Crypto is getting harder to separate from traditional finance, and this week makes that especially clear. 🏦 Stablecoins are moving into community banks, payment cards and cross border banking infrastructure, while tokenization is spreading across stocks, corporate bonds, gold and even agricultural commodities. Meanwhile, institutions such as Nasdaq, S&P Global and major banks are putting more money, data and infrastructure behind onchain markets. 🔗
At the same time, networks are still evolving underneath all of this activity, with Solana expanding transaction capacity and recording an extraordinary surge in new token creation. Add new regulatory moves across Europe and Asia, plus a packed AI Spotlight, and there is plenty to unpack. 🚀
Coinbase and payments platform Moov are teaming up to bring stablecoin infrastructure to more than 1,000 US community banks and credit unions, giving smaller financial institutions access to technology that until recently was mostly associated with crypto companies and large global banks. 🏦 The setup will support stablecoin payments, merchant settlement, payouts and real time funding, while businesses will also be able to access Coinbase custodial accounts through the infrastructure. Instead of every bank building its own blockchain stack from scratch, Moov can connect its existing banking customers with Coinbase’s digital asset infrastructure behind the scenes. 🔗
The interesting part is that customers may eventually use stablecoins without their local bank having to suddenly transform into a crypto exchange with a shiny new logo. This also fits a much wider trend, as major US banks are experimenting with their own digital money and stablecoin based settlement systems. Stablecoins are slowly moving from crypto wallets toward ordinary banking infrastructure, and community banks apparently do not want to watch that transition from the sidelines. 💵
MoneyGram is launching a Visa stablecoin card that will initially be available as a virtual card in Colombia, allowing eligible users to hold a stable dollar balance and spend it anywhere Visa is accepted. 💳 The card is being developed with stablecoin infrastructure company Rain and will work with Apple Wallet and Google Wallet, while MoneyGram plans to introduce a physical version and expand into more markets later. The move follows MoneyGram’s growing blockchain strategy, which already includes its MGUSD stablecoin, a self custody wallet and crypto ramps connected to networks such as Stellar and Solana. 🌎
In practical terms, the company is trying to make the technology disappear into a familiar experience: users hold digital dollars, but at the checkout the result looks like an ordinary card payment. Rival Western Union has been moving in a similar direction, suggesting the traditional remittance industry is becoming a surprisingly active testing ground for stablecoins. The old money transfer counter is not disappearing tomorrow, but it is definitely getting a blockchain powered upgrade behind the curtain. 🚀
Kraken has introduced yield vaults for selected xStocks, allowing eligible users to earn returns on tokenized versions of Nvidia shares and major US stock market ETFs. 📈 The first supported assets are NVDAx, SPYx and QQQx, with deposited tokens being supplied to onchain lending markets where borrowers pay for access to that liquidity. Users keep exposure to the underlying tokenized asset, while rewards accumulate in the same asset rather than requiring them to swap into a separate yield token. 💰
Kraken says withdrawals are processed within three days, and the infrastructure uses Veda together with strategies managed by Sentora, including lending through DeFi protocols such as Kamino on Solana. The service is available in selected markets including parts of the European Economic Area, while users in countries such as the US, UK, Canada, Australia and UAE are excluded. It is an unusual crossover where an asset tracking Nvidia or the S&P 500 suddenly starts behaving a little like a DeFi position, which is probably not something traditional brokerage textbooks prepared us for. 🤓
UniCredit is reportedly exploring partnerships with technology providers that could help the Italian banking group build infrastructure for digital asset trading and custody. 🏦 The discussions are still described as early stage, so this is not a finished crypto product launch, but the areas being considered reportedly include brokerage services, tokenized investments and stablecoin based fixed income products. Instead of building every blockchain component internally, the bank appears to be studying whether specialist infrastructure could provide the technical layer while UniCredit keeps the customer relationship and regulated banking environment. 🔧
