ChainFeeds on July 15 published a new research brief spanning Robinhood’s tokenization strategy, the market significance of Robinhood Chain, OpenAI’s shifting relationships with major technology companies, the latest direction of Lean Ethereum, and a16z’s view on how traditional finance is approaching blockchain.
Vlad Tenev says humans will keep trading as Robinhood pushes stock tokens
In a conversation with Robinhood co-founder and CEO Vlad Tenev, one of the clearest takeaways was his view that no matter how far AI develops, humans will still trade for themselves.
Tenev said the market keeps returning to the same question: whether the industry is in an AI bubble. What makes this cycle harder to judge, in his view, is that many AI companies already have real business models. Foundation model firms sell tokens to enterprise customers and individual users, while OpenAI has also built a subscription business at meaningful scale. That changes the debate. The key question is no longer whether AI companies can generate revenue, but whether companies now spending heavily on AI will move from a learning phase with limited cost sensitivity to one centered on return on investment. If that shift happens, revenue per customer may either keep rising or begin to fall.
He also argued that large parts of the market still have not adopted AI in a serious way. The same applies to consumers. He pointed to Claude Code and said its user base appears to be in the tens of millions, still far from the hundreds of millions or even billions. For that reason, he said the AI market remains at a very early stage and still has a long runway.
On IPO cycles, Tenev said people tend to believe they are living through uniquely important moments, but the longer view looks more cyclical. He noted that the IPO window closed at the end of 2021 and had begun reopening by 2023, which he described as closer to a sinusoidal cycle than a permanent shift.
Tenev also linked the launch of Robinhood Chain to the company’s broader tokenization strategy. He said Robinhood laid out its long-term tokenization roadmap at an event in Cannes, France, last year. The point of tokenization, in his framing, is similar to what stablecoins did for access to dollars. Stablecoins made U.S. dollar exposure easier to obtain across hundreds of countries and regions. Tokenization, he said, could do something similar for U.S. equities by bringing the value of American stocks to users in markets where financial systems are less developed than those in the U.K. or the U.S.
Robinhood therefore plans to launch stock tokens on Robinhood Chain. According to Tenev, those tokens will be available in more than 120 countries and regions. Users will be able to access them through non-custodial wallets or through Robinhood Wallet, the company’s own wallet product. The goal is to offer a full trading and swap experience around tokenized equities and give users exposure to the broader U.S. listed stock market.
In the first stage, Robinhood expects to support about 2,000 U.S.-listed stocks. Those stock tokens are intended to trade 24/7 and be portable. Tenev said that portability matters because users would no longer be fully dependent on a single broker as counterparty. As long as the blockchain remains live, the tokens can be transferred and exchanged freely.
\nWhy Robinhood Chain is being framed as a boost for Ethereum
In a separate long-form commentary, Haotian argued that Robinhood Chain introduces a new narrative because it comes from a compliant and regulated TradFi platform. In that framing, future DeFi lending and perpetuals markets could be connected more directly to liquidity tied to real-world assets.
Haotian’s main point was structural. Robinhood, despite being seen as a representative of traditional finance or Wall Street, did not choose to build an independent Layer 1. Instead, it built Robinhood Chain on Arbitrum Orbit, a Layer 2 stack in the Ethereum ecosystem. He said that choice suggests Ethereum mainnet’s role as the Layer 1 that gives Layer 2 networks security and finality is still intact. In his view, Robinhood’s decision effectively validates Ethereum’s rollup-centric route and offers a reference case for other institutions.
If more institutions follow that pattern, he wrote, they may prefer to build application-specific chains on Ethereum rather than start from scratch. That would matter for Ethereum because it reinforces its position as a security and settlement anchor.
The commentary also focused on Robinhood’s user base. Haotian said the platform already reaches a large Gen Z audience, a group he described as more open to crypto assets and high-volatility assets such as meme coins. He added that Robinhood’s CEO has shown openness toward memes and that CASHCAT has already become an attention point on Robinhood Chain.
