Cryptopolitan

Chinese electric vehicle maker Xpeng flew into Munich this week with more than just a new car. The company used the European launch of its L03 SUV to tell the world it is building robots and flying cars too.
The launch event was branded a “Physical AI” event, a term Xpeng uses to describe its approach of combining self-built AI chips and large language models with physical products like cars, robots, and power systems.
The company says this technology currently drives three product lines: electric vehicles, humanoid robots, and flying cars.
Xpeng is no small player in EVs. It has shipped more than a million vehicles in China and sold over 60,000 cars in Europe since entering the market in 2024. It now sells in 65 countries worldwide.
At the event, the CEO He Xiaopeng, said that humanoid robots and flying cars are not sci-fi anymore. In fact, both the robot, named Iron and the flying car named Aridge are soon going to be mass-produced.
Iron has also made appearances at the company’s previous events for about a year. As previously reported by Cryptopolitan China is prioritizing robots more than AI models unlike the U.S.
Even though Aridge’s has been in development for 13 years now, He said it will be released in the near future for European customers.
Meanwhile, L03 is a compact SUV producing 180 kW of power with a WLTP range of up to 445 km. At €35,600, it is priced below rival models from Tesla and Hyundai.
All three of Xpeng’s product lines, the car, the robot, and the flying car, run on Turing chips, a processor the company developed in-house, featuring 750 TOPS per chip and a 40-core processor.
Chasing Tesla on self-driving
On the autonomous driving side, Xpeng laid out a roadmap to reach Level 4 autonomy by 2028.
That would mean the car can drive itself without human input under defined conditions, a step beyond the Level 2 systems that Tesla’s Full Self-Driving and comparable Chinese systems currently represent, where drivers must remain alert and ready to take control at any moment.
Xpeng’s CEO has said the company’s VLA system already outperforms Tesla’s FSD on narrow roads and in tight situations.
The autonomous driving technology has already moved beyond Xpeng’s own vehiclesVolkswagen began mass-producing its first jointly developed model in March, using Xpeng’s driving systems and Turing chips. The company is also building robotaxis on the same technology framework.
Google maps deal makes Xpeng a first in Asia
Xpeng also announced a partnership with Google Maps for vehicles sold outside China, making it the first car company from the Asia-Pacific region to ship vehicles with Google Maps built directly into the navigation system. Rather than requiring drivers to download the app or mirror their phone screen, the Google Maps technology will be embedded natively into Xpeng’s own map interface through the Google Maps Auto SDK.
The integration will support real-time traffic guidance, EV trip planning, and energy estimation, and will also feed into Xpeng’s NGP assisted driving system. Overseas buyers of the L03 will be the first in the Xpeng lineup to get the upgraded experience.
The company has a dedicated research and development center in Munich, and the L03 will go on sale across 64 countries and regions.
Sven De Smet, Xpeng’s head of product in Europe, summed up the company’s pitch simply: “We are a technology company that makes cars. Not a car company that uses tech.”
The smartest crypto minds already read our newsletter. Want in? Join them.

The on-chain analytics firm BubbleMaps reports that most people trading the biggest memecoins on Robinhood Chain are underwater.
Reportedly, only 37% of the 164,538 traders active across the network’s top 50 tokens have made a profit, while the remaining 63% have lost money.
Are traders making money on Robinhood Chain?
“Robinhood trenches are brutal,” BubbleMaps, an on-chain analytics firm, wrote on X, calling the memecoin market “a tough game.”
Robinhood Chain, an Arbitrum Orbit layer-2 blockchain built on top of Ethereum to cut fees, went live on July 1, 2026.
However, of the 164,538 traders active across the network’s top 50 tokens, only 37% have made a profit. The remaining 63% have lost money.
Cryptopolitan reported CASHCAT (CASHCAT), the token modeled on Robinhood’s old cat mascot, jumped 718% in 24 hours to a $68 million market cap on July 8, when Robinhood CEO Vlad Tenev publicly warmed to memecoins on the network. However, since then, the token has bled out.
The token is trading at roughly $0.0557, down about 75% from its July 11 all-time high of $0.2252, with a market cap near $55.7 million and about 4,840 holders. Around 25,000 wallets reportedly held CASHCAT on July 13, when the token carried a market cap near $150 million.
Are Robinhood Chain meme price drops coordinated dumps?
Beyond tallying winners and losers, BubbleMaps ran distribution checks on individual tokens and found that CASHCAT shows no major holder clusters and a solid spread of ownership.
A handful of clusters that were present at token launch have already sold out of their positions. The checks show that the losses on CASHCAT come from ordinary price action, not from a concentrated group sitting on the supply.
CASHDOG, on the other hand, was flagged and labeled as heavily bundled. BubbleMaps said its holders were funded through one-time contracts, implying a coordinated setup rather than organic buying.
Robinhood is currently covering gas fees for wallet transactions during the first 90 days in order to pull in more users. That incentive, paired with the CASHCAT frenzy, has brought $3.1 billion in volume over a seven-day stretch to the decentralized exchanges on the chain.
Meme tokens such as Wen Lambo, Tendies, and Hoodrat follow closely behind. DeFiLlama pegged the total value locked on the chain at about $227 million on Saturday.
The tokenized stocks Robinhood actually built the chain to sell have drawn far less interest so far, accounting for just $12.66 million in market cap on the network. At its peak, CASHCAT alone was worth twelve times that figure.
The industry is observing whether activity on the chain remains constant once the gas subsidy expires in late September and traders start covering their own fees.
The smartest crypto minds already read our newsletter. Want in? Join them.

Kimi-K3 has claimed top spot on a closely watched coding leaderboard. Kimi-K3 is a Chinese-built AI model, and this has prompted David Sacks, President Trump’s former AI czar, to believe that American AI labs are being hindered by US rules.
Moonshot AI built Kimi-K3, and the AI model became number one on July 16. It took top spot on the Frontend Code Arena with 1,679 points. Frontend Code Arena monitors how well AI models write front-end code, and Kimi-K3 climbed 17 places from its predecessor to finish at the top spot.
In an X post, David Sacks expressed concern over Kimi-K3’s rise, stating the result was “concerning.” He linked the result to politicians and officials, more or less inflicting damage on American AI models by restricting the construction of new data centers.
What does Kimi K3 bring?
Kimi K3 runs on 2.8 trillion parameters with a one-million-token context window, and Moonshot intends to make the model’s open weights public by July 27. Kimi-K3 is available on Kimi.com and Moonshot’s API.
Sack’s argument also includes the price of AI models. His friend and co-host of the All-in podcast, Chamath Palihapitiya, wrote on X that a buyer can now “spend $0.50 per 1MM leading edge tokens or spend $56 for those same 1MM tokens,” adding that such discrepancies mean “the math ain’t mathing.”
Sacks went further to explain that American models prioritizing wokeness could also become a problem. He stated this after a user had switched to Kimi from Claude because Kimi “just does the thing instead of lecturing you.” In another post, he argued against gatekeeping AI models.
A policy fight from outside the White House
Sacks served as the AI and crypto czar. He was a special government employee until March of this year. He’s to co-chair the President’s Council of Advisors on Science and Technology. The council makes recommendations to the president but does not set policy. So his arguments are definitely worth listening to.
This is not his first time criticizing the administration’s AI policy. He disagreed with the Trump administration after the US approved limited access to Anthropic’s Claude Mythos 5 to ~100 businesses and federal agencies.
According to Sacks, America would lose the AI race if it resorts to broad regulation, comparing it to the internet era, where America was the bastion of innovation.
Sacks’ approach to AI regulation, while in office, was not exactly welcomed by all. He had a huge hand in the executive order that blocked states from regulating AI.
Moonshot’s open-weights release is on the 27th of July, and it will allow developers to run Kimi-K3 away from the company’s servers.
The smartest crypto minds already read our newsletter. Want in? Join them.