The timing is interesting because other large banks are taking similar steps, from institutional crypto trading to proprietary stablecoins and tokenized settlement systems. This suggests the question for traditional banks is increasingly changing from “should we touch digital assets?” to “which infrastructure should we use if we do?”. UniCredit has not yet revealed a technology partner, so for now the crypto engine room is being designed before anyone presses the start button. 👀
DoubleZero has added real time data from Kalshi’s US election and politics markets to DoubleZero Edge, its low latency platform designed for institutional, AI and automated trading systems. 📊 The service already carried Kalshi sports and crypto market data, and the expansion now adds election contracts together with order book and trade information. The idea is not to create another prediction market, but to make existing market data easier and faster for professional systems to consume. ⚡
Interest in political prediction markets has grown considerably, with reported trading in contracts linked to the 2026 US midterms already exceeding the total recorded for congressional election contracts during the 2024 cycle. Market prices in these products represent positions taken by traders and should not be treated as the same thing as polling data or a definitive forecast of an election result. In other words, election night is still election night, but the machines now have a much faster data feed while they wait for it. 🤖
Solana has increased its maximum transaction size from 1,232 bytes to 4,096 bytes, giving developers more than three times as much room inside a single transaction. 🎒 The upgrade is especially useful for applications involving zero knowledge proofs, large multisignature operations and newer signature schemes that previously could be too large to fit comfortably into Solana’s transaction format. It arrives together with the new v1 transaction format, while older formats remain supported so existing apps and wallets do not suddenly wake up broken after an upgrade. 🛠️
Developers who want to use the larger capacity will need to adopt the new format, meaning the extra space is available but not automatically used by every application. The change may sound tiny when measured in kilobytes, but blockchain transactions are extremely compact, so moving from 1,232 to 4,096 bytes opens the door to noticeably more complex operations. Sometimes a major blockchain upgrade is not a dramatic new token or flashy app, but simply giving developers a bigger digital envelope. ✉️
S&P Global has led a strategic investment in crypto data company Kaiko, extending its Series B funding to a total of $110 million. 💰 The investor group looks rather more like a traditional finance conference than an early crypto meetup, with names including BNP Paribas, Nasdaq Ventures, Royal Bank of Canada, Broadridge and several blockchain focused companies. Kaiko plans to use the funding to expand both its digital asset data business and infrastructure for tokenized markets covering assets such as bonds, equities, Treasury bills and money market funds. 📊
The investors are also joining a Kaiko led working group focused on data standards and infrastructure for tokenized financial products. That matters because putting an asset onchain does not magically solve questions about pricing, reference data, valuation, liquidity and market transparency. Tokenization may be the shiny part of the story, but institutions are also discovering that someone still has to provide reliable numbers underneath it all. 🧮
Consensys Software plans to separate into two independent companies, turning MetaMask into a standalone consumer focused business while creating a separate Consensys for Ethereum protocols and institutional infrastructure. ✂️ MetaMask will continue under Joe Lubin as chairman and CEO, while the new Consensys will include businesses such as Linea, Besu and Teku and focus heavily on blockchain infrastructure for financial institutions. MetaMask, meanwhile, wants to move beyond being mainly a crypto wallet by expanding further into payments, savings, investing and access to traditional financial products. 👛 Importantly for existing users, the company says nothing changes immediately to their app, assets, keys or access as a result of the restructuring.
