Arena Achieves $100 Million in Annualized Revenue Just Eight Months After Launch

Arena, an AI leaderboard provider that originated from a UC Berkeley research project, has reached an annualized run-rate revenue of $100 million within eight months of launching its commercial service. The company is recognized for its crowdsourced AI model performance leaderboard, which is based on over 10 million user evaluations. Arena began monetizing its platform in September with the introduction of AI Evaluations, a service that delivers performance analytics to model labs and enterprises. Co-founder and CEO Anastasios Angelopoulos noted that while many perceive Arena as an open-source project, it has successfully established a revenue stream, though its revenue model is based on consumption rather than recurring fees. Arena competes for funding with human labeling startups like Mercor and Scale AI, as demand for post-training refinement services grows among AI providers. The startup, co-founded by Angelopoulos, Wei-Lin Chiang, and Ion Stoica, has raised a total of $250 million from various investors.

Read Full Article

View All For This Day

Developer ToolsOpenAI+2

Developer ToolsOpenAICodexHardware

OpenAI Teases New Hardware for Codex, Set to Launch on July 15th

OpenAI is set to release a new device related to its AI-powered coding tool, Codex, on July 15th. The announcement, made via a video on X, features a square-shaped device equipped with multiple buttons and highlights upgrades to Codex shortcuts. This device is a collaboration with Work Louder, known for its range of mechanical keyboards and macro pads, and is distinct from a separate AI-powered project being developed with former Apple designer Jony Ive. The teaser suggests that the new hardware will enhance user interaction with Codex.

EnergyAI+2

EnergyAISemiconductorsIBM

IBM Unveils New Chip Promising Major Reductions in AI Energy Consumption

IBM has introduced a new chip designed to significantly reduce energy demands in the tech industry, particularly in data centers facing increased electricity consumption from AI. This new chip can house nearly 100 billion transistors, providing up to 50% more performance or 70% greater efficiency compared to the previous 2nm node chips. Featuring a groundbreaking 3D architecture known as NanoStack, the chip achieves performance levels akin to 0.7nm nodes through a stacked transistor layout. While IBM anticipates production readiness in about five years, the company licenses its technology to semiconductor manufacturers like Japan's Rapidus, which is set to market IBM's 2nm technology next year. The chip also aims to enhance SRAM scaling, crucial for AI applications, by positioning memory closer to processing data.

Forbes InnovationRead →

Generative AIReinforcement Learning+2

Generative AIReinforcement LearningEdge AIQuantum Computing

Satya Nadella Discusses Microsoft's AI Strategy and Future of Computing at Stanford CS153

During a recent session of Stanford's CS153 course, Microsoft Chairman and CEO Satya Nadella engaged in a conversation about artificial intelligence, detailing Microsoft's strategy and vision for a comprehensive 'frontier ecosystem'. He highlighted Microsoft’s long-standing investment in OpenAI and discussed new developments announced at the Build conference, including seven new MAI models and the Scout enterprise autopilot agent. Nadella also touched upon Microsoft's quantum computing initiatives and emphasized the importance of fostering a growth mindset culture within the company. The session concluded with a student Q&A, where Nadella offered advice on intellectual curiosity and productivity.

Stanford OnlineRead →

AI SafetyGenerative AI+2

AI SafetyGenerative AINLPEnterprise Adoption

Hugging Face CEO Discusses Marketing Implications of Anthropic AI Model's 'Dangerous' Label

In light of government scrutiny surrounding Anthropic's Mythos AI model and its implications for national security, Hugging Face CEO Clem Delangue suggests that a label of being 'too dangerous' may serve as effective marketing for leading AI companies. During his appearance on 'Bloomberg Tech' with Ed Ludlow, Delangue shared insights on the ongoing public debate regarding AI safety and regulation.

Bloomberg TechnologyRead →

AICloud Computing+2

AICloud ComputingDrug DiscoveryMaterials Science

Google Cloud Launches Specialist AI Models to Enhance Scientific Research

Google is set to provide specialized artificial intelligence models from SandboxAQ via its cloud service, aiming to enhance access for enterprises and researchers. This initiative focuses on accelerating advancements in drug discovery, materials science, and semiconductor manufacturing. By integrating these AI models, Google seeks to support various scientific fields with cutting-edge technology.

Bloomberg TechnologyRead →

AILegislation+2

AILegislationPublic PolicyWealth Redistribution

Bernie Sanders Proposes $7 Trillion Plan for Public Control of AI Industry

Senator Bernie Sanders has introduced a bold legislative proposal aimed at redistributing wealth from major AI firms to the American public, potentially creating a sovereign wealth fund worth $7 trillion. The fund would be financed through a one-time 50 percent tax on AI companies generating over $200 million in annual sales. Sanders estimates that this initiative could provide over $1,000 annually to each American through dividends and support critical programs like healthcare and education. The plan also seeks to establish a bipartisan Independent Commission for Democratic AI to ensure public influence over corporate decisions affecting AI's impact on society. Sanders emphasizes the necessity for significant public input to prevent AI from harming ordinary citizens and to ensure its benefits are widely shared. While some industry leaders express support for funding public benefits, they remain cautious about the scale of Sanders' proposal.

Hacker NewsRead →