Chamath Palihapitiya is asking if the money being poured into artificial intelligence is paying off for anyone beyond the small group of companies already collecting it.
The Social Capital founder, in two posts made on X on July 17 and 18, called out AI labs over how they trained their own models versus how they treat others that copy their work. Chamath then pointed out what he noticed about buzzwords that dominate SEC filings and how they tend to deflate over time, a reference to the current buzz of AI and agentic technologies.
He wrote, “Right now everyone is glomming onto AI like it’s a life raft. But these same folks have not yet shown repeatable, audited, verifiable ROI even as their CapEx and OpEx are increasing with token costs on all things AI.”
Uber, Microsoft, and Meta are already reining in AI budgets in response to findings from a McKinsey survey that reported that most companies see no earnings impact from generative AI. Other critics just don’t see a future where the whole market is not dragged down by one big stumble.
Did the AI Labs play by different rules?
Palihapitiya threw a jab at Anthropic and, by extension, other players in the frontier model space, like OpenAI, on Friday when he posted an assessment from Anthropic’s Fable model about distillation, the practice of using one model’s outputs to train a cheaper rival.
Palihapitiya stated distillation as a moral problem is genuinely contested. He pointed out that the labs themselves built their systems on the open internet, including copyrighted books, articles, and code, among others.
Now the same firms that relied on these resources scraped across the world to train their frontier models are now against others doing the same thing to them.
Are big tech budgets finally being reined in?
The froth is what large tech firms are now trying to drain. Uber’s AI budget for 2026 was exhausted in roughly four months. The company had to cap coding tools at $1,500 per employee per tool, and it is reportedly tracked on an internal dashboard.
Microsoft is phasing out Claude Code licenses in its Experiences and Devices division. It is now encouraging its engineers to use GitHub Copilot CLI, a move an internal memo referred to as deliberate benchmarking.
In an April memo, Meta’s CTO Andrew Bosworth said, “All motion is not progress and token usage alone is not a measure of impact of any kind.”
MIT’s NANDA initiative looked at enterprise AI pilots across industries and found that 95% of them produced no measurable financial return at all.
Is the AI industry too concentrated to fail safely?
Technology critic Ed Zitron warned this week that OpenAI has become “one of the largest liabilities in recent economic history,” stating that its failure would be the AI era’s Lehman Brothers moment.
By his accounting, OpenAI plans to spend more than $50 billion on compute this year, has taken on around $748 billion in obligations to Microsoft, Amazon, and Oracle, and posted a $38.5 billion net loss in 2025 on $13.07 billion in revenue.
Oracle, which has committed over $340 billion to build capacity for OpenAI, saw its credit rating cut to the lowest investment-grade rung by S&P Global, with OpenAI named as a key risk.
That is the setup Palihapitiya is prodding. A few firms sit at the center of trillions in committed spending; however, the returns remain unproven for most buyers.
Palihapitiya acknowledged that AI is real, adding that “it is the defining change of our lifetime.” However, what he wants to change is that returns should start coming in for more companies than a few, writing, “We are in the early phase where a few companies are making all the money from our largesse. This needs to be reset for everyone to win.”
The smartest crypto minds already read our newsletter. Want in? Join them.

OpenRouter, the aggregation platform that routes developers’ API calls across dozens of AI systems, reports that open-weight models built by Chinese labs now run more than three times the weekly token volume of American models.
The reason is because Chinese models that are nearly equivalent to their U.S. counterparts cost a fraction as much. Price has become a deciding factor for developers optimizing with AI.
Chinese models hold the top spots in LLM adoption. Source: OpenRouter.
Why are Chinese AI models becoming more popular?
OpenAI and Anthropic sell access to closed models at a premium per token, while labs such as DeepSeek, Moonshot, and Zhipu publish weights developers can download and run without paying a per-token toll.
As of late June, the six most-used models on OpenRouter were all open releases from Chinese firms, including Tencent, Xiaomi, DeepSeek, MiniMax, and Z.ai.
For context, in a Cryptopolitan comparison, Zhipu GLM 5.2 runs at $1.40 per million input tokens and $4.40 per million output tokens compared to $5 and $25 for Anthropic’s Opus 4.8.
During the week of February 9 to 15, 2026, Chinese models handled 4.12 trillion tokens on OpenRouter while U.S. models handled 2.94 trillion. This was the first week that the American side of the industry lost. Through most of 2025, U.S. systems had held close to 70% of top-model usage.
Chinese open-weight models are also dominating downloads. Data shows that they made up 41% of Hugging Face downloads this spring, bypassing U.S. releases.
Vercel’s Production Index, published in July, found that open-weight models accounted for 29% of all tokens on its AI gateway in June 2026, up from 11% in April. Those models handled close to a third of tokens while drawing under 4% of spending. DeepSeek alone drove 22.6% of the gateway’s token volume, placing it third behind Google.
Why are industries adopting Chinese technology?
Cryptopolitan reported that a KPMG survey of 2,145 senior leaders found 29% of them cannot understand or control what their AI systems cost to run. Uber burned through its 2026 AI coding budget by April and now caps engineers at $1,500 per tool each month, while one unnamed firm ran up a reported $500 million Claude bill in a single month.
Z.ai’s cloud API, for instance, falls under China’s National Intelligence Law, meaning any code or data sent through it could be accessed by Chinese authorities. U.S. lawmakers opened an inquiry in May into risks from Chinese-origin models in critical infrastructure.
Notably, Moonshot AI is preparing the launch of Kimi K3, a 2.8-trillion-parameter open-source release that the company says competes with top U.S. systems.
On overall benchmarks, the model sits below Anthropic’s Claude Fable 5 and OpenAI’s GPT-5.6 Sol, but Cryptopolitan reported that it beat Claude Opus 4.8 and GPT-5.5 on coding and agent tests. Kimi K3 is priced at $3 and $15 per million tokens, undercutting GPT-5.6 Sol at $5 and $30 and Fable 5 at roughly $10 and $50.
If you're reading this, you’re already ahead. Stay there with our newsletter.

For a while, stock trading in Asia appeared to be unstoppable and has grown to be one of the largest sources of income for American banks.
Then, practically overnight, the atmosphere was altered by a Chinese artificial intelligence business.
Asian equity markets have been bringing in large sums of money for Wall Street’s main banks, and the area is getting closer to becoming Europe’s second-largest source of trading income behind the US.
Large institutional clients have invested heavily in Asian companies involved in the supply chain for AI chips over the past year, including China’s Cambricon Technologies, Taiwan’s TSMC, and South Korea’s SK Hynix.
The figures are startling. Asia continued to be identified as a major contributor to the largest investment banks’ record-breaking $25.7 billion in revenue from equity trading in the most recent quarter.
Clients wanted a piece of everything related to AI in Asia, not just in the US, according to Denis Manelski, co-president of global markets at Bank of America.
“We saw strong demand for financing, cash and derivatives in Asia,” he stated.
Coal fills the gap as power demand surges
However, a lot of electricity is needed to keep that machine operating. Data centers don’t sleep. They require power constantly, and Asia has a plentiful supply of coal that can provide it.
Nearly three-fifths of the world’s known coal deposits are located in this region. It is less vulnerable to disruptions from foreign conflicts, less expensive than importing gas or oil, and has emerged as the preferred fuel for the region’s data center expansion despite its well-known environmental harm.
The issue is straightforward, according to Alexander Kheder, a market researcher at BMI who monitors global investment in AI infrastructure.
“AI demand is materialising faster than clean energy generation can be commissioned,” he stated.
Despite significant advancements, solar and wind power are still unable to consistently provide the steady supply of electricity needed by AI data centers.
Cyberjaya, a technology hotspot in Malaysia, currently has dozens of data centers, and more are on the way.
Adit Rahim, a 49-year-old communications professional from the area, is uneasy about the expansion. He stated, “The impact is going to be very, very visible,” highlighting the strain that this development is placing on nearby utilities.
A Chinese startup spooks the markets
Then the shock arrived.
A new model named Kimi K3 was unveiled by the Chinese startup Moonshot AI, which described it as an open-source system that nearly matches the performance of top Western AI products like ChatGPT from OpenAI and Claude from Anthropic.
The subscription business models developed by US AI businesses are immediately threatened by open-source models, which are free for everyone to use and modify.
Businesses may spend less on paid services and the chips that power them if they can obtain similar outcomes for free.
The markets reacted quickly. On Friday, the Nasdaq dropped 1.4%. The S&P 500 fell 1%. The Dow ended the day down 407 points, or 0.77%.
Japan’s market ended the day down 4%, while Taiwan’s primary stock index fell more than 6%.
After falling 1.6% on Friday alone, an index that tracks semiconductor chip companies is now 20% below a record high it reached in late June, firmly placing it in bear market territory.
Senior bank executives expressed their worries openly. The global co-head of equities at Goldman Sachs, Dmitri Potishko, pointedly questioned what would happen if the AI trade reversed or if all quantitative funds began selling at once.
“There are correlated risks that drive exposure in prime books,” he stated.
The chairman of the Wells Fargo Investment Institute’s global equities and real assets division, Sameer Samana, stated that his team was already concerned that technology stocks, especially those of chipmakers, had risen too rapidly.
In an email, he stated, “Really, markets were just looking for any excuse to sell.”
Asia’s AI gold rush is far from over. However, for the first time in months, those in charge of the funds are openly questioning what would happen if the tide turned.
Don’t just read crypto news. Understand it. Subscribe to our newsletter. It's free.

