Rice University Breakthrough: New AI Method Accelerates Protein Engineering
Scientists at Rice University have developed a novel method to generate large datasets for training artificial intelligence models in protein engineering. A significant bottleneck in AI-guided protein engineering has been the lack of sufficient experimental data to train accurate models. The new technique utilizes an activity-based barcoding system that records the activity of individual protein variants, creating the necessary large-scale data for machine learning. As a proof of concept, the team applied this method to a small CRISPR-Cas protein, and the resulting AI model was able to predict mutations that significantly enhanced the protein’s activity. This breakthrough could dramatically accelerate the development of optimized proteins for various applications.
Sources:
Kia Commits $33 Billion to EV, Autonomous Driving, and Robotics Future
Kia has announced a substantial investment of 49 trillion won ($33 billion) through 2030 to advance its position in the future mobility sectors of electric vehicles, autonomous driving, and robotics. The automaker revealed these plans during its 2026 CEO Investor Day in Seoul, outlining a strategy to deploy the humanoid robot Atlas in its U.S. factories and to complete the development of autonomous driving vehicles by 2027. A significant portion of the investment, 21 trillion won, is earmarked for future technologies. In collaboration with Nvidia, Kia aims to deliver vehicles with Level 2+ highway autonomous driving by 2027, with plans to expand to Level 2++ for city driving by early 2029. The company also intends to utilize its purpose-built vehicles combined with robots like Stretch and Spot to enter the last-mile delivery market.
Sources:
Uber and VW’s MOIA to Launch Autonomous ID. Buzz Ride-Hailing in Los Angeles
MOIA America, a Volkswagen Group company, is partnering with Uber to deploy a fleet of autonomous ID. Buzz vehicles on the Uber ride-hailing platform in Los Angeles, with rides expected to be available by the end of 2026. The companies have begun on-road validation testing of the purpose-built autonomous vehicles in the city. MOIA America plans to scale its test fleet to over 100 autonomous ID. Buzz vehicles, which will have human operators on board to supervise the technology’s validation. This collaboration marks the next phase of a strategic partnership, combining MOIA’s autonomous mobility solution with Uber’s extensive network to create a scalable path for driverless ride-hailing services. Formerly known as Volkswagen ADMT, the company was renamed MOIA America in early 2026.
Sources:
Generative AI Diverges: Anthropic Restricts Claude Mythos as Open-Source Models Rise
In early April 2026, the generative AI landscape saw a significant strategic divergence in release strategies. Anthropic confirmed the existence of its most powerful model, Claude Mythos, but announced it will not be publicly available. Instead, it will be offered to a limited group of 50 organizations for defensive security research. In stark contrast, Zhipu AI released GLM-5.1, a 744-billion-parameter model, under the permissive MIT open-source license. This model has reportedly outperformed both Claude Opus 4.6 and GPT-5.4 on a software engineering benchmark. Additionally, Google contributed to the open-source movement by releasing its Gemma 4 family of models under the Apache 2.0 license. This trend highlights a growing split in the industry between companies keeping their most advanced models proprietary and those embracing open-source distribution.
Sources:
Meta Launches Muse Spark: A Powerful New Multimodal AI from its Superintelligence Lab
Meta has introduced Muse Spark, its first proprietary AI model developed by the Meta Superintelligence Labs division formed in the summer of 2025. This new large language model (LLM) is described as the most powerful Meta has released and is the first in a new ‘Muse’ family of models. Muse Spark boasts advanced multimodal reasoning capabilities, tool use, visual chain of thought, and multi-agent orchestration. The model is being positioned as a foundation for ‘personal superintelligence’ and is currently available through the Meta AI app and website, with a private API preview for select users. Meta has acknowledged that while the model shows strong performance in multimodal reasoning, it has performance gaps in areas like coding and complex multi-step tasks that are currently being addressed.
Sources:
Meta Signs $21 Billion Deal with CoreWeave for Next-Gen AI Infrastructure
Meta has committed to a landmark $21 billion agreement with AI cloud provider CoreWeave for dedicated AI infrastructure capacity, extending through December 2032. This deal dramatically expands upon a previous $14.2 billion contract between the two companies and is designed to support Meta’s escalating AI development and deployment needs. As part of the agreement, CoreWeave will establish dedicated capacity across multiple locations. Notably, CoreWeave will be among the first to deploy NVIDIA’s upcoming Vera Rubin platform, providing Meta with early access to next-generation AI chip technology. This move is part of Meta’s broader strategy to rapidly scale its AI inference capacity, complementing its own extensive data center investments.
Sources:
Microsoft Boosts Singapore’s AI Readiness with Expanded Cloud Services
Microsoft today announced the general availability of Microsoft Fabric Go Local in Singapore, a strategic move to support the data residency requirements of government and regulated industry workloads. The company also launched Windows 365 Link devices in the country, enabling secure connections to Cloud PCs from any location. These announcements, made at the Microsoft Singapore Public Sector Solutions Day, underscore the company’s commitment to advancing national AI readiness. This news follows Microsoft’s recent commitment to invest $5.5 billion in its Singapore-based cloud and AI infrastructure and operations between 2025 and the end of 2029. Microsoft Fabric provides a unified analytics platform, integrating data engineering, data warehousing, and business intelligence into a single SaaS experience.
Sources:
Kubernetes v1.36 Arrives in April with Enhanced Device and Volume Management
The Kubernetes project is preparing for the release of version 1.36, scheduled for the end of April 2026. This upcoming release will introduce key enhancements, alongside planned API removals and deprecations in line with the project’s documented deprecation policy. One notable feature graduating to beta is device taints and tolerations, which offers improved scheduling control for specialized hardware resources. Additionally, faster SELinux relabeling for volumes will become generally available, a critical change that reduces Pod startup delays on SELinux-enforcing systems. The release cycle for v1.36 began on January 12, 2026, with the documentation freeze set for April 8, 2026, ahead of the planned release on April 22, 2026.
Sources:
Infleqtion Projects $40M Revenue in 2026 Amid Growing Demand for Quantum Technology
Global quantum technology company Infleqtion has announced a strong revenue forecast of $40 million for 2026. The company attributes this positive guidance to surging customer demand for its quantum computing and sensing solutions across the defense, telecommunications, and critical infrastructure sectors. This financial outlook signals an increasing commercial adoption of real-world quantum applications. The announcement follows a pivotal 2025 for the company, which included the delivery of the United Kingdom’s only 100-qubit quantum computer in operation. Infleqtion also reported a narrowing of its operating losses in 2025 compared to the previous year.
Sources: