Liquid AI announced “Hyena Edge,” a new convolution-based, multi-hybrid model designed for smartphones and other edge devices. Hyena Edge is engineered to outperform strong Transformer baselines on both computational efficiency and language model quality. In real-world tests on a Samsung Galaxy S24 Ultra smartphone, Hyena Edge achieved up to 30% faster prefill and decode latencies compared to its Transformer++ counterpart, with speed advantages increasing at longer sequence lengths. Unlike most small models designed for mobile deployment, Hyena Edge steps away from traditional attention-heavy designs. Instead, it strategically replaces two-thirds of grouped-query attention (GQA) operators with gated convolutions from the Hyena-Y family. The new architecture is the result of Liquid AI’s Synthesis of Tailored Architectures (STAR) framework, which uses evolutionary algorithms to automatically design model backbones. STAR explores a wide range of operator compositions, rooted in the mathematical theory of linear input-varying systems, to optimize for multiple hardware-specific objectives like latency, memory usage, and quality. With mobile devices increasingly expected to run sophisticated AI workloads natively, models like Hyena Edge could set a new baseline for what edge-optimized AI can achieve.
NeuroBlade’s Analytics Accelerator is a purpose-built hardware designed to handle modern database workloads delivering 4x faster performance than leading vectorized CPU implementations
As Elad Sity, CEO and cofounder of NeuroBlade, noted, “while the industry has long relied on CPUs for data preparation, they’ve become a bottleneck — consuming well over 30 percent of the AI pipeline.” NeuroBlade, the Israeli semiconductor startup Sity cofounded, believes the answer lies in a new category of hardware specifically designed to accelerate data analytics. Their Analytics Accelerator isn’t just a faster CPU — it’s fundamentally different architecture purpose-built to handle modern database workloads. NeuroBlade’s Accelerator unlocks the full potential of data analytics platforms by dramatically boosting performance and reducing query times. By offloading operations from the CPU to purpose-built hardware — a process known as pushdown—it increases the compute power of each server, enabling faster processing of large datasets with smaller clusters compared to CPU-only deployments. Purpose-built hardware that boosts each server’s compute power for analytics reduces the need for massive clusters and helps avoid bottlenecks like network overhead, power constraints, and operation complexity. In TPC-H benchmarks — a standard for evaluating decision support systems — Sity noted that the NeuroBlade Accelerator delivers about 4x faster performance than leading vectorized CPU implementations such as Presto-Velox. NeuroBlade’s pitch is that by offloading analytics from CPUs and handing them to dedicated silicon, enterprises can achieve better performance with a fraction of the infrastructure — lowering costs, energy draw and complexity in one move.
FurGPT’s Web3 virtual pet platform each pet’s behavior securely on-chain creating a unique personality and memory profile, while providing verifiable continuity for users
FurGPT, a leading player in Web3 virtual pet innovation, has introduced emotion-driven mechanics to its AI-powered pets. The new features allow users to engage with their pets in more lifelike and emotionally responsive ways, enhancing user experience and deepening bonds. The AI pets can interpret and respond to user behavior in a nuanced and adaptive manner, allowing them to develop emotional states like curiosity, joy, or loyalty. The platform also integrates blockchain technology to securely record each pet’s behavioral data, ensuring a unique personality and memory profile that evolves over time. The emotional AI system also introduces mood-based quests and emotion-driven animations, making user interactions more intuitive and emotionally rich. This fusion of tech and sentiment in decentralized play represents a shift in the Web3 gaming space towards experiences that prioritize personalization, emotional depth, and interactive narrative.
Deloitte study shows tokenized ownership of loans and securitizations to grow to US$2.39 trillion by 2035 while tokenized private real estate funds expected to grow to US$1 trillion by 2035
Deloitte Center for Financial Services predicts that real estate tokenization could become a core pillar of property financing, ownership, and trading. The market of tokenized real estate could reach $4 trillion by 2035, growing at a 27% compound annual rate from under $300 billion. This sector combines crypto tech and traditional finance, creating digital versions of assets like bonds, funds, and real estate on blockchain rails. It offers operational efficiencies, cheaper settlements, and broader investor access. The report outlines three-pronged evolution of tokenized property: private real estate funds, securitized loan ownership, and under-construction or undeveloped land projects. Tokenized debt securities are expected to dominate, reaching $2.39 trillion in value by 2035. Challenges remain, including regulation, asset custody, cybersecurity, and default scenarios.
Mastercard is partnering crypto exchanges and fintechs to unveil end-to-end acceptance and payments capabilities of stablecoins such as wallet enablement, card issuance and digital commerce
Mastercard is doubling down on stablecoins, unveiling new global end-to-end acceptance and payments capabilities with an integrated 360-degree approach to allow consumers and businesses to use stablecoins as easily as the money in their bank accounts. The company has partnered with a host of crypto natives such as MetaMask, Kraken, Gemini, Bybit, Crypto.com and Binance on wallet enablement and card issuance and acceptance. Now it is working with OKX to launch the OKX Card, providing millions with easy access to their funds. Mastercard is also teaming up with Nuvei and Circle to give merchants the option to receive their payments in stablecoins such as USDC, regardless of how a consumer chooses to pay. It already offers this functionality across Paxos-issued stablecoins. “When it comes to blockchain and digital assets, the benefits for mainstream use cases are clear,” says Jorn Lambert, chief product officer, Mastercard. “To realise its potential, we need to make it as easy for merchants to receive stablecoin payments and for consumers to use them. We believe in the potential of stablecoins to streamline payments and commerce across the value chain.
