ChainFeeds on July 26 published a new edition of its “VC Says” series, collecting recent views from crypto venture firms on AI, crypto’s long-term narrative, enterprise chains, and pending legislation.
The issue is organized into three sections: what crypto VCs are discussing, what research they have released recently, and a list of major primary-market financings disclosed from July 13 to July 19.
Main themes in the latest issue
In the discussion section, ChainFeeds highlighted two topics: “Trump taking stakes in AI companies” and “crypto’s long-term narrative.”
In the research section, it pointed to three recent outputs: Blockchain Capital’s discussion of whether developers should embrace distribution from large platforms or choose neutrality as major companies begin launching chains; an a16z partner’s argument that it is time to pass the CLARITY Act; and Bitwise’s view that the next bull market opportunity may be found in the convergence of onchain finance and traditional finance.
In the financing section, ChainFeeds said 21 funding events were publicly disclosed over the past week, with total financing of more than $752 million. The companies listed were Crypto.com, Alpaca, Flex, ADI Chain, Velocity, Cyclops, Pascal, Pact Labs, Glacis Labs, Alsa, Credible Finance, AXON Finance, Trasia, ILITY, Universe Pro, NOBI, Glide, Coinhako, bloXroute, MasterDEX, and Sovereign Labs.
Delphi Ventures and Dragonfly on Trump’s reported AI-company stakes
For the topic of Trump taking stakes in AI companies, ChainFeeds framed the question this way: if Trump plans to take stakes in AI companies including OpenAI, how would the U.S. respond to open-source AI from China?
Tommy of Delphi Ventures said one of the most interesting possible outcomes of the AI capex cycle could be Trump taking stakes in OpenAI, Anthropic, and possibly a range of other companies, including Oracle.
Tommy said he does not think the government truly trusts Sam or Dario, but argued that “losing to China” is politically unacceptable, especially if slowing AI capital spending and fundraising also hit the stock market. He added that most AI-related sectors are already down more than 20%, and said Trump has previously shown a willingness to take stakes in companies, citing Intel.
He also said that if Trump did take a stake, he would likely demand unrestricted access to models for war use and push to prevent model distillation. In his view, suppressing distillation would limit the development of Chinese models to some extent. If those models cannot reach the same level, users would keep paying for subscriptions from frontier labs, preserving the capex flywheel toward AGI.
Tommy described government share purchases as an extension of large-scale tech buildout driven by national security spending. In his phrasing, the government once funded NASA’s moon mission, and now it is funding Sam to build AGI.
On Chinese open-source models, Tommy said a ban remains unlikely because cheap intelligence is valuable to consumers and enterprises outside frontier labs, and would be very hard to enforce in practice. He added that such a move could instead pull in more talent and make models such as GLM 5.2 and Kimi K3 easier to run locally, producing the opposite effect.
He went on to argue that banning Chinese models would allow companies in other countries to outcompete U.S. firms, because for 90% of everyday AI use cases they would face only one-tenth of the cost. For that reason, he said Trump is more likely to target model distillation than ban Chinese models outright, since that would slow progress globally rather than only inside the U.S.
Tommy also said China will continue to produce innovation surprises, because many people are still in the camp that sees Chinese progress as entirely dependent on distillation rather than innovation. He pointed to the DeepSeek R1 paper as a reminder. Taking those factors together, he said he still believes U.S. open-source AI companies will become a market favorite over the next 12 months.
Haseeb of Dragonfly said the current situation resembles electric vehicles or solar power: the U.S. fears that if Chinese open-source technology spreads fully and is widely adopted, domestic investment could fall sharply and the country could lose its lead.
Haseeb said the best solution is to keep competition fully open and, if needed, support frontier research with government subsidies when capital markets weaken. He added that true nationalization should be delayed as long as possible, though he worries it may be unavoidable in the long run.
He said a second-best option, if restrictions must be imposed, would be to ban only inference services. In practice, that would mean U.S. companies could not directly buy inference services from Chinese labs, but could still use those models through U.S.-based inference providers. He compared that setup to China’s market, where U.S. companies that want to sell goods must work with Chinese firms.
Haseeb added that a blanket ban on Chinese models would be the worst option of all.
Crypto’s long-term narrative was listed as the other focus
Beyond AI, ChainFeeds identified “crypto’s long-term narrative” as the other major discussion topic in this issue. The framing question was whether the crypto sector is being priced with excessive pessimism and whether the industry’s long-term narrative can win market recognition again.
The input provided here ends at that point, so no further detail from that section was available.