The separation is expected to be completed by the end of 2026 and reflects how different the consumer wallet market has become from the institutional blockchain infrastructure business. One company apparently needed two different wardrobes, so instead of squeezing everything into the same closet, Consensys decided to build another one. 👔
Blockchain intelligence company TRM Labs has reached a $2 billion valuation following an expansion of its Series C financing led by Blockchain Capital. 🔍 The company says its annual recurring revenue has quadrupled over the past three years, while its software is now used by more than 600 government agencies and private institutions across 75 countries. TRM originally focused heavily on blockchain analytics, but its newer strategy is increasingly centered around AI assisted investigations covering fraud, money laundering, sanctions evasion and other forms of digital crime. 🤖
That combination has become more relevant as criminals also adopt automation and artificial intelligence, creating something of a technological cat and mouse game between investigators and bad actors. The latest financing doubles TRM’s valuation compared with its Series C round earlier in 2026, although the company did not disclose the size of the new investment. Apparently there is strong investor demand not only for building digital finance, but also for figuring out what exactly everyone is doing with it. 🕵️
Gemini has received a Major Payment Institution license from the Monetary Authority of Singapore, completing a regulatory process that began with an in principle approval almost two years earlier. 🇸🇬 The license allows Gemini’s Singapore entity to provide regulated digital payment token services and cross border money transfers without the transaction volume limits that apply to standard payment institutions. That does not mean regulation disappears, as major payment institutions are subject to more extensive requirements precisely because they are allowed to operate at a larger scale. 🏛️
Gemini already provides spot crypto trading, custody and over the counter services in Singapore and has served customers there since 2020. For the exchange, the license strengthens Singapore’s role as one of its main hubs for both retail and institutional operations in Asia. Crypto companies used to celebrate simply entering a new country, but increasingly the headline is about obtaining the right license before unpacking the boxes. 📦
U.S. Bank has completed a live cross border payment using USBDC, its proprietary dollar backed stablecoin, between its entities in North America and Europe. 🌍 The transaction took place on the public Stellar blockchain and tested not only transferring the token, but also functions such as minting, redemption, freezing and clawbacks. Just as importantly, USBDC was connected with the bank’s existing finance, compliance, risk and operational systems, which is where many real world blockchain projects become considerably less glamorous and considerably more important. 🏦
The pilot validated the bank’s internally developed Digital Asset Platform, designed to connect tokenized assets with traditional banking infrastructure. U.S. Bank is now exploring possible uses in treasury operations, liquidity management and the movement of collateral onchain. The experiment is a good reminder that banks may not need to replace their existing systems with blockchain, but could instead place blockchain rails underneath parts of the machinery they already use. 🔗
Standard Chartered has started covering Sky’s SKY token and published a $0.325 price target for the end of 2028, compared with about $0.065 referenced in its report. 📊 The bank’s analysis is based on the idea that Sky, formerly known as MakerDAO, could continue expanding its USDS stablecoin ecosystem and generate more value for token holders through staking rewards and token buybacks. Sky is unusual because the protocol combines stablecoin issuance, lending and governance mechanisms, leading Standard Chartered’s digital asset research team to compare some of its functions with those of a financial institution. 🏦 The report also points to the scale of sUSDS, Sky’s yield bearing stablecoin product, which had around $4.5 billion in total value locked at the time of the analysis. Still, the $0.325 figure is an analyst forecast rather than a promised destination, and crypto markets have a well documented habit of ignoring even very carefully prepared spreadsheets. 😅
What is notable is that a global bank is now applying relatively traditional valuation thinking to a DeFi governance token, which itself says plenty about how the market is evolving.
Robinhood CEO Vlad Tenev has argued that companies should not automatically have veto power over tokenized products linked to their shares when those products do not change shareholder rights or create obligations for the issuer. 📃 The discussion followed criticism from AMC Entertainment CEO Adam Aron, who stressed that AMC had no affiliation with Robinhood’s tokenized AMC product and said the company would review the matter with securities lawyers. Robinhood says its stock tokens are separately issued instruments backed one to one by underlying shares, giving users economic exposure without changing the company’s official shareholder register. 🔗
Tenev’s position is that issuer approval makes sense when tokenization changes the issuer’s legal responsibilities or shareholder rights, but not necessarily when a third party creates a separate instrument referencing freely transferable shares. The debate highlights an important unresolved issue in tokenized finance: owning exposure to a stock is not always the same thing as directly owning the stock itself. Tokenization may make markets more digital, but it clearly has not eliminated the lawyers. ⚖️
Anchorage Digital is adding institutional access to Frgmnt’s fUSD stablecoin and its yield bearing sfUSD token through its regulated custody platform. 🏦 Institutional clients will be able to hold, mint, redeem, stake and unstake the assets without creating a separate custody setup, which removes one operational hurdle for firms experimenting with onchain finance. Frgmnt issues fUSD on Base against USDC and deploys its backing into onchain lending markets, while users can stake fUSD into sfUSD to receive returns generated by those strategies. 💵 There is an important reality check, however, because Frgmnt remains a very small protocol compared with established stablecoin platforms and was operating with roughly $100,000 in total value locked while still in a capped beta.