Grayscale has filed with the SEC to convert its Solana Staking ETF into a product that pays shareholders regularly. The fund outlined amendments to GSOL, allowing it to convert staking rewards into cash at least once a quarter and to pay the net proceeds to stakeholders.
Grayscale Asset Manager filed the new proposal with the Securities and Exchange Commission this week. The filing sought to change the ETF’s original structure, requiring the trust to convert its staking profits into cash at least quarterly.
The firm aims to distribute those proceeds, net of expenses, to shareholders.
Grayscale restructures GSOL to offer frequent rewards
Grascale began staking all of its SOL holdings, and those staking rewards currently yield approximately 6.1% annually, according to filings. Under the old structure, the yield accrued within the fund was reflected gradually in net asset value.
However, under the new proposed trust agreement, Grayscale will liquidate those rewards into dollars on a quarterly cadence, subtract trust expenses and sponsor fees, and distribute the profits directly to investors.
The U.S. Securities and Exchange Commission filing is careful to note that nobody should expect a fixed payout. The amount to be distributed will depend on the staking consideration actually received by the trust during each period and cannot be predicted with certainty.
According to the filing, the amount to be paid out each quarter will vary depending on validator performance on Solana and the prevailing staking yield at the time.
Grayscale used the same filing to formalize fee changes it had already begun phasing in before the amendment was announced. The sponsor fee dropped from 0.35% to 0.19% as of June 25. Grayscale also reduced the staking fee from 23% to 7%, increasing the potential payouts for investors.
However, Grayscale also noted that the new structure may result in different tax implications for investors. The firm advised its shareholders to consult their own tax advisors about the consequences of the new fund agreements.
The proposed changes will be effective on August 7, giving shareholders a few weeks to weigh what a quarterly Solana income stream actually means for their portfolios.
Grayscale’s SOL ETF
Grayscale launched the SOL ETF in November 2021 as a private placement and spent years trading over the counter before finally uplisting to NYSE Arca on October 29, 2025. The fund gave ordinary investors direct exchange access, and shortly after that transition, Grayscale began staking its SOL.
Earlier that January, Grayscale had already implemented the same strategy for its ETH ETF, making the implementation of the SOL ETF more of a copy-and-paste of an already working blueprint.
Don’t just read crypto news. Understand it. Subscribe to our newsletter. It's free.

A negotiated settlement has once again outpaced a long asset recovery process as one of the attackers of TrustedVolumes, a liquidity resolver linked to 1inch Fusion, has returned 1,122.12 ETH in an Ethereum transfer, more than two months after the incident. The hacker has allegedly retained a similar amount as part of a negotiated bug bounty agreement, as reported by Defimon Alerts.
When the deal was made, Ether’s price was roughly estimated at $1,843, thus making the total value of the recovery approximately $2.07 million and clarifying why this event is usually called the “$2 million recovery.”
The importance of the deal is not defined by how large it is, but by the fact that it reflects the increasing tendency in decentralized finance. Increasingly, organizations are turning to negotiating with their attackers as opposed to relying only on law enforcement or lengthy court cases. While this approach may provide quicker results, it raises the fear that extortionists will begin to view bug bounty negotiations as a way to avoid being caught rather than as an exception.
Both parties posted a message on-chain confirming the settlement.
“We have finalized the negotiations with the original exploiter,” the message said
More than 2 months after $5.8M @trustedvolumes exploit one of the attackers returned 1,122 ETH ($2M):https://t.co/N2nZ3xSVPn https://t.co/auPXtWRLT1
— Defimon Alerts (@DefimonAlerts) July 17, 2026
Moreover, the exploiter “returned the funds and received their bug bounty,” the message added, encouraging any remaining attackers to write to tvbugbounty@proton.me. TrustedVolumes hinted at this approach right after the exploit, saying that the company was willing to have “constructive communication regarding a bug bounty and a mutually acceptable resolution.”
What was taken in May
TrustedVolumes worked as a resolver in the 1inch Fusion environment, allowing for the operation of a Request-For-Quote (RFQ) market, which provided liquidity for token exchanges when needed. On 7 May, an attack took place whereby the resolver was exploited and approximately $5.87 million worth of digital assets were withdrawn in a single Ethereum transaction. Later data by the protocol suggested that the damage after accounting for asset values and related losses could reach about $6.7 million.
According to the analysis of Blockaid, in total $5.87 million was lost, which included 1,291 WETH, 1.26 million USDC, 206,282 USDT, and 16.93 WBTC. The compromised resolver contract was determined to be 0x9bA0CF1588E1DFA905eC948F7FE5104dD40EDa31, while the attack was connected to a bespoke RFQ proxy contract launched at 0xeEeEEe53033F7227d488ae83a27Bc9A9D5051756. The main attacker wallet 0xC3EBDdEa4f69df717a8f5c89e7cF20C1c0389100, however, was classified as a TrustedVolumes exploit address by Etherscan.
Cybersecurity researchers found that methods that were used in this attack were very similar to those used in the hack from March 2025 involving the 1inch Fusion V1 system.
Investigators determined that the hack happened due to an access-control vulnerability rather than the theft of private keys or some previously unknown exploit. Halborn discovered that due to a public function, any individual could register as an authorized order signer. The attackers then approved unauthorized orders transferring money that had already been approved for the proxy contract. Blockaid was able to identify the exploit while it happened and noted that neither 1inch’s infrastructure nor user funds were affected.
Why the incentives cut both ways
The settlement showcases the challenges the DeFi industry faces. It is often more desirable to negotiate and reclaim stolen funds rather than face the possibility of losing everything at the end of long and unpredictable investigations. However, these settlements can create a financial rationale that leads criminals to believe that they can steal money and then negotiate for a large payoff afterward.
Calculating the theft is becoming increasingly complex as the forensics in blockchain keep getting better. TRM Labs found that cryptocurrency scams robbed people of $2.87 billion in 2025 through approximately 150 incidents, but at the same time, investigators have been rather successful in tracing the stolen crypto and following the laundering routes. For instance, in the TrustedVolumes case, Blockaid, CertiK, and SlowMist were able to monitor the theft as soon as it happened and to track how the hacker converted the stolen goods into ETH, which helped make the criminals feel more uneasy before the deal was cut.
Only part of the case was settled through the deal. One attacker returned 1,122.12 ETH or approximately $2.07 million at the moment of the transaction; however, held a similar amount as stated in the proposed bug bounty. Recovery of the rest of the stolen funds may come from either reaching a settlement in the future or laundering the stolen money and will be an excellent gauge of how future hacks’ offenders weigh their chances of being traced by blockchain security against the possibility of negotiating their way out.
The smartest crypto minds already read our newsletter. Want in? Join them.