For that reason, he argued that any TradFi-DeFi crossover cannot rely only on putting traditional products on-chain. It also needs crypto-native forms of participation if the chain is going to unlock the full potential of on-chain economic activity. If Robinhood Chain produces more breakout meme assets, he wrote, the impact may extend beyond the existing crypto user base and pull in new retail users from traditional markets.
Haotian also cited fee data as an early sign of that effect. He said Robinhood Chain has brought Uniswap protocol roughly $3 million to $4 million in trading fees in a single day, while newer meme launchpad NOXA Fun has been generating around $1 million in daily fees. Those figures remain far below DeFi’s peak era, he said, but they still show how a highly active Layer 2 can feed activity back into the Ethereum ecosystem.
He added one qualification: protocol revenue does not necessarily translate directly to gains for token holders. Even so, protocols that make money are better placed to attract and retain strong developers.
\nOpenAI’s relationships with Apple, Microsoft and Anthropic keep turning competitive
Biteye’s essay on OpenAI and large technology companies described a pattern that keeps repeating: OpenAI partners with major players, expands toward the edge of their businesses, and then ends up competing with them.
Apple and OpenAI
The piece described Apple’s relationship with OpenAI as a classic AI-era case of a rapid tie-up followed by a rapid split. In 2024, Apple was clearly behind in generative AI. ChatGPT had already captured user attention globally, Google was pushing Gemini, and Microsoft had woven Copilot into office software, while Apple was still trying to prove Siri could understand users well enough.
To close its model gap, Apple partnered with OpenAI and integrated ChatGPT into Apple Intelligence. When Siri could not answer a question, it could hand the request to ChatGPT. For OpenAI, that meant access to one of the world’s most important consumer electronics gateways. For Apple, it bought time.
But the two companies were not aiming at the same destination. Apple wanted ChatGPT to remain a system-level plugin hidden behind the iPhone, with Apple retaining control over users, hardware, operating systems and distribution. OpenAI’s ambition was broader. It wants to become the next user interaction layer. If people stop opening apps and instead ask an AI assistant to act for them, the company controlling that assistant could route around the operating system and become a new traffic hub.
Biteye wrote that Apple began to feel the threat more directly when OpenAI brought former Apple chief design officer Jony Ive’s team into consumer hardware efforts. The article said Apple is now planning to bring in Gemini as a core capability for a new Siri while also suing OpenAI, accusing it of obtaining internal confidential information through the hiring of Apple employees.
Microsoft and OpenAI
The article framed Microsoft and OpenAI as a relationship between a financial backer and a potential rival. Microsoft supplied early funding, compute, cloud services and enterprise distribution, making it a central partner in OpenAI’s rise.
After ChatGPT took off, Microsoft quickly integrated OpenAI models into Copilot, Office, Bing and enterprise offerings, helping return the company to the center of the technology race. But as OpenAI’s own influence expanded, the balance started to shift. Microsoft wants OpenAI to keep helping drive Azure, Office and enterprise AI sales. OpenAI wants its own consumer entry point, enterprise platform, developer ecosystem and hardware presence.
The article said both companies still talk about strategic cooperation in public, but both are reducing dependence in practice. Microsoft is building its own AI model capabilities, adding other model suppliers such as Anthropic and cutting single-source reliance on OpenAI. OpenAI, for its part, has been expanding its cloud relationships with Oracle, CoreWeave and even Google Cloud, weakening Azure’s exclusive position.
The core question in that contest, the piece argued, is who controls entry points, infrastructure and commercial value in the AI era. Microsoft has strengths in enterprise distribution and cloud computing. OpenAI is trying to become a direct platform for end users.
OpenAI and Anthropic
Biteye described OpenAI’s competition with Anthropic as an internal family dispute that evolved from philosophical disagreement into direct business rivalry. Anthropic’s founding team includes several former OpenAI figures, among them co-founder and CEO Dario Amodei.
The break, according to the article, centered on AI safety, the pace of commercialization and corporate governance. Anthropic placed greater emphasis on safety, interpretability and long-term risk control, favoring a more cautious approach. OpenAI put more weight on product deployment and market competition, seeking advantage through speed.
