to handle AI’s computational demands?”

Jiahao Sun, CEO of Flock.io and a former financial infrastructure lead, argued that traditional blockchain architectures are fundamentally mismatched for AI workloads. “The public chain design predates the AI boom,” he explained. “Even if on-chain transaction speed is fast, a single consensus layer cannot solve the demands of AI.” Sun’s solution lies in modular consensus: “We’re using a multiple and modular consensus mechanism. We built single processors for decentralized storage and computing, but we align all different modules, data service, cloud service, and computation on top of a PoS system. This creates unlimited transaction possibilities and aligns computing with storage.”

Anthurine Xiang of Quarkchain added nuance, distinguishing between monolithic (e.g., Solana) and modular (e.g., Ethereum) chains: “For modular ecosystems, we need a shared data availability (DA) layer. Solutions like Celestia or EigenDA help store data on-chain forever, making it traceable and preventing losses like the infamous NFT storage failures.” Her point was stark: “When centralized storage fails, like when a team stops paying for AWS, your NFTs become broken links. For AI, this is unacceptable.”

JT Song of 0G Labs (ZG) took this further, announcing their new IFT standard (likely “Immutable File Token”): “For AI agents, all data must be stored on our decentralized service and trace the entire training process. This makes data verifiable and tradeable on-chain, a radical shift from traditional ERC-721.” Crucially, Song revealed ZG’s collaboration with China Mobile: “We ran decentralized training for a 100-billion-parameter model faster than centralized alternatives. Decentralized computing isn’t slower, it’s a different paradigm.”

Data Sovereignty: The Privacy Imperative

The conversation pivoted to AI’s data crisis: Big Tech’s monopolization of user data for training models. “How can infrastructure enable true user ownership while allowing decentralized training?” asked Lee.

Jiahao Sun spotlighted federated learning – a Google-originated technique now supercharged by blockchain. “Your phone predicts your typing locally; raw data never leaves your device. But Google controls the aggregation – it’s still centralized. Blockchain changes this: none of the users’ raw data is ever submitted. Instead, we submit model gradients – changes to the AI itself – which merge into a larger model. Everything is transparent on-chain.” He emphasized the breakthrough: “You don’t have to trust a third party; you see the transactions.”

JT Song reinforced this with ZG’s vision: “We’re building full-chain data services. If an AI project uses our IFT standard, all training data is stored in a decentralized manner. Even if the operation team disappears, the AI agent and its data remain self-sovereign and verifiable.” This tackles the “black box” problem of open-source AI: “Models claim transparency, but the data and process remain hidden. Blockchain forces process transparency.”

Anndy Lian, Intergovernmental Blockchain Advisor, injected pragmatism: “Full decentralization remains a big challenge. Security must be managed effectively, no hacks, no losses. But I’ve discussed zero-data AI architecture with Southeast Asian governments. Blockchain can enforce rules and enable fair audits, creating a win-win for AI and Web3.”

The Killer App: Why Decentralized AI Isn’t Optional

The panel’s most heated debate centered on the “killer app” for decentralized AI: Why bother with Web3 when centralized AI works?

Jiahao Sun targeted enterprise pain points: “Privacy isn’t just ‘nice to have’, it’s necessary in banking, healthcare, and public sectors. But mass adoption needs retail applications. Imagine a virtual companion where conversations are secured on-chain. You know no one, not even the platform, can access your private chats. That’s a healing application blockchain enables.”

Anthurine Xiang pushed for Web3’s evolution beyond finance: “Ethereum aimed to be a ‘world computer,’ but most apps are still token-trading. We need diversified use cases: AI agents, decentralized content platforms. Our ‘supercomputer’ infrastructure must enable non-financial apps with mass appeal, faster speeds, more capacity, lower costs.”

JT Song unveiled ZG’s “Air Wars” AI agent marketplace (boasting 2.3 million testnet users): “Agents can evolve, be verified, and classified. This isn’t just about functionality, it’s about ownership. Users control their AI’s data and evolution.”

But Anndy Lian delivered the most provocative insight: “The best way to onboard people to AI + Web3? Teach them how to make money. AI agents that help users make smart trades or generate income will drive adoption faster than ideology. And let’s be honest: today’s ‘Web3’ isn’t truly decentralized. We need Web4, a more decentralized, less controlled, AI-driven future.”

The Road Ahead: Beyond the Hype

As the session concluded, a clear consensus emerged: The “infra wars” aren’t about which chain wins, but how Web3’s core innovations – decentralization, transparency, and user sovereignty – can solve AI’s existential flaws. Federated learning plus blockchain enables private AI training; modular data layers prevent catastrophic data loss; and new consensus models unlock scalable compute.

The panelists acknowledged the journey is nascent. “Papa, this will be a slow process,” admitted JT Song. Anndy Lian tempered expectations: “From a productivity standpoint, putting everything on-chain remains challenging. But give us time.”

The most profound takeaway? Decentralized AI isn’t a niche experiment, it’s the only path to an AI future where users own their data, models are transparent, and infrastructure serves people, not platforms. As Jiahao Sun succinctly stated: “We’re not just building faster chains. We’re rebuilding the entire operating system for decentralized AI.”

In the battle for AI’s soul, Taipei Blockchain Week 2025 made one thing clear: Web3’s infrastructure warriors aren’t just participants in the AI revolution, they’re building its foundation. The “infra wars” have just begun, but the stakes, a truly user-owned digital future, couldn’t be higher. As Lee Ting Ting closed the session: “This isn’t about technology alone. It’s about who controls the future.” With 2.3 million testnet users already engaging with decentralized AI agents, that future may arrive sooner than we think.

Anndy Lian is an early blockchain adopter and experienced serial entrepreneur who is known for his work in the government sector. He is a best selling book author- “NFT: From Zero to Hero” and “Blockchain Revolution 2030”.

Currently, he is appointed as the Chief Digital Advisor at Mongolia Productivity Organization, championing national digitization. Prior to his current appointments, he was the Chairman of BigONE Exchange, a global top 30 ranked crypto spot exchange and was also the Advisory Board Member for Hyundai DAC, the blockchain arm of South Korea’s largest car manufacturer Hyundai Motor Group. Lian played a pivotal role as the Blockchain Advisor for Asian Productivity Organisation (APO), an intergovernmental organization committed to improving productivity in the Asia-Pacific region.

An avid supporter of incubating start-ups, Anndy has also been a private investor for the past eight years. With a growth investment mindset, Anndy strategically demonstrates this in the companies he chooses to be involved with. He believes that what he is doing through blockchain technology currently will revolutionise and redefine traditional businesses. He also believes that the blockchain industry has to be “redecentralised”.

j j j