Venture capital’s biggest winners are changing the rules of the exit game, and 2026’s IPO boom is allowing them to cash in. US IPOs have reportedly already raised more than $141.2 billion this year. This puts the market within striking distance of the $142.4 billion record set in 2021. The biggest artificial intelligence listings are still expected to come.
For private equity firms investing in these firms, the time is virtually perfect. However, this might not be the case for long. Brookfield-backed data center operator Csquare is looking to raise up to $1.35 billion under its IPO scheduled for July 15, which may push the fundraising this year to a level never seen before. Nuclear startup Standard Nuclear is also aiming for its debut on that day and is trying to raise the amount of $384.3 million.
Widening the gap with the AI elite
There are even greater names ready to enter the scene. Stelios Saffos, partner at Latham & Watkins, told Axios that the recent boom of large offerings has been “as much of a green light as you could possibly get.” According to him, the investors jumped into these deals partially due to their beliefs that such companies as Anthropic and OpenAI will follow suit.
That momentum contributes to widening the gap between the select few elite venture firms and the rest of the sector. Reuters reported that the major gains made by AI leaders such as SpaceX, OpenAI, and Anthropic are distributed among a limited number of investors who have funded them in the early stages. Meanwhile, many traditional venture firms struggle to obtain their next funding rounds.
This level of concentration alters the power dynamics. Once most venture capital funds get invested into a handful of big investment deals, the companies that take part in them obtain more power in determining company valuations, seats on the Board, and exit points, while smaller investors benefit less from controlling any processes.
AI IPO growth will benefit only a few
The National Venture Capital Association maintains that venture capital investing still brings about important economic benefits.
Venture investing generates billions of dollars for investors, their institutions and creates millions of jobs. Many venture-backed companies have scaled, gone public, and become household names, and at the same time have generated high-skilled jobs and trillions of dollars of benefit for the U.S. economy.
It states that jobs at VC-supported companies increased by 960% between 1990 and 2020 compared to 40% in the overall economy and estimates that three-quarters of the largest venture capital-funded companies would not have grown to that extent had they not received venture financing. However, at this point one question becomes practically impossible to evade: how much of the current AI-driven growth is going to benefit companies that are not in the top rankings?
Still, the rally is fraught with dangers. As Axios pointed out, there’s growing interconnectedness to AI investment. The same hype that has reignited the IPO market can just as quickly reverse if investor mood soured.
A case in point: Tim Draper is a third-generation Silicon Valley venture capitalist and founder of Draper Fisher Jurvetson, Draper University and Draper Associates. A Stanford University electrical-engineering graduate with an MBA from Harvard Business School, he founded Draper Associates in 1985. He’s since been a backer of more than 60 “unicorns” at the seed stage, including Tesla, SpaceX, Twitch, Coinbase Global, Robinhood and Baidu. He weighs in on the AI bubble and four huge opportunities he missed.
In a recent interview, Draper shares how he identifies ideas that seem impossible, where he sees the next wave of AI opportunity, whether AI stocks are in a bubble and why he refuses to make a bear case — for anything. He compared AI’s current trajectory to the early internet era. His notable remark: “I invest in heroes.”
Is there a new order in the venture capital sector?
Other areas still continue to face difficulty. The Financial Times states London has achieved its lowest amount of money raised from IPOs in the last 30 years, which means investment banks and advisers have to count more on private markets, while asking policymakers to make changes. The situation has created the greatest difference between regions: record fundraising in New York due to AI while listing markets are still stuck elsewhere.
The next significant trial will occur when OpenAI and Anthropic eventually enter the public market. Their IPOs will show if the venture capital companies that supported them from the beginning can turn years of unrealistic profits into actual gains and solidify a new order in the venture capital sector.
The smartest crypto minds already read our newsletter. Want in? Join them.

Consensys, the blockchain firm behind MetaMask crypto wallet, has confirmed it accidentally brought a North Korea-linked software developer onto its team. The firm confirmed that they gave the operative access to the core wallet code before the company caught on and shut him down after a month.
Internal Slack messages revealed that the North Korea-linked operative worked on core MetaMask code for approximately one month before being terminated. Although Consensys confirmed that the infiltrator was stopped before any damage was done, the market remains skeptical of MetaMask’s ability to guarantee the safety of its users’ funds.
North Korean developer spent a month inside MetaMask
According to the report, the North Korean software engineer worked under the alias “Tyler Knapp” and used the GitHub handle “imyugioh.” He was hired as a consultant through a third-party service provider with a long-standing relationship with Consensys. Consensys confirmed that the software engineer was not hired directly through its internal hiring pipeline, insisting that the third-party hiring agency may have been responsible for the breakdown in proper screening.
Internal Slack messages reviewed show that Tyler Knapp worked on the core MetaMask platform code. He had access to the core MetaMask codebase that converts crypto to fiat currency via third-party payment providers and vice versa. Tyler also contributed to MetaMask’s mobile wallet codebase on GitHub.
Those contributions began on March 9 and abruptly stopped in April, the same month Consensys cut off his access, meaning the operative had roughly a month of activity within the company’s systems.
Consensys general counsel Matt Corva revealed that the company discovered the threat quickly after Tyler was hired. The company followed its security protocols and terminated access immediately upon identifying the threat. He said a subsequent investigation found no misappropriation of assets or data, no malicious code pushed into production, and no impact on user safety.
In April, Corva sent a company-wide alert ordering all product releases suspended pending investigation and instructing staff not to interact with the individual. He also asked employees to keep the matter internal while the probe continued, a request that suggests Consensys was trying to control the narrative well before the story became public this week.
Crypto remains a favorite target for Pyongyang
North Korean operatives posing as remote software engineers have repeatedly landed real jobs at American companies. These companies achieve this with the help of US-based facilitators running laptop farms that make it appear the worker is logging in from within the country.
One Arizona woman was sentenced last year for running such an operation, which prosecutors say generated more than $17 million for North Korea-linked entities, according to reporting from The Guardian.
Earlier this year, two more American nationals were sentenced for facilitating similar schemes that the Department of Justice says touched close to 70 US companies.
Crypto firms are an especially attractive target because a developer’s ordinary access can extend well beyond source code into transaction signing infrastructure, the layer where stolen funds actually move.
Blockchain analytics firm TRM Labs has estimated that North Korea-linked actors were behind roughly two-thirds of all crypto stolen in hacks last year, a figure that includes the $1.5 billion Bybit theft widely attributed to Pyongyang.
Don’t just read crypto news. Understand it. Subscribe to our newsletter. It's free.