As AI moved into commercialization, those philosophical differences hardened into direct market conflict. The article listed several areas where the companies now meet head-on: ChatGPT versus Claude, OpenAI API versus Anthropic API, Codex versus Claude Code, along with competition for enterprise customers, research talent and developers.
Once they entered the same commercial lane, the older debate over what direction AI should take gave way to a struggle over market share, revenue and leadership. The contest now is not only about technical paths, but about who becomes the core infrastructure provider of the AI era.
\nLean Ethereum returns with a sharper institutional and protocol focus
In its review of Lean Ethereum, imToken Labs argued that a set of organizational and protocol-level changes across Ethereum point in the same direction: reducing redundancy and friction in how the network operates.
The article said outside observers have long tended to equate the Ethereum Foundation, or EF, with Ethereum itself. Questions about upgrades, research priorities, ecosystem funding and public communication often collapsed into a single line of thinking: what is EF going to do? But the foundation is not a normal company. It has no shareholders in the traditional sense, does not optimize for market share or quarterly profit, and does not literally own the Ethereum network. That creates a built-in tension. Ethereum needs long-term work on protocol research, upgrades and public goods, yet if too much research, capital, talent and decision-making sit inside the foundation, EF itself can become a centralization risk.
According to the article, recent organizational changes are meant to break that pattern. In the latest round of adjustments, the Ethereum Foundation reduced headcount by about 20% while narrowing internal focus around protocol, users and institutions. EF said the goal was to become leaner and more focused, prioritizing tasks that only the foundation can and must take on.
At the same time, capabilities once concentrated inside EF have begun moving outward. On June 22, five former core Ethereum Foundation researchers announced the creation of Ethlabs, an independently operated nonprofit R&D lab that will work on protocol research, infrastructure and institution-grade technical needs. On July 1, Ethereum Institutional formally launched and took over the institutional outreach work previously handled by the EF market expansion team, becoming an independent entry point for traditional financial institutions looking at Ethereum.
The article stressed that Lean Ethereum is not a brand-new concept. In July 2025, Ethereum Foundation researcher Justin Drake published a ten-year “lean Ethereum” vision covering lean consensus, lean execution and lean data. The aim was to improve the scalability of the base layer and Layer 2 networks while preserving decentralization and system stability.
What changed this week, imToken Labs wrote, is that Vitalik used the latest strawmap to move those scattered research threads into a clearer position. Lean Ethereum is not one hard fork. It is a sequence of protocol changes expected to unfold over the next three to four years and is described by Vitalik as Ethereum’s third major iteration.
By Vitalik’s summary, the roadmap touches several core parts of the protocol. One is simplification: moving from heavy execution toward light verification. Through technologies such as recursive STARKs, proof verification would replace part of the need to re-execute transactions, while client architecture, the state model and gas design would also be adjusted to make the protocol more compact and easier to verify formally.
Another focus is quantum resistance. What had been treated as a longer-term issue is now being pulled forward. Cryptographic components that could be threatened by quantum computing are expected to be replaced gradually with quantum-resistant schemes, and quantum-safe blob design has also been highlighted.
Privacy is the third major pillar. The article said privacy will stop being only an application-layer function and will increasingly be treated as a core protocol capability. New frame, mempool and state-tree designs are intended to support a more native environment for private transactions.
It also connected these changes to validator economics. The 0x02 compounding validators mechanism is presented as another form of optimization because it lets staking rewards participate in continuous compounding more effectively. Previously, smaller stakers were constrained by the 32 ETH effective balance cap, meaning rewards above that threshold could not keep compounding inside the same validator, while large staking providers could aggregate rewards faster and launch new validators. After Pectra introduced the 0x02 mode, the maximum effective balance for a single validator rose to 2048 ETH, and rewards can continue to be staked.