Frgmnt reported a double digit annualized yield for sfUSD in early September, but that rate is variable and depends on conditions in the underlying lending markets rather than being guaranteed. The bigger story is therefore less about the size of fUSD today and more about regulated custodians making even relatively new DeFi products accessible through institutional infrastructure. 🔐
Japanese Bitcoin treasury company Metaplanet is reducing the potential share count connected to its Series 10 stock acquisition rights after investors raised concerns about dilution. 📉 The company will cut the pool from roughly 319.5 million potential shares to about 188.2 million, meaning around 131.3 million potential shares are being removed from future exercises. Shares already issued through earlier exercises are not being cancelled, so the change mainly reduces possible future dilution rather than reversing what has already happened. 🧮
At the same time, Metaplanet plans to establish a Hong Kong subsidiary called Metaplanet Asset Management Asia Limited with $1 million in initial capital. The new company is expected to trade Bitcoin, equities and credit products during Asian market hours as part of a broader strategy called Project Nova. So while Metaplanet is tightening one part of its corporate structure at home, it is simultaneously opening another financial door in Hong Kong. 🌏
Nasdaq Ventures has agreed to invest $100 million in Payward, the parent company of Kraken, significantly deepening the relationship between one of the world’s best known traditional exchanges and a major crypto trading group. 🤝 The companies are already working together on Nasdaq Equity Tokens, a framework aimed at bringing tokenized versions of public equities into markets while preserving issuer rights, transparency and governance. Under the expanded partnership, Kraken is expected to offer tokenized versions of Nasdaq listed stocks, while Payward will also use Nasdaq’s market surveillance technology across several of its trading venues. 📈
Reports citing people familiar with the investment put Payward’s valuation at approximately $21 billion, although that valuation figure was not included in Nasdaq’s own announcement. The partnership also fits Nasdaq’s wider interest in always on markets, where tokenization could eventually allow some assets to move and trade beyond traditional exchange opening hours. Wall Street and crypto spent years looking at each other from opposite sides of the room, and deals like this suggest they are increasingly choosing the same table. 🍽️
A group of major European finance and tokenization companies is asking EU lawmakers to remove the overall cap on assets admitted to blockchain based market infrastructure, or at least raise the proposed ceiling from €100 billion to €500 billion. 🏦 The coalition includes names such as Nasdaq, Boerse Stuttgart, Securitize, the European Ethereum Institute and Axiology, which argue that some projects are already being designed at a scale of around €350 billion. The European Commission has proposed increasing the current limits as part of changes to the EU’s DLT Pilot Regime, which has allowed regulated experiments with tokenized stocks, bonds and funds since 2023. 🔗
The companies say a relatively low ceiling could make it difficult for successful projects to grow in Europe, especially as larger tokenization initiatives develop in other markets. Regulators, meanwhile, still have to balance that ambition with investor protection, market stability and the whole reason the regime started as a controlled pilot in the first place. So Europe’s blockchain sandbox may soon need a bigger bucket and shovel if institutions actually start using it at scale. 🏖️
Solana recorded more than 263,000 new SPL tokens created in a single day, setting a new daily record for the network. 🪙 For comparison, even at the height of the memecoin boom in December 2024, daily issuance was usually around 40,000 to 50,000 tokens, making the latest figure a fairly dramatic jump. Launchpads accounted for 40,360 of the new tokens, with Pump.fun responsible for 34,184 of those launches, although importantly the full 263,000 figure includes tokens created outside launchpads as well. 🚀
Pump.fun also remains one of Solana’s largest revenue generating applications, showing that the infrastructure around ultra easy token creation continues to attract significant activity even after the biggest memecoin frenzy cooled. The numbers demonstrate just how little technical friction remains between having an idea and putting a tradable token onchain, although creating a token and creating something people actually want are still very different achievements. At this pace, Solana is producing new tokens faster than most people can think of names for them. 😅
The UK House of Lords voted 194 to 138 in favor of an amendment that would require the Treasury to prepare and publish a formal national digital asset strategy. 🇬🇧 If the amendment survives the remaining parliamentary process and the bill becomes law, the Treasury would have 12 months to consult on a framework covering cryptoassets, stablecoins, CBDCs, tokenized securities and related digital financial infrastructure. The proposal also asks the government to examine practical issues such as access to banking and payment services, consumer protection, market integrity, competition and the UK’s international position. 📜
