Meta Tests Self-Made AI Training Chips to Break Free from NVIDIA

Hamid Siddiqui / March 12, 2025 3:20am News

Meta is testing its new in-house AI training chips to reduce dependency on NVIDIA's technology. The company aims to lower infrastructure costs and improve efficiency by 2026. Their custom chips are designed for AI tasks and have completed their initial design phase. With a major focus on AI developments, Meta plans to reveal more at the upcoming Llamacon.

Merantix Capital Launches €103 Million Fund to Boost AI Startups Across Europe!

Merantix Capital has successfully raised a €103 million fund aimed at investing in early-stage AI startups throughout Europe. This new fund is over three...

Intel's Bold Move: One-to-One CPU to GPU Ratio for AI Revolution

Intel, alongside SambaNova and Foxconn, announced a groundbreaking partnership aimed at redefining AI workloads, shifting from a four-to-one GPU-CPU rati...

Post-Smartphone Revolution: A Glimpse into the Future of Tech

In 2026, the rise of post-smartphone devices is transforming tech interaction through spatial computing, smart glasses, and wearable displays. These inno...

5G-Advanced Set to Revolutionize Connectivity by 2026

5G-Advanced, or 5.5G, focuses on smarter network behavior rather than peak speeds. Set for rollout in 2026, it enhances responsiveness, lowers latency, a...

Google Unveils AI Edge Gallery for macOS: Local AI is Here!

Google has launched the AI Edge Gallery for macOS, allowing users to run AI models directly on Apple devices. This shift towards decentralized computing ...

Alphabet Shatters Records with $85 Billion Equity Offering!

Alphabet has successfully raised $85 billion in the largest equity offering ever, supported by a $10 billion investment from Berkshire Hathaway. The fund...

EU Unveils Ambitious Tech Sovereignty Package to Diminish US Dependence

The European Commission has launched its long-awaited technological sovereignty package, aiming to reduce reliance on American and Asian tech. The propos...

AI Saves Time, But Companies Let It Slip Away

A new study reveals that while 85% of employees save hours weekly using AI, nearly 40% of this time is lost to rework. The findings highlight the detrime...

AI Security Demands New Tactics as Threat Landscape Evolves

With companies rapidly integrating AI, traditional security methods are struggling to keep pace. New vulnerabilities like prompt injections and adversari...

Revolutionizing Material Science: Apoha Launches Liquid State Intelligence

Apoha, a London-based startup, has emerged from stealth with $36M funding to unveil its innovative data layer, Liquid State Intelligence, which measures ...

Megaport Shifts Gears: From Connectivity to AI Cloud Powerhouse!

Megaport, after a decade of connecting clouds, has pivoted to become an AI cloud player, securing four contracts worth A$458.9M. The Australian firm aims...

Microsoft Unveils Project Solara: A New AI Agent OS

At Build 2026, Microsoft introduced Project Solara, an innovative OS tailored for AI agent devices. This platform, built on Android, features two excitin...

Supermicro Unveils Game-Changing Dielectric Fluid for AI Cooling at Computex 2026

At Computex 2026, Supermicro introduced a groundbreaking coolant, SMC PG25-A, boasting 1,000 times higher electrical impedance than standard coolants. Th...

Nvidia's Vera Rubin AI Platform Enters Full Production, Boosts Korean Tech Stocks

At his GTC Taipei keynote, Jensen Huang announced Nvidia's Vera Rubin AI platform is now in full production, confirming Samsung, SK hynix, and Micron as ...

Intel Unveils Crescent Island GPU: A Game Changer for AI Inference!

At Computex 2026, Intel's CEO Lip-Bu Tan introduced the Crescent Island GPU, targeting the AI inference market. Built on the Xe3P architecture, it suppor...

Microsoft Launches WSL 3, Supercharging Windows for AI Development!

At Build 2026, Microsoft unveiled Windows Subsystem for Linux 3, offering near-native GPU and NPU access for AI developers on Windows. This new architect...

Microsoft Unveils Project Polaris, Transforms GitHub Copilot Experience!

At Build 2026 in San Francisco, Microsoft launched Project Polaris, its in-house AI model set to replace GPT-4 Turbo for GitHub Copilot users by August. ...

Arm Revolutionizes Market as ByteDance and Oracle Join AGI Chip Roster

At Computex, Arm CEO René Haas revealed that ByteDance and Oracle are now leveraging Arm's AGI data-centre CPU, joining Meta as users. This shift marks A...

GeForce Now Revolutionizes Linux Gaming on Low-Spec PCs

Cloud gaming is transforming Linux gaming, enabling users with low-spec PCs to enjoy high-end titles through services like GeForce Now. This platform all...

Anthropic Grants EU's ENISA Access to Groundbreaking Mythos AI Model!

Anthropic has granted the EU's cybersecurity agency, ENISA, access to the Mythos AI model, which has autonomously discovered over 10,000 zero-day vulnera...

Microsoft Set for Major AI Reveal at Build 2026!

Microsoft's Build conference is primed for exciting announcements, including a new Copilot super app and groundbreaking AI models. As trust in Windows an...