The controversy regarding Base’s loss of credibility in the crypto community has now moved into the higher levels of Coinbase, creating doubts that go beyond social media. This controversy is important because today Base isn’t just another layer-2 blockchain anymore. Now it is processing more of the onchain stablecoin volume than any other blockchain and is also securing $11.61 billion in total value, as per L2BEAT, thus making it one of the main infrastructure pieces of the industry.
The conversation took place on X on July 18, after a crypto trader named Rune (@RuneCrypto_) had a chat with Cobie, who has just assumed the lead of the Coinbase Base application, regarding the reasons why users should pick the application. Cobie responded to Rune’s key concern that Coinbase has often been “a little bit in an ivory tower and a little bit distant from users,” especially those who are into crypto-centric businesses.
Coinbase admits trust gap
Coinbase’s admission marks a major transformation in the company, which recently reported a record high of 8.6% of the total crypto trading volume in the world as of Q1 2026. By admitting the gap between them and the users driving onchain adoption, the company is trying to win back the trust of the crypto sector.
Rune expressed doubts. In a X post, he described Coinbase as “a consumer app that doesn’t care about their consumers” and went on to say that recent developments have proved that “believing in anything @base-related for more than 24 hours is a mistake.” The two messages received hundreds of likes, which indicates that Rune was not the only one seeing things that way.
I took over the baseapp a few days ago yeah (as well as trading products in Coinbase – the Coinbase app, Advanced trading, etc). I don’t run Base chain.
I think that typically CB has been a little bit in an ivory tower and a little bit distant from users, particularly crypto…
— Cobie (@cobie) July 18, 2026
Cobie made sure to say that the application should not be confused with the blockchain technology. Currently, he is in charge of the Base consumer app and the trading products for Coinbase but insisted that he is not involved in the workings of the actual Base network.
Jesse Pollak is still in charge of the operations of the Base network. On July 16, Pollak announced that he would be handing the consumer app over to Cobie, so that he could focus on Base, which he calls the biggest chance ever to build a blockchain for the global financial system.
The change in leadership came after a difficult quarter. Just a day earlier, Pollak had remarked that the quarter had been “a punch in the face” and had unveiled his plans for the company, which would entail trading, a payment function, and AI agents. Moreover, he has acknowledged that Base is not successful in its foray into social apps and thus the project will now be focusing on its other activities which have potential in the longer run.
Armstrong already conceded the misstep
Brian Armstrong, the chief executive officer of Coinbase, admitted that the organization has made a mistake before the latest developments.
Armstrong talked about the failure of the “content coins” project of Base on July 13, saying, “We have messed up, time to turn the page.” According to him, the company shifted the focus away from the project at the beginning of the year and now prioritizes trade, payments, and AI agents.
His remarks mirrored Pollak’s assessment and seemed to indicate that Coinbase executives are united in their desire to move past the previous strategy based on social media. But whether this reset will lead to renewed confidence will depend more on execution than on public relations.
Why the dispute matters
For the larger cryptocurrency market, what is more important is how Base is growing as a player in on-chain finance. In its Q1 2026 letter to shareholders, Coinbase noted that Base handled 62% of the global on-chain stablecoin transaction volume–more than the total volume of all other blockchains together. The firm also said that Base processed above 90% of the global on-chain stablecoin volume with regard to agentic stablecoins, while the volume of stablecoin trades on Base grew tenfold compared to the previous year.
The mentioned numbers make Base one of the key settlement layers in the industry. A blockchain with such a huge volume of stablecoin transactions benefits from the confidence of developers, liquidity providers, and crypto-savvy users responsible for creating applications and transferring money in the networks. Any evidence of lack of trust in Base would have an impact beyond the limits of Coinbase.
Additionally, the discussion takes place while Base is dealing with another technical milestone. Currently, L2BEAT credits the network with the designation of a Stage 1 optimistic rollup, although it is likely to come down to Stage 0 in roughly 30 days due to its proof system not complying with the upcoming trusted-setup requirements.
While the governance review is not connected to the public row, they do raise the same issue of whether Base is capable of enhancing its technical roadmaps as well as the ties with the cryptocurrency community as it develops further. In the next days, we will be able to see whether Coinbase’s new approach towards crypto-native clients brings about certain product improvements and whether the anticipated L2BEAT reclassification proceeds as expected.
The smartest crypto minds already read our newsletter. Want in? Join them.

Pump.fun transferred another 81,712 SOL, worth roughly $6.15 million, to crypto exchange Kraken on July 18, 2026, according to on-chain analyst EmberCN. On its own, the move is just another wallet transfer. But in the context of slowing memecoin activity and weakening Solana network revenue, it offers another glimpse into how one of crypto’s biggest success stories is changing.
Amidst the memecoin trend, Pump.fun emerged as one of the most popular apps on Solana, turning over millions in fees as users created and traded tokens. As trading opportunity slows down, however, the application continues to convert part of its revenue into cash through transfers to Kraken.
As estimated by EmberCN, Pump.fun has reportedly sold approximately 4.81 million SOL tokens in the period between January 2024 and July 18, 2026. This translates to sales of around $812 million, with each token being sold at an average cost of $168.70. Together with this milestone, Pump.fun is now reputed to be among the top players in the field of long-term SOL selling in the current walking stick.
What was sold, and how much has left
The latest transaction is part of an emerging pattern, not an isolated incident. According to its X account, Pump.fun launched in January 2024, charging a trading fee of 1% for the swap of tokens. This business strategy has proved to be highly successful in the heyday of the memecoin boom.
According to CoinMarketCap, Pump.fun has generated more than $1 billion in cumulative revenue since its launch, including approximately $664 million in 2025 alone. The platform also produced about $124.7 million in revenue during the first quarter of 2026, underscoring the success of its 1% fee-based business model despite a broader cooling of the memecoin market.
Why do the numbers look different?
When examined closely, the figure put out by EmberCN is much lower than that provided by DefiLlama. This difference is due to the fact that they measure different aspects.
EmberCN’s figure ($812 million as of July 18, 2026) is simply the estimated number of SOL coins (around 4.81 million) that Pump.fun is believed to have sent to Kraken and supposedly sold.
DefiLlama’s figures (as of July 18, 2026) indicate a cumulative fee of around $1.86 billion, while cumulative revenue from the protocol stands at over $1.2 billion, representing all fees earned from Pump.fun‘s products.
In other terms, both sets of figures show different information. EmberCN sees the figures of actual sales of SOL, while DefiLlama evaluates the total fee income and revenue of the protocol.
The drag on Solana
Pump.fun‘s slowdown is also becoming visible at the network level.
According to an article published on June 16, 2026, by The Block, the platform suffered 80% decline in activity over the past three months. Average revenue per day in June dropped to around $800,000 compared to approximately $4.8 million six months before. Furthermore, only 0.26% of the newly issued tokens managed to reach the market capitalization sufficient to reach other exchanges.
The downturn is about more than merely a reduced interest in memecoins. Traders have been switching to perpetual futures protocols such as Hyperliquid, leading to a reduced amount of on-chain activity that had once made Solana one of the most active networks in crypto. Whenever there is one application that leads to a significant portion of blockchain activity, a slowdown in its usage will also affect the economics of the blockchain.
A business past its peak
As per The Block, the price of PUMP has reduced by nearly 40% over past six months and DefiLlama has suggested that as of July 18, 2026, the token is priced at nearly $0.0016, considerably lesser than its peak value of $0.0088. The total market cap of the token is around $655 million.
The difference is quite striking when we consider how quickly the interest of investors in the project increased. According to CoinMarketCap, Pump.fun managed to raise $600 million in an ICO that was finished in just 12 minutes.
The overall concept of Pump.fun has come under fire as well. According to STORM Partners, Pump.fun is the center of the memecoin boom of 2024-2025, with estimates suggesting the platform is responsible for releasing approximately 11 million tokens by mid-2025. However, in the end, just under 1% of released tokens actually make it to exchanges, such as Raydium. During spikes of activity, this constant stream of new tokens translates to significant profits. However, now that the trading activity has eased up, the same model has become reliant on decreasing activity levels.
What to watch next
The question that follows is whether or not Pump.fun is going to continue selling SOL at this rate even as protocol revenue slows down and whether Solana’s network activity starts picking up steady pace.
According to the wallet tracking by EmberCN, as far as July 18, 2026, Pump.fun still sends SOL to Kraken from time to time. Such transfers will surely continue being an essential indicator, but they are just part of the total situation. It is DefiLlama’s revenue stats and Solana’s network fee metrics that will tell us better whether user activity stops slowing down and whether the memecoin boom that helped the blockchain achieve astonishing growth is still continuing to fade away.


Don’t just read crypto news. Understand it. Subscribe to our newsletter. It's free.