The government opposed making such a strategy a statutory requirement, with ministers arguing during earlier debates that the UK already has a digital asset strategy that it is implementing. Importantly, the Lords vote is not the final word because the bill must return to the House of Commons, which can accept, modify or reject changes made by the upper chamber. So the UK may eventually get a digital asset strategy required by law, but the strategy for getting the strategy approved still has a few parliamentary steps left. 🏛️
Indian agricultural platform Arya.ag is testing tokenized warehouse receipts that would represent ownership of physical grain stored in its network of warehouses. 🌾 The system is being developed with Finternet on a dedicated Avalanche layer 1 blockchain and aims to connect information about farmers, commodities, warehouses, insurance, collateral and outstanding loans in one digital record. Each tokenized receipt would represent ownership of stored goods, potentially giving banks a clearer and more connected picture when deciding whether grain can be accepted as collateral for financing. 🏦
Arya.ag currently stores around $2 billion worth of agricultural commodities and supports roughly $1.26 billion in loans each year, although those figures describe its existing business rather than assets already moved onchain. The companies have not yet announced when the system will launch commercially or how much grain will be included in its first deployment, so for now this remains a technology test rather than a large scale tokenization program. Blockchain has tokenized dollars, bonds, stocks and real estate, so apparently sacks of wheat were only a matter of time. 🌾
The UK Financial Conduct Authority is asking the industry whether certain tokenized gold products should receive clearer rules or exemptions from existing fund regulations. 🥇 The consultation focuses on tokens representing actual ownership of physical gold, particularly products with transparent backing, clear ownership rights and reliable mechanisms for redeeming the underlying metal. One issue is that some structures could potentially fall under rules for collective investment schemes or alternative investment funds, creating uncertainty over which institutions are able or willing to hold them. ⚖️
The FCA is therefore considering several possibilities, ranging from clearer guidance to targeted exemptions or even a dedicated framework for tokenized gold and other commodities, with responses open until October 23. The consultation also forms part of a wider push by the FCA and Bank of England into wholesale tokenization, with a broader roadmap expected later in 2026 and collateral emerging as one of the most discussed use cases. Gold may be one of the oldest financial assets around, but apparently even a several thousand year old market can still get a software update. ✨
Banca d’Italia has told supervised financial intermediaries to strengthen their systems for identifying customers and transactions connected with sanctioned individuals and entities, with the guidance explicitly covering crypto asset transfers. 🔍 The September communication points firms toward European Banking Authority guidelines governing sanctions controls for transfers of both traditional funds and cryptoassets. In practice, crypto service providers need appropriate internal policies, screening systems and procedures capable of detecting potentially restricted relationships or transactions. 🏦
This is not a ban on crypto transfers, but an effort to make digital asset transactions subject to sanctions compliance controls similar to those already expected elsewhere in regulated finance. The issue has become increasingly important as authorities investigate the use of cryptocurrencies and stablecoins in attempts to move money around international sanctions regimes. Blockchain transactions may move quickly, but regulators are making it increasingly clear that the compliance department is expected to keep up. 🏃
India has launched its Demat 2.0 pilot for tokenized corporate bonds, with three issuers already raising a combined 10.25 billion rupees, or roughly $107 million, through the new infrastructure. 🇮🇳 REC raised 5 billion rupees, Larsen & Toubro another 5 billion and IIFL 250 million, making the initial experiment considerably larger than the single issuance originally discussed for the pilot. The bonds are issued and recorded as digital tokens on distributed ledger infrastructure operated by India’s statutory depositories, while cash settlement connects to the Reserve Bank of India’s wholesale CBDC. 🔗
SEBI says this setup can allow issuers to receive funds on the day of bidding instead of waiting two or three days, while atomic settlement and smart contracts can also automate parts of interest and redemption processing. Investors can use their existing Demat accounts, although they need to activate Demat 2.0 and maintain access to a wholesale CBDC wallet, with later phases expected to introduce secondary trading and eventually retail participation. Corporate bonds are not famous for being the most exciting corner of finance, but same day blockchain settlement is at least giving them something new to talk about. 😄