Put together, the article said, the Ethereum Foundation’s downsizing, the emergence of Ethlabs and Ethereum Institutional, the protocol shift toward simplification, privacy, quantum resistance and light verification, and the 0x02 validator design are not isolated developments. They all point to the same broader effort to make Ethereum easier to operate, easier to verify and better suited for the long term.
a16z says TradFi wants blockchain infrastructure, not DeFi ideology
In a separate piece, a16z argued that traditional finance is not adopting blockchain because it has embraced decentralization as an idea. It is adopting blockchain because the technology can improve how existing financial businesses run. In that framing, the future is less about a neat DeFi-TradFi merger and more about a new category built on blockchain rails: programmable financial infrastructure designed around institutional constraints.
The article began by revisiting a familiar market assumption. For a long time, many believed financial systems would converge around a blend of DeFi and TradFi, with decentralized finance supplying open liquidity and traditional finance providing institutional distribution, eventually replacing older financial models. a16z said reality looks different.
When traditional financial firms find that blockchain can lower cost, improve settlement efficiency, expand distribution and strengthen control over customer relationships, they will use it. That does not mean they are merging with DeFi. What they are actually doing is adopting selected DeFi modules that fit their operating requirements while discarding the pieces that do not work within their risk frameworks.
That is why a16z suggested the market may end up with a different category altogether: programmable financial infrastructure built on blockchain but optimized for institutional demands.
The article said institutional adoption of blockchain usually depends on two conditions. First, the technology has to improve cost, risk or distribution efficiency. Second, it has to fit institutional requirements around control, compliance and accountability. Open access, anonymity and irreversible execution may be strengths in native DeFi systems, but those traits often fail to meet the regulatory and operating needs of large institutions. That is also why institutions are more interested in core capabilities such as atomic settlement, programmable money and tokenized collateral than in reproducing today’s full DeFi stack.
a16z also argued that institutional blockchain adoption is not simply DeFi at larger scale. Enterprises evaluate technology differently from crypto-native users. When companies choose software vendors and infrastructure partners, they focus on operational risk, compliance controls, long-term maintenance and system ownership. A protocol that succeeds in DeFi does not automatically fit institutional markets. Businesses rarely buy the “most advanced” technology. They buy technology that fits existing workflows, risk models and procurement systems.
The article compared this process with earlier technology shifts. The internet was reshaped through enterprise firewalls and internal networks. Cloud computing evolved into private clouds and VPCs. AI brought requirements around internal deployment, data isolation and model governance. Blockchain, it said, will go through a similar process.
a16z divided that reconfiguration into two main areas. The first is compliance. KYC, AML, sanctions screening, investor accreditation and regulatory reporting are unavoidable for most financial institutions. Traditional DeFi systems were not originally designed around those requirements, so institutions usually need identity controls, asset freezing capabilities and transaction tracing layered in. The second is enterprise value creation. Institutions do not adopt blockchain because they believe in permissionless finance. They adopt it because it can lower operating costs, reduce reconciliation friction, open new distribution channels or deepen customer relationships. If the commercial value is not clear, the technology is unlikely to survive enterprise procurement.
At the industry level, the article said institutional adoption and open-network development are not mutually exclusive. They can advance at the same time. For individual startups, though, they usually imply very different business models. Products built for institutions require an understanding of procurement processes, compliance structures, risk controls, channel partnerships and long sales cycles. Products built for open networks depend more on developer ecosystems, liquidity, composability and network effects. The customer base, distribution path, product requirements and success metrics differ sharply.
a16z did not frame one path as superior to the other. Instead, it said builders need to be clear about which market they serve and where the real connection lies: public blockchains as neutral settlement infrastructure. In the best-case scenario, institutional systems bring capital, trading scale and broader market recognition, while open networks continue producing new financial primitives and innovation that institutions later absorb.
The likely long-term direction, the article concluded, is that public blockchains increasingly become an important base layer for financial settlement, while the application layer built on top evolves in more varied ways depending on user needs. Some of those applications will remain more open. Others will be designed more explicitly around institutional regulatory requirements. In that sense, the next generation of programmable financial infrastructure is not a copy of DeFi. It is a redesign of financial systems using blockchain technology under institutional constraints. Some projects will build institution-ready networks from scratch. Others will package existing DeFi protocols into products institutions can use.