From the transaction cost revolution to the rise of AI agents, stablecoins are not just a payment tool, but an underlying technological innovation that redefines the capital market, banking services, and global financial flow architecture. Venture capital Foundation Capital recently published a report titled " Rewriting the Way Money Flows with Stablecoins " to deeply analyze the value, limitations, and application scenarios of stablecoins.
( Arthur Hayes: Forget CRCL, buy JPMorgan! Stablecoins will drive bank stocks up 184% )
Table of Contents
ToggleFoundation: The era of stablecoins has just arrived, and there is huge room for innovation
One of the most eye-catching highlights in the cryptocurrency field in the past 18 months has been the explosive growth of stablecoins.
From payment processing giant Stripe 's $1.1 billion acquisition of Bridge , stablecoin issuer Circle's five-fold increase after its IPO, to Tether 's astonishing profit of $14 billion in a single year, these figures reflect a signal: "Stablecoins have gradually come into the world's attention and are expected to become a new financial infrastructure."
However, the real potential has just been released. Foundation revealed that even though the transaction volume on the chain continues to rise, the integration of stablecoins and traditional finance is still in its early stages, and there is a lot of room for entrepreneurship and innovation.
Transaction costs reduced by 99%? Stablecoins’ financial efficiency revolution
The charm of stablecoins lies in the fact that they combine the stability of the US dollar with the programmable capabilities of blockchains, enabling near-zero-cost, instant fund transfers. Traditional transaction channels require many stakeholders, and everyone has to do some supervision, take a share, and take days. However, on chains such as Solana, the transfer cost is only a fraction of a cent and the speed is less than 1 second.
If the 400 billion transactions processed by Visa and Mastercard each year were put on the chain, the total transaction fees would drop from US$64 billion to US$400 million, which would save merchants 99% of their costs.
Foundation believes this is why Circle's IPO performance far exceeded expectations, even though 99% of its revenue came from interest income and 50% was distributed to Coinbase, and its market value once reached $43 billion. If the same logic is used to estimate, Tether'smarket value can even reach hundreds of billions of dollars.
Debunking myths: Stablecoins are not a panacea, and there are still practical challenges
Despite the strong advantages of stablecoins, Foundation said there are still many misunderstandings and limitations in their application and promotion:
Stablecoins are cheaper than traditional payments, but compliance costs such as KYC and risk control cannot be ignored
Supports 24-hour operation, but if traditional financial institutions are still connected, the actual experience is still limited
The immutable and public nature of blockchain improves transparency, but is not conducive to refunds or other disputes.
Cross-border payments are the main application scenario, but in the long run, the profits of these markets will be squeezed out, leaving only a few winners
The company stressed that the most critical aspect is the disruptive potential brought about by the “composability” of stablecoins:
Most of the applications on the chain can communicate with each other. Once this interoperability successfully connects the off-chain world, it is expected to create a new financial experience. It can not only improve efficiency, but also breed new financial infrastructure.
From AI to corporate cash flow, stablecoins build the future financial framework
Foundatio also mentioned that the future financial world will no longer be a person-to-person transaction, but a "software-to-software, AI-to-AI" financial network. Stablecoins, as programmatic currencies, are a necessary tool for this transformation. From on-chain credit scoring, automatic tax compliance, to smart contract-driven governance and risk control frameworks, stablecoins will become the core of new financial services.
( When AI meets Crypto: a16z analyzes 11 key intersections that are reshaping the Internet economy )
The treasury management and foreign exchange settlement of multinational companies can also achieve higher efficiency and automation through stablecoins. And financial flow scenarios such as one-to-many salary payment, subsidy allocation and e-commerce settlement will also be redefined by stablecoins.
In addition, the financial needs of AI agents will give rise to new transaction logic and application scenarios. From micropayments (small payments) to instant settlements, low-cost, auditable transaction mechanisms that do not require bank accounts are required.
Foundation Capital predicts that, based on pre-programmable and scalable financial flows, stablecoins will not only be a payment tool, but also the core of the reshaping of the financial network. With the emergence of many new financial experiences that seamlessly bridge the on-chain and off-chain in the future, it is expected to bring about a group of giant companies that have never been imagined.
Risk Warning
Cryptocurrency investment carries a high degree of risk. Its price may fluctuate drastically and you may lose all your capital. Please assess the risk carefully.
Well-known semiconductor analyst Dylan Patel was interviewed on July 1 to analyze the current competition among AI giants. He not only talked about the reasons for the failure of GPT-4.5 development, the acquisition of Meta and Scale AI, and the inside story that Apple's lagging behind in AI competition is actually due to its feud with Nvidia, but also shared why AMD still cannot beat Nvidia's powerful CUDA ecosystem. At the end of the interview, Patel even said that he is most optimistic about OpenAI's future development in superintelligence.