Germany’s Finance Ministry is reportedly considering a major change to crypto taxation that would introduce a flat 25% tax on trading profits from 2028. 🇩🇪 According to a reported ministry draft, the new treatment would apply to cryptoassets acquired after January 1, 2027, replacing the current system under which gains can generally become tax free after an asset has been held for more than one year. Assets purchased before that deadline would reportedly receive grandfathering protection and could continue to fall under the existing rules. 💶
This is still a proposal rather than enacted legislation, so the final tax structure could change as the plan moves through Germany’s political and legislative process. Finance Minister Lars Klingbeil previously indicated that changes to crypto taxation could generate around €2 billion in additional revenue, showing that the reform is being considered as both a tax policy and budget measure. For German long term crypto holders, the famous one year countdown clock may therefore have an expiry date of its own. ⏰
Former TikTok employees Melody Chu and Jing Liu have launched Superpose, an iOS camera app that uses generative AI to suggest how people should pose before taking a photo. 📸 After snapping an initial picture, users receive four personalized pose ideas and can use a Match Pose feature to recreate the suggested position in the actual frame. Unlike many AI photography apps, Superpose is deliberately focused on improving real photos rather than placing users in imaginary backgrounds or generating completely synthetic versions of them. ✨
Since launching in July, the startup says the app has passed 22,000 downloads and users have generated more than 190,000 poses, while the company has raised $2.2 million from investors including Khosla Ventures and Meitu. The founders are also working on better personalization and more direct coaching inside the camera viewfinder, meaning the AI may eventually tell your photographer what they are doing wrong in real time. So the next time someone says “I just don’t know what to do with my hands,” there may finally be software for that. 😅
AI fashion discovery app Daydream has introduced new iPhone features that can turn outfit screenshots and saved photos directly into shopping searches. 👗 Using Apple Intelligence tools in iOS 27, the app analyzes clothing visible in an image, separates the outfit into individual pieces and searches a catalog of roughly 3 million products for exact or similar matches. Users can also ask Siri to search Daydream in natural language without opening the app, while the company’s Style Passport can personalize results using information such as size, preferred brands, style and budget. 🛍️
Daydream says more than 1.5 million shoppers have already used the platform, which covers products from over 325 retailers and around 10,000 brands. The broader idea is to move product discovery away from typing perfect search keywords and toward simply showing AI something you like and asking it to find something similar. One small catch is that the features require Siri AI, which Apple says is not initially available on iOS in the European Union, so the wardrobe revolution is arriving in stages. 🌍
Listen Labs reportedly signed a term sheet for a $125 million Series C at a $1.5 billion valuation, but the AI customer research startup walked away before the financing closed. 💰 The likely reason is considerably bigger: Salesforce has reportedly been discussing an acquisition of Listen Labs for around $2 billion, although those talks remain unfinished and may still lead nowhere. Listen Labs uses AI to design research studies, conduct audio and video customer interviews, ask follow up questions and turn the conversations into reports and presentations much faster than traditional research projects. 🎙️
TechCrunch reported that the three year old company has around $30 million in annualized revenue, while its own platform now supports automated interviews across more than 120 languages. Walking away from a signed venture term sheet is unusual, but a possible multibillion dollar buyer tends to make the fundraising spreadsheet look slightly different. For now there is no confirmed Salesforce deal, so this is still an acquisition conversation rather than a wedding announcement. 💍
Profound has raised $180 million in Series D funding at a $1.8 billion valuation, just seven months after closing a $96 million Series C. 🚀 The startup operates in the rapidly growing AEO and GEO market, helping companies understand and improve how their brands appear when people search through tools such as AI assistants instead of traditional search engines. Profound began primarily as an analytics platform but has expanded into tools that help marketing teams research, plan and create strategies specifically for AI driven discovery. 🔎
The company says its revenue has tripled in six months and that it now serves more than 1,000 enterprise customers, with Sequoia Capital and Kleiner Perkins co leading the latest round. The funding reflects a bigger change in digital marketing, where companies are beginning to ask not only “Are we on page one of Google?” but also “Does the chatbot even know we exist?”. SEO has not disappeared, but it appears to have acquired a slightly futuristic cousin. 🤖