Piero Cipollone, member of the executive board of directors of the European Central Bank (ECB), warned on Friday that the increasing popularity of stablecoins threatens Europe’s banking system because retail deposits will be detached from commercial banks. Speaking at a gathering of Italy’s cooperative banks in Rome on July 17, he said the trend adds to the pressure lenders already face from mobile payment platforms that have chipped away at payment revenues and customer data.
According to Cipollone, the ECB’s idea of the digital euro is the solution. The warning has special significance for smaller banks, which depend on deposits from people in order to finance local loans. If clients start to keep their money in stablecoins held in digital wallets rather than in their bank accounts, banks will need to substitute this source of income with more expensive wholesale borrowing and this may lead to the increase of lending rates in the economy.
According to him, the deposit risk is simply the latest step in the ongoing process of moving away from traditional banking. He stated that mobile payments are already covering more than 10% of the point-of-sale transactions in Ireland, the Netherlands, and Finland. Banks usually receive higher fees for mobile payments than they receive for debit card fees, but they also typically lose access to valuable customer transaction data that comes with those payments.
According to Cipollone: “If the use of stablecoins increases in the future, the banks will also lose retail deposits.”
The ECB emphasizes the dependence of Europe on foreign payment infrastructure. Approximately two-thirds of euro area card transactions are processed using non-European networks. Further, 13 out of the 21 eurozone countries lack a domestic card facility, thus, proving the need for a European-controlled payment system.
What stablecoins are, and why the dollar ones worry the ECB
Stablecoins are cryptocurrencies tied to fiat currencies, with the US dollar being the most common. Most of the market is rooted in Tether’s USDT and Circle’s USDC stablecoins, both of which are issued from outside the EU, which contributes to the ECB’s worries.
EU legislation regulates euro-denominated stablecoins under the Markets in Crypto-Assets (MiCA) framework but leaves US-dollar-pegged stablecoins mostly outside the framework’s direct oversight. It should be noted that euro stablecoin issuers must, under MiCA, keep at least 30% of their reserve assets in the form of bank deposits, with that figure rising to 60% for issuers valued as “significant.” These requirements, while helping banks operate under normal conditions, might lead to sudden bank outflows during mass redemptions activity.
Several ECB officials have voiced similar worries to Cipollone’s, including board member Isabel Schnabel, who stated in June that stablecoins pose a risk to both financial stability and monetary sovereignty and equated their emergence to the use of money-market funds that diverted banks’ deposits in the 1970s.
The digital euro as the ECB’s counter
The ECB highlights that digital euro will protect public funds in an increasingly digital economy while avoiding any dangers for commercial banks. The suggested central bank digital currency (CBDC) will not pay interest and will limit the amount of money each individual can keep in an account so as to prevent major withdrawals, while banks will still provide this service instead of customers doing it directly with the ECB.
“The digital euro would both preserve the role of public money and ensure banks remain involved in the payments ecosystem while continuing to meet their customers’ needs,” said Cipollone.
The ECB has recently made many remarkable moves to implement the plan by selecting 36 payment service providers, including Deutsche Bank, UniCredit, and Revolut for a 12-month pilot involving 19 national central banks. The beta pilot is scheduled for launch in the second half of 2027, and the European Parliament has voted for the initiation of formal legislative processes. Nevertheless, the bank is not expecting complete launch before 2029.
What it signals for the crypto market
According to the latest statement from ECB, Europe prefers token bank deposits and digital euro instead of stablecoins issued by private companies as the foundation for digital payment in the future.
This shift could limit the usage of stablecoins, even the ones that comply with MiCA’s requirements. Circle invested a lot of funds to meet the requirements of Europe, but representatives of ECB believe that regulation cannot solve every problem regarding monetary sovereignty and the movement of deposits. Circle has achieved compliance with the Markets in Crypto-Assets regulation (“MiCA”), the EU’s landmark crypto law. Of the top ten stablecoins by market cap, only USDC is in compliance with the new EU rules, and while smaller in market cap, EURC is also MiCA compliant. USDC and EURC are uniquely positioned to provide solutions for the European Union’s 450 million residents.
The strategy reaches beyond retail payments as well. The European Central Bank is working on Project Pontes, which aims to settle tokenized assets using central bank money. The bank is also working on Project Appia, which aims to create a public-private marketplace for tokenized finances, highlighting the bank’s effort to maintain the central role of commercial banks and central bank money in the digital financial system of Europe.

The smartest crypto minds already read our newsletter. Want in? Join them.

GameStop (NYSE: GME) now owns almost one-tenth of eBay (NASDAQ: EBAY) after turning a large set of financial contracts into actual stock.
A filing released late Friday showed the retailer holds 43.39 million eBay shares, equal to 9.8% of the online marketplace. Ryan Cohen is still chasing a takeover even though eBay rejected his first proposal in May.
The size of the holding has nearly doubled since early May. At that point, GameStop controlled about a 5% financial interest through derivatives and beneficial ownership.
Ryan had already told eBay’s board chairman that he wanted to buy the business for about $56 billion. The target is roughly five times larger than GameStop, so the funding plan has drawn heavy attention from Wall Street.
GameStop turns its eBay contracts into stock and gains voting power
GameStop bought about 3.5 million eBay shares for nearly $381 million in mid-June. It then completed the physical settlement of put and call pairs tied to around 39 million shares on July 15.
Those deals took its total position to just under the 10% line. The ownership position disclosed on Friday had 43.4 million shares outstanding, while the precise number of transactions was 43.39 million when the two legs were added up.
Prior to settlement, the bulk of the position was held in derivatives. The derivative contracts enabled GameStop to gain from any movements in the stock price of eBay, but without having the shares in their possession or owning them outright. Taking delivery of the stock reversed this situation.
A second filing published Friday included comments from Ryan’s Bloomberg Television interview from Thursday. “I’m not going to call my shots, but we’re coming for eBay one way or another,” he said.
Ryan first pitched the deal two months earlier. He said a combined GameStop and eBay, under his control, could compete more directly with Amazon (NASDAQ: AMZN). eBay turned down the mix of cash and stock and called it “neither credible nor attractive.”
Ryan did not drop the idea after the rejection. He repeated in several interviews that he still planned to pursue the company.
GameStop investors also approved a larger authorized share count last month, giving the board more room to use stock while arranging a deal. Ryan has also committed $500 million of his own money.
Ryan keeps the bid unchanged while Wall Street questions the financing
The biggest concern is still the debt package. GameStop is leaning heavily on a nonbinding letter from TD Securities for as much as $20 billion. That money depends on the combined business receiving an investment-grade credit rating, so the financing is not locked in.
The letter does not force TD Securities to provide the cash, and no final loan agreement has been announced yet.
Ryan pushed back during the Bloomberg interview. “There has been a complete failure by the media to explain why this transaction makes sense,” he said. He also said GameStop has a “highly confident” letter from its bankers and added, “We have a lot of parties that are interested in this transaction.”
Bloomberg also asked whether he would offer more money for eBay. “I’m not going to negotiate against myself,” Ryan replied.
Retail traders were split across the two stocks. Stocktwits showed GME sentiment in bullish territory during the past 24 hours, while message activity stayed at normal levels. EBAY sentiment remained bearish.
The market performance also favored eBay. EBAY shares were up 29% year to date, while GME stock had gained 9%.
NOTE: The author of this article owns GME stock.


The smartest crypto minds already read our newsletter. Want in? Join them.