Table of Contents
ToggleSuper intelligence is the way to go, if you don’t keep up, you will lose
Patel was the first to say that the entire AI industry has shifted from pursuing general artificial intelligence (AGI) to pursuing super intelligence. Since Illya Sutskever, co-founder of OpenAI, founded Safe Super Intelligence (SSI), it has been seen as a key turning point. It was also from that time that Meta, OpenAI, and xAI all began to shift. He pointed out that if we don’t catch up now, we will be the losers in the end.
Why did GPT-4.5 fail?
When asked why OpenAI GPT-4.5 failed, Patel said that the problem was that although the model parameters became larger and smarter, they were actually too slow, too expensive, and had low user adoption. The main reason for the failure was that "not enough data" caused GPT-4.5 to be "overparameterized," which means:
“The bigger the model is, the more hungry it is for data. If it is only given a limited amount of information to memorize, it will not be able to truly understand.”
Patel said that looking back, we can find that the key to the failure of GPT-4.5 is not the scale of the model, but the failure to keep up with the data and reasoning architecture.
( GPT becomes a bit weird? Three major events reveal the potential risk of AI getting out of control )
Why does Meta's AI keep losing?
Talking about Meta's Llama 4 and the delayed Behemoth model, Patel believes that the problem lies in the overall technical decision-making. Although Meta also has strong researchers and a lot of GPUs, if there is no leader who really understands technology and can make decisions, it is easy to go in the wrong direction.
He gave an example, in the process of developing Meta's AI model, some teams tried to imitate China's DeepSeek, but due to improper training design, some "expert modules" in the model were not used effectively, resulting in a waste of training resources. The problem lies in the lack of a leader who understands technology and can make decisions to lead the direction.
Why OpenAI and Microsoft are drifting apart
Patel believes that the cooperation between OpenAI and Microsoft has passed the honeymoon period. Although Microsoft invested more than $10 billion in OpenAI, it only received a commission and profit sharing from revenue, and did not get any shares in OpenAI, which is a bit embarrassing. Microsoft originally had the exclusive computing rights for OpenAI, but it was cancelled this year.
OpenAI is now cooperating with suppliers such as Oracle and CoreWeave to build data centers to get rid of its dependence on Microsoft. Patel also revealed that OpenAI is the most expensive startup on the planet and will not make a profit in the next few years. It only wants to continue to expand its scale and valuation.
There is a reason why Apple lags behind. It is conservative and refuses to cooperate with Nvidia.
Patel pointed out that Apple's culture is too conservative, resulting in no major breakthroughs in the field of AI. The company only acquires some small and insignificant startups, and lacks the "guts" to acquire an entire top team at one time.
In addition, Apple previously had the "Bumpgate" incident with NVIDIA, because a certain generation of NVIDIA GPU caused device failure due to poor welding, and the two sides blamed each other and created a "beef". Later, due to patent litigation issues, Apple stopped using NVIDIA chips.
Although Apple's emphasis on "device-side AI" has advantages in terms of security and response speed, Patel believes that the overall market demand will still tend to favor cloud services. He said:
“Everyone says they value privacy, but in reality they still prefer free and easy-to-use services.”
Is Grok interesting? At least some of the questions are open to discussion.
Patel's evaluation of Grok is better than expected, especially when it comes to checking breaking news, discussing topics such as geography and history, his answers are more natural and he is not afraid to talk about sensitive topics. He mentioned that when he was asked about the changes in the black-white population ratio in the southern United States, the history of slavery, or the history of oil monopoly, Grok was willing to give a more in-depth background instead of just telling the safe version.
As for Musk's claim that Grok is "the smartest AI in the world", Patel is reserved. He said Grok might be really good, and Musk is indeed very executive in building data centers and buying power plants, but he thinks it will take some time to see.
( Musk xAI wants to retrain Grok, experts criticize: deliberately distorting history )
AMD is catching up, but Nvidia still holds the lead
Patel finally analyzed the current chip war between Nvidia and AMD. He agreed with AMD's attitude of striving to catch up, and it did perform well in some situations, but overall Nvidia was still ahead.
Patel gave an example, Nvidia's NVLink can interconnect 72 GPUs at high speed, while AMD can only do 8 at this stage. In addition, Nvidia's software and development ecosystem are quite complete, and users can execute models with just a few clicks, while AMD still needs to manually adjust a lot of parameters.
However, Nvidia recently launched its own cloud rental platform Lepton and DGX Cloud Lepton AI , which made many cloud providers who started out by selling Nvidia chips feel like they were backstabbed, and also gave AMD the opportunity to grab some market share.
Who is most likely to take the top spot in superintelligence?
Patel finally believes that OpenAI is most likely to take the throne of superintelligence because all major breakthroughs come from them. Next is Anthropic, which has been steady and recently became more open, and then Meta and xAI, which are still working hard.
Risk Warning
Cryptocurrency investment carries a high degree of risk. Its price may fluctuate drastically and you may lose all your capital. Please assess the risk carefully.