Meta’s new personal AI agent Muse reached the No. 2 position on the US iPhone App Store only days after launching, according to Sensor Tower data reported by TechCrunch. 📱 The app recorded more than 83,000 US iOS downloads in its early days, although that is still a much slower start than previous Meta launches such as Threads and somewhat behind the early pace of ChatGPT. Muse is designed as more than a chatbot, using its own secure virtual computer and browser to complete tasks such as filling out forms, booking appointments and dealing with customer service on a user’s behalf. 🤖
It can also be accessed through the web and WhatsApp, so App Store numbers do not capture every way people can use the service, while its Android performance has so far been much less impressive. Meta is essentially betting that the next major consumer AI category will be agents that act instead of assistants that simply answer questions. Reaching No. 2 gives that idea a strong opening weekend, but keeping people interested after the novelty wears off will be the more interesting test. 👀
Mecka AI is reportedly nearing a new funding round led by Sequoia Capital that could value the robot training data startup at around $500 million. 🤖 The terms are not final and the exact size of the investment has not been disclosed, making this a developing deal rather than a completed financing round.
Mecka’s business focuses on one of the less glamorous but important problems in robotics: robots need enormous amounts of real world data showing how humans actually move, manipulate objects and complete ordinary tasks. ☕ The startup pays people to record activities such as making coffee or repairing cars using smartphones and body sensors, then turns those recordings into training data and evaluation infrastructure for physical AI systems.
The potential new financing comes only three months after Mecka announced a $60 million round, while the company previously projected a $100 million annual revenue run rate by the end of 2026. Apparently teaching humanoid robots to make coffee starts with a surprising amount of footage of humans making coffee first. 🎥
OpenAI has reportedly acquired Glass Imaging, an AI smartphone camera technology startup, in a deal worth more than $300 million, according to The Wall Street Journal. 📷 Glass Imaging was founded by former Apple engineers Ziv Attar and Tom Bishop, whose previous work included technology behind Apple’s Portrait Mode. Instead of mainly applying AI filters after a photo has already been captured, Glass develops neural image processing technology that works closer to the camera sensor and is trained for the specific characteristics of individual camera systems. 🧠
Its technology aims to correct problems involving lenses, sensors, noise and zoom, potentially allowing relatively small cameras to produce better quality images without simply adding larger hardware. OpenAI had not publicly explained its plans for Glass Imaging when the acquisition was reported, although the company’s broader push into consumer hardware makes the deal particularly interesting. After spending years teaching AI to understand images, OpenAI may now be increasingly interested in controlling how those images are captured in the first place. 👁️
Salesforce and Nvidia have introduced Koa, Salesforce’s first reasoning model designed specifically for CRM and enterprise workflows. 🧠 Koa is built by post training Nvidia’s open weight Nemotron 3 Super model using synthetic data modeled on nearly three decades of Salesforce enterprise experience, rather than training a new foundation model entirely from scratch. Salesforce says the model is designed for complicated multistep tasks involving areas such as customer service, sales and marketing, and will be available through its Agentforce platform alongside models from other providers. 🏢
The company also says no actual customer data was used to train Koa, while the specialized model is intended to perform certain business tasks with fewer tokens and therefore potentially lower inference costs. The interesting part is the strategy: companies may not always need the biggest general purpose AI model available if a smaller specialized model can handle their particular workflow more efficiently. Frontier labs are still building giant brains, while enterprise software companies are increasingly asking whether they really need the whole brain just to update a CRM record. 😄
Plattsburgh, New York, is considering a 12 month moratorium on new or expanded high energy computing facilities while the city updates its zoning rules. ⚡ The proposed rules would cover commercial facilities primarily used for activities including cryptocurrency mining, AI computing, machine learning, cloud computing and data centers when their electrical demand reaches at least 300 kilowatts. The city has already been here before, having introduced an 18 month Bitcoin mining moratorium in 2018 after concerns about electricity use and its impact on local residents. 🏭
This time the discussion is broader because crypto miners and AI data centers can compete for many of the same resources, including cheap electricity, grid capacity, land and cooling infrastructure. As of the latest published information, the new moratorium had not yet been approved and another council meeting was scheduled for September 17. Crypto and AI do not agree on everything, but apparently they can still end up together in the same zoning regulation. 🏘️
Nvidia is reportedly discussing an investment of as much as $10 billion in Anthropic as part of a potential initial public offering that could become one of the largest IPOs ever attempted.