Etched, which is a four-year-old chip startup, claims to have received more than $1 billion worth of orders of hardware that is basically designed to run AI models after they have already been trained. The process of running AI models is known as inference and this has become one of the costliest operations for companies using AI in their business operations, and Etched believes that chip technology can do the job more quickly, cheaply, and with less energy use than Nvidia’s general-purpose GPUs.
The company disclosed its orders together with a valuation of $5 billion. It added that its first chip is already made by Taiwan Semiconductor Manufacturing Company (TSMC), as per a report by TechCrunch. If the chip works as expected, a portion of the money currently flowing to Nvidia would be diverted, leading to a change in the economics around deploying frontier AI models.
“We’ve hit a point in the evolution of AI where specialized chips that can perform better than general-purpose GPUs are inevitable — and the technical decision-makers of the world know this,” Etched co-founder and CEO Gavin Uberti told TechCrunch in 2024.
Unlike training where models learn from huge datasets, inference takes place whenever a user submits a prompt. Currently, large language models make use of the time looking for weights and updating their key-value (KV) cache which increases the importance of the memory capacity and bandwidth as context windows get bigger. As a result, there is a demand for chips designed for inference only as opposed to GPUs capable of handling heavier workloads.
A paper published in July 2026 by Michael J. Yuan and Ju Long states that mainstream GPUs can be characterized as “compute-heavy and capacity-light,” owing to their combination of high processing power with relatively low memory. According to the authors, during inference, some computations are often left unused while waiting for data, which creates a possibility to use hardware designed specifically for optimizing memory consumption rather than arithmetic efficiency.
What Etched is selling
Instead of marketing a standalone chip, Etched provides “frontier inference clusters” which are an assemblage of systems functioning together comprising of custom silicon, networking units and software. The firm claims that over 400 engineers are employed by it who have been recruited from Nvidia, Google TPU team, Broadcom, SK Hynix and TSMC.
The firm has attributed the expected enhancement in performance to two proprietary technologies. The first one is Low Voltage Inference (LVI), a technology meant to allow computing blocks operate at lower voltage levels to achieve continuous operation with lower heat production. The other technology is the Cluster Scale Memory (CSM), which free chips in a low-latency storage system and thus eliminates problems with long-context inference and KV-cache size growth.
These technologies, however, are still considered claims made by the company rather than being confirmed independently. Etched has claimed that it expects to make some of its benchmark results and technical information public as soon as it starts shipping its first product racks later this summer.
From near-bankruptcy to a crowded cap table
Etched has raised roughly $800 million to date, including a $500 million financing completed in December at a $5 billion post-money valuation, according to TechCrunch. Stripes led the round, with participation from Jane Street, Hudson River Trading, Two Sigma, Ribbit Capital and VentureTech Alliance.
Its investor roster also includes prominent AI researchers Andrej Karpathy, Geoffrey Hinton, Fei-Fei Li, Arthur Mensch and Scott Wu, as well as Stanley Druckenmiller and Peter Thiel.
This backing represents a complete change of direction for the startup. Etched, which was established in 2022 by Harvard dropouts and Thiel Fellows Robert Wachen and Gavin Uberti, ran into trouble trying to attract investors for most parts of 2023. The creators of the company have stated that almost every organization that they turned to dismissed their requests before they managed to get over $125 million in funding in 2024.
A widening race the industry is watching
Etched is entering an already populated market. Cerebras had one of the most highly publicized AI chip IPOs this year, Groq got a $650 million raise recently, while Amazon, Google, and Microsoft are creating their AI custom chips for their own infrastructure. OpenAI has also hired Broadcom to create a custom chip, pointing to the fact that the inference market has reached a level where it can support other players apart from Nvidia.
The manufacturing process continues to be a struggle. As reported by Cryptopolitan earlier, advanced packaging technology, especially TSMC’s CoWoS method of joining processors with high-bandwidth memory, has become one of the biggest obstacles to the industry.
The demand for AI wafers is expected to increase nearly eleven times from 2022 to 2026. Meanwhile, TSMC has about seventy-two percent of the market of the pure-play foundry. Since Etched uses TSMC’s N4P process, it ends up competing with the likes of Nvidia, AMD, and other AI chipmakers for the same manufacturing resources.
At the moment, the focus is on implementation. Reports indicate that production has already begun and that the first inference racks will be shipped in the coming months to get about $1 billion in sales orders. The extent to which hardware has delivered on the company’s promises of improved features will only be subject to independent verification once it reaches customers.
For the time being, the company’s claims about LVI and CSM should be regarded as promises rather than confirmed results. If the results turn out to be true, the start-up is likely to become one of Nvidia’s fiercest competitors in the AI inference sector.


The smartest crypto minds already read our newsletter. Want in? Join them.

The FTX Trading Ltd. company and the FTX Recovery Trust announced on Friday that about $900 million would be paid out to creditors on July 31. This is the fifth major payout made by the company since the Chapter 11 restructuring of the failed exchange commenced in early 2025.
Creditors eligible for payment under the Convenience and Non-Convenience Classes of the Plan who have completed all pre-payment procedures by the June 16 record date will be paid in 1-3 business days following July 31 via BitGo, Kraken, or Payoneer, according to FTX’s official statement. The total repayments of the estate will exceed $10 billion after this payment round, The Block reports.
FTX creditors get another class-based payout
The fifth distribution’s incremental percentages vary by class. Dotcom Customer Claims (Class 5A) receive an additional 9%, bringing cumulative recovery to 105% of allowed claim value.
US Customer Claims (Class 5B) get an additional 5% to reach the same 105% level. General Unsecured Claims and Digital Asset Loan Claims each receive 3%, reaching 103% cumulative. Convenience Class creditors, who represent the smaller retail claimants that comprise the majority of FTX’s creditor base, sit at a cumulative 120%.
An additional $18 million distribution to Preferred Equity Holders is scheduled to be made on July 31 by the Preferred Shareholder Remission Fund Trust, resulting in a total of $95 million in PSRFT payments. This should not be added to the creditor distribution as far as customer recovery amounts are concerned.
Fifth Distribution has become possible partly because of the court-approved lowering of the amount reserved against disputed claims from $2.4 billion to $1.8 billion.
Creditors still lose the crypto rally despite 105% recovery
The above-100% recovery figures come with an important qualification. All FTX recoveries are calculated using the dollar-denominated value of each creditor’s holdings as of November 11, 2022, when the exchange filed for bankruptcy. Bitcoin traded at roughly $16,000 that month.
A creditor who held 1 BTC on FTX at collapse would receive approximately $16,800 in cash under the 105% recovery today. That same 1 BTC would be worth over $65,000 at current market prices. Similar math applies to ETH, SOL, and other tokens the estate liquidated to fund distributions.
As Cryptopolitan earlier reported in third-wave coverage, the 118-142% recovery framing has been a consistent point of contention among the FTX creditor community. The estate has drawn specific criticism for repaying in dollars rather than in kind, effectively locking creditors out of the 2023-2026 crypto bull run. Distribution mechanics matter too. Payoneer largely delivers fiat directly to bank accounts, meaning creditors receiving payments through that channel get dollars rather than crypto.
Remaining claims will stretch the process into 2027
The fifth distribution’s smaller size compared to earlier waves reflects the winding-down of the bankruptcy process. Wave one distributed roughly $1.2 billion in February 2025 to Convenience Class creditors. Wave two reached $5 billion in May 2025. Wave three delivered $1.6 billion in September 2025. Wave four was the largest single distribution at $2.2 billion in March 2026. This fifth wave at $900 million is the smallest major distribution since the process began.
The trust is now processing the remaining disputed claims, illiquid venture investments, and international recovery matters that will extend timing into 2027 for some classes. FTX has not set a record date or a size for its next creditor distribution, which will depend on available cash, disputed-claim resolution, and further recoveries.
The recovery trust continues to caution creditors against fake claim portals that impersonate the official process.
Sam Bankman-Fried, whose fraud collapsed the exchange in November 2022, is serving his 25-year sentence. Former Alameda Research CEO Caroline Ellison was released in May 2026.

If you're reading this, you’re already ahead. Stay there with our newsletter.