Scientists develop OS that allows quantum computers to connect with each other, paving the way for a quantum internet
Scientists have developed the world’s first operating system for quantum computers, QNodeOS. This system allows quantum computers to connect with each other, paving the way for a quantum internet. QNodeOS operates by combining a classical network processing unit (CNPU) with a quantum network processing unit (QNPU), which controls the quantum code. The QNodeOS connects to a separate quantum device called the QDevice, which is responsible for executing quantum operations. The QDriver is a key component of QNodeOS, enabling it to control different types of quantum computers. The QNodeOS was demonstrated by connecting different quantum computers together and running a test program. Further experimentation is required, including using more quantum computers of different types and increasing the distance between them. The architecture could be improved by having the CNPU and QNPU on a single system board to avoid millisecond delays in communication. A quantum computer operating system represents a major step forward in their development, with potential applications for distributed quantum computing and potentially laying the foundations for a quantum internet.
Sprive’s tech automates using cash-back offers, discount vouchers and spare cash from everyday shopping towards mortgage overpayments
Fintech startup Sprive is a “mortgage overpayment” platform aimed at the U.K. market that helps users repay their mortgages faster using automation and cash-back rewards. It claims to save users an average of £10,000 each over the lifetime of the mortgage. Sprive said the app works by letting homeowners pay off their mortgage faster through their everyday shopping, by automatically putting spare cash toward overpayments and continuously scanning the market to help customers find better mortgage deals. Users can pay toward their mortgage from a bank account linked to the Sprive app by shopping as they would normally with mainstream U.K. supermarkets. They then use cash-back offers, discount vouchers, and more in order to pay off a mortgage faster. It does this by allowing the user to save money on the interest on the mortgage and shave potentially years off the mortgage term. CEO Jinesh Vohra said Sprive gives lenders a more efficient digital means of being able to acquire new customers. “Every time someone shops or switches mortgages, we make money. Within 15 minutes of shopping, you get money towards your mortgage, and we scan the market every day for better mortgage deals. Every time a customer refinances, we get commission from the mortgage lender.”
25% of Americans are now using buy-now, pay-later for groceries compared to 14% in 2024: Lending Tree survey says
Consumers are feeling the squeeze from economic uncertainty and are turning more to financing essential purchases like groceries. An April survey from Lending Tree shows an increase in Americans using buy-now, pay-later services for groceries—25% this year compared to 14% in 2024. Meanwhile, 41% of respondents said they’ve paid back their loan late over the past year, though the vast majority have paid back the money in a week or less. Men, young people, and higher-income borrowers were more likely to make late payments. The increase in popularity of BNPL has coincided with growing fear of a recession as a result of uncertainty around tariffs and interest rates. Consumer sentiment has continued to weaken, with the index falling to 52.2 last week from 57 a month prior. “It’s pretty clear that as people struggle with inflation and other kinds of economic uncertainty, people are looking to things like BNPL loans to help them extend their budget,” Matt Schulz, Lending Tree chief consumer finance analyst, told. “When buy-now, pay-later started, it was typically about designer handbags and appliances and things like that,” Schulz said. “But now people are looking at it for things like groceries and food delivery.” The ubiquity of BNPL may indicate a growing comfort with financial risks, especially among younger spenders. The rise of BNPL opportunities signals that consumers interpret the service as less risky, despite it not helping shoppers build credit that could help them down the line financially. With the current economic backdrop, expect these loans to remain popular, Schulz predicted.
AR app facilitates ageing-in-place, uses LiDAR scanners to identify potential accessibility issues through a step-by-step assessment and digital overlays
The University of Wisconsin-Madison is developing a mobile application called the Augmented Reality Home Assessment Tool (ARHAT) that uses AR and smartphone features like LiDAR scanners to help people assess their homes and identify potential accessibility issues as they age. The app “takes users through a step-by-step process that measures relevant parts of a living space and offers suggestions to make it more accessible. While using it, a user is prompted to select one of 14 limitations they may have — including visual impairment, mobility challenges or balance issues — and “then select the areas of the home to evaluate. Follow-up questions and prompts for measurements are tailored to the user’s answers, with the option to add notes and take photos. The app then employs a LiDAR scanner to measure a home’s elements and to visualize recommended changes or renovations. By superimposing visuals in the space as seen through the device’s camera, the tool will instantly let you know if something in the home is an accessibility barrier. The assessment then creates a detailed report listing identified barriers and potential issues with Americans with Disabilities Act (ADA) compliance, and makes recommendations on how to best address them. The researchers believe ARHAT can be a faster and more accurate alternative to traditional manual home assessments, potentially saving time and costs for healthcare agencies and systems.
Google Wallet may let you add a card just by NFC tapping it to your Android phone, instead of scanning or manually entering card information
Google appears to be working on a new feature that would allow users to add a credit card to Google Wallet simply by tapping the card to their phone. The current process of adding a card to Google Wallet typically involves either scanning or manually entering card information, then going through some sort of verification with your bank. It’s not a difficult process, but it’s one that Google seemingly wants to simplify by letting you add a card by tapping it to your phone. Code added to the latest Google Play Services update directly explains a process in which you “hold your card” to the back of your phone, using NFC to add it to Wallet. This new code was apparently found alongside the existing “Add payment method” screen, taking out any doubt that it’s for anything but adding a new payment card to the app. Presumably, users will still need to verify with their banking institution and complete other steps, but this would provide a quick and easy way to end the card information. It’s also a fairly familiar process as some banking apps, the Capital One app as one example, use NFC to interact with physical cards.