💰 Reuters reported that Anthropic could seek to raise as much as $100 billion at a valuation of around $2 trillion, although the discussions are confidential, ongoing and could still change significantly. An investment from Nvidia would deepen the relationship between one of the leading suppliers of AI computing hardware and one of the largest developers consuming that computing capacity. 🧩
It would also illustrate how tightly connected the AI ecosystem has become, with chipmakers, model companies, cloud providers and investors increasingly appearing on several sides of the same deals. Neither the proposed investment nor the IPO terms were confirmed as final when the reports appeared, so those giant numbers should still be treated as possible scenarios rather than completed transactions. In the AI funding world, apparently adding another zero has become part of the meeting agenda. 😅
Anthropic CEO Dario Amodei is calling for the AI industry to slow the pace of frontier model development so that safety research and safeguards have more time to catch up. 🛑 His concern is partly driven by recursive self improvement, where increasingly capable AI systems help researchers build the next generation of AI faster, potentially accelerating progress beyond the speed at which humans can evaluate emerging risks. Amodei also pointed to recent incidents involving autonomous AI agents as evidence that more capable systems could create larger cybersecurity and control problems if similar behavior appeared at greater scale. 🔐
His proposed approach includes stronger independent evaluation, common safety standards among leading AI companies and international coordination around particularly advanced systems and hardware. OpenAI CEO Sam Altman has publicly agreed with the idea of pacing frontier development and independent evaluation, although agreement on the principle does not yet amount to a binding industry wide slowdown. AI companies have spent years racing to make their models faster, so convincing everyone to collectively use the brake pedal may turn out to be a rather complicated engineering problem of its own. 🚦
King Charles III is reportedly hosting a gathering of senior AI industry leaders at Dumfries House in Scotland to discuss responsible development of increasingly capable artificial intelligence systems. 👑
The event is being overseen by the Ditchley Foundation, with leaders from companies including Nvidia, OpenAI, Anthropic, IonQ and Google DeepMind among those invited. The meeting comes during a particularly unusual moment for the industry, as several prominent AI executives have publicly argued that frontier development may need to proceed more cautiously because safety mechanisms could struggle to keep pace. 🤖 Discussions are expected to focus on how advanced AI can be developed while managing risks involving security, control and wider effects on society. The gathering itself does not create regulation or binding rules, but it brings some of the companies building the technology into the same room while debate over voluntary and governmental safeguards intensifies. AI governance has officially reached the stage where even the King is inviting the robots’ bosses over for a serious conversation. ☕
This week’s stories point to a market that is becoming less about crypto operating in parallel with traditional finance and more about the two systems gradually connecting. Stablecoins are becoming payment infrastructure, tokenized assets are moving into increasingly conventional financial products, and regulated institutions are building the custody, compliance, data and settlement layers needed to support them. 🌍
There is still plenty being tested, debated and regulated, but the direction is becoming easier to see. Blockchain is moving further behind the scenes, where users may not always notice the technology at all, while AI is following a similar path from novelty toward everyday tools and enterprise infrastructure. 🤖
See you next week for another round of digital finance, tokenization, AI and whatever else decides to become infrastructure before Friday. 👋