The president of France’s National Gambling Authority instructed the country’s internet service providers to block access to Polymarket on July 16, according to a statement issued by the authority on Friday.
The order escalates France’s November 2024 action, when regulators barred financial transactions from French accounts to the site. Even that did not stop traffic. According to the ANJ, Polymarket drew 578,751 visits and 205,057 unique visitors from France in June, French media reported.
Advertising, by any means whatsoever, in favour of an unauthorised betting or gambling site is a criminal offence, with fines reaching €100,000 ($114,000).
– ANJ
France targets Polymarket at the network level
The move pushes French law enforcement from the transactional level to the network level. According to the order issued in November 2024, the French banks had to block any transactions to Polymarket, but people would still be able to access the website and see the odds. That reading pattern is what the ANJ now classifies as illegal advertising, per the regulator’s statement.
The novel legal theory here is that the mere display of market prices in real-time to French citizens constitutes solicitation for participation in an illegal gambling market. According to this theory, mere accessibility becomes the crime. The regulator said the site would remain blocked “for as long as authorities considered the platform noncompliant with the country’s gambling regulations.”
As of last month, Polymarket has generated over $1 billion in annualized revenue through activities predominantly based outside of the 33+ jurisdictions that have banned it.
One French whale case became a wider traffic problem
The France crackdown started with a single French trader who placed roughly $30 million in bets on the 2024 US presidential election. As Cryptopolitan earlier reported, the ANJ opened its formal investigation in November 2024, and Polymarket responded by implementing IP-based geo-blocking that displays a restricted-jurisdiction notice to French users. Bettors kept accessing the platform through VPNs.
France’s national weather agency Meteo-France filed a criminal complaint in April 2026 after one of its weather probes was hacked in an attempt to manipulate weather-related bets on Polymarket.
Similar concerns are running through parallel enforcement in the US, where a soldier was recently charged with using classified information about the January operation to capture former Venezuelan president Nicolás Maduro to place bets on prediction markets, allegedly netting more than $400,000.
The White House said Thursday it suspended a teleprompter operator over allegations of placing prediction market bets.
US regulators protect Kalshi as Europe blocks Polymarket
France’s escalation occurs three days after the CFTC used its emergency powers to trump a decision by a Michigan state court to void the executed trades of Kalshi. The two sets of regulatory approaches to prediction markets have officially diverged.
France, Spain, Netherlands, Belgium, Germany, Italy, Portugal, Switzerland, Brazil, Indonesia, Singapore, South Korea, and Japan have moved to block or restrict access, per Cryptopolitan’s June coverage of the FIFA World Cup crackdown.
The US Commodity Futures Trading Commission has moved the opposite direction, defending federally registered prediction platforms from state-level enforcement while writing new rules to expand event-contract trading.
Polymarket finds itself on that dividing line.
The legal jurisdiction of Polymarket is getting narrower, but at the same time, its profits are growing. It was this very contradiction that French regulators found to be unreasonable in the Friday ruling: the website was still making money and attracting French users without using French payment systems.
If you're reading this, you’re already ahead. Stay there with our newsletter.

Amazon-owned Zoox issued a software recall on Friday for its entire fleet of 105 public-road robotaxis, after one of its vehicles drove into a smoke-obscured active fire scene on June 20, per the NHTSA recall document.
U.S. Department of Transportation Part 573 Safety Recall Report | Source: NHTSA
The vehicle was carrying no passengers. It encountered heavy smoke that obscured an active emergency scene that had not been cordoned off with traffic cones. The robotaxi braked hard while attempting to steer away, then came to a stop inside the scene.
A Zoox teleguidance tactician took remote control of the vehicle and reversed it out of the area, allowing first responders to close two of the three lanes running through the scene with cones after the fact, per Engadget.
Zoox fixes how its robotaxis respond to heavy smoke
Zoox investigated the incident and held several rounds of talks with NHTSA through late June and early July on severity, recurrence patterns, and root cause. The company settled on the recall July 7 and shipped the software update to its fleet on July 15, per the NHTSA filing.
The company said that the update “enhances existing capability of detecting and responding to heavy smoke”, adding the ability to spot and react to thick smoke in certain conditions. Zoox owns and directly controls the entire recalled fleet, so no third-party owners or dealers had to be notified under federal recall rules.
Zoox currently offers free rides in parts of Las Vegas and San Francisco and lets select users hail rides in small zones in Miami and Austin. As Cryptopolitan earlier reported, the Las Vegas free-ride operations began in September 2025 with Zoox’s purpose-built vehicles that have no steering wheel or pedals.
NHTSA warns AV firms after emergency-scene failures
The Friday recall filing arrived a day after NHTSA Administrator Jonathan Morrison sent a public letter to autonomous vehicle developers warning them to stop interfering with first responders. Morrison identified “a clear pattern of driverless AVs interfering with law enforcement and other first responders,” citing incidents in which AVs drove into active emergency scenes, blocked ambulances, or failed to recognize flashing lights, flares, smoke, fire, or traffic cones.
Morrison told developers that emergency scenes are “not rare or extreme edge cases” and called on companies to focus resources on the problem, with solutions required by end of month. That reframing matters because it moves emergency-scene detection from an area where AV companies can defend gaps as unusual circumstances to one where NHTSA treats gaps as functional failures.
Zoox had settled its recall internally by July 7. Morrison’s letter went out July 8. The Friday filing under NHTSA’s recall process converts an internal software update into a public accountability act.
Waymo recalls show the same first-responder risk
Zoox is not alone. Alphabet’s Waymo, the leading paid robotaxi operator in the US, recently recalled thousands of vehicles and suspended freeway operations after its cars drove through construction zones at speed.
Reports had previously counted at least six incidents through March 2026 in which first responders had to physically move Waymo robotaxis away from an emergency scene.
Zoox previously issued several software recalls last year to address unrelated issues, including a March 2025 update addressing a hard-braking issue NHTSA had been probing since 2024, and two May 2025 recalls after a Las Vegas collision with a passenger car and an e-scooter strike.
Zoox’s commercial launch hinges on NHTSA cooperation. Because its purpose-built robotaxis ship without a steering wheel or pedals, the company needs an exemption from certain Federal Motor Vehicle Safety Standards. NHTSA has separately floated dropping the brake-pedal requirement for vehicles designed to run fully autonomously. The FMVSS exemption path is the specific reason Zoox filed a Friday recall on top of the July 15 software deployment.
Don’t just read crypto news. Understand it. Subscribe to our newsletter. It's free.

A Dutch court has declared the Knaken crypto exchange bankrupt after prosecutors claimed it failed to account for over $8 million in customer funds. Knaken was declared bankrupt on July 16th, having previously sponsored Ajax and Feyenoord.
The Rotterdam court declared both Knaken Cryptohandel B.V. and its client-fund foundation bankrupt. The prosecutors determined that over $8 million in client funds belonging to more than 30,000 customers had disappeared without explanation. Knaken had already closed down earlier this month, and Dutch authorities had raided the premises to confiscate assets.
Dutch authorities point out suspicious activities at Knaken
Dutch authorities flagged suspicious activities at Knaken earlier last month. The Dutch Authority for the Financial Markets referred to the activities at the exchange as very concerning. As a result, the Public Prosecution Service petitioned for bankruptcy in late June following a criminal probe into the missing funds.
The court ruled that Knaken has many customers and a significant cash deficit, of which the customers were never informed. The judges argued that declaring the exchange insolvent was the best option that served the public interests following the misappropriation of user funds.
The prosecutors estimated that roughly 30,000 customers used the platform, and that they may recover only a fraction of what they deposited. The court’s ruling noted that Knaken customers have been shut out of the trading platform entirely and can no longer see their accounts or balances, and the company simply doesn’t have enough capital left to make everyone whole.
The exchange had reportedly been struggling in the lead-up to this final blow: the bankruptcy ruling. Before then, the native app and website were shut down, and Dutch authorities had also raided the company premises and seized computers, phones, and part of the company’s assets.
Knaken reportedly asked authorities to hold off filing damage claims after the shutdown, a move that did little to calm nerves among customers already locked out of their holdings.
Knaken’s defense points to a structural safeguard
Knaken’s defense argued in court that the exchange had a plan in place to protect customers in the event of such a scenario. According to their defense team, the customer funds were supposed to sit in a separate foundation, Stichting Knaken Payments, created so clients wouldn’t lose their funds if the company failed.
However, the foundation never began paying anyone out, with Knaken citing the need for careful legal and operational preparation first.
The court wasn’t persuaded by that explanation and declared both the trading company and the foundation bankrupt in the same ruling.
The exchange had pushed back against the bankruptcy request, arguing there were better ways to unwind the business. The company maintained that customers’ interests were already protected through other criminal law measures, including assets seized by FIOD, and proposed simply distributing available funds among its customers instead.
For now, the criminal investigation into where the €7 million actually went remains open, and thousands of former Knaken users are left waiting to see how much, if anything, comes back to them through the bankruptcy proceedings.
If you're reading this, you’re already ahead. Stay there with our newsletter.