- Future of Computing Newsletter
- Posts
- 🤖🦾 Photonic Quantum Computing, Next-Gen EUV, and Happy New Year
🤖🦾 Photonic Quantum Computing, Next-Gen EUV, and Happy New Year
A Newsletter for Entrepreneurs, Investors, and Computing Geeks
Happy New Year, and welcome to the first Future of Computing newsletter for 2025!
We’re excited to kick off the year with a fresh update: the newsletter will now arrive on Monday mornings. This will give you an early start on all the key developments in computing from the past week and a glimpse of what’s ahead.
After the success of the Future of Computing Conference in Munich last December, we’re already gearing up for the next editions—London in March and Berlin in June. Stay tuned for more updates coming soon!
Spotlights
⚛️⚡️ QC82: Shaping the Future of Photonic Continuous-Variable Quantum Computing (Future of Computing)
Check out our latest interview with Hussain Zaidi, co-founder and CEO of QC82. The startup is developing integrated photonic chips for fault-tolerant quantum computing at room temperature. We talked about how they tackle various challenges, from building the architecture for a large-scale photonic quantum computer to engineering groundbreaking individual components like photon detectors that operate at room temperature and are commercial game changers in the short term.
This paper introduces a new architecture called a Large Concept Model, which operates on higher-level semantic representations called "concepts" instead of individual tokens, addressing a limitation of traditional LLMs that generate output at the token level.
🤖 Notes on the new Deepseek v3 (Composio Blog)
Chinese DeepSeek v3, a 607-billion parameter mixture-of-experts model, achieves high performance in reasoning and mathematical tasks while being cost-efficient, with training costs of only $6 million. It rivals proprietary models like GPT-4o and Claude 3.5 Sonnet, and its advanced architecture enables efficient memory usage and strong generalization capabilities across tasks.
🧠 How a neuromorphic chip could benefit industry (Techxplore)
Neuromorphic chips integrate processing and memory to enhance energy efficiency and computing power. Heidemarie Krüger and her Dresden-based startup, Techifab, develop memristor-based components that process and store data locally, reducing the need for energy-intensive data transfers. This technology holds promise for applications in autonomous vehicles and industrial plants, enabling complex, real-time decision-making with minimal energy consumption.
⚛️ IBM will release the largest-ever quantum computer in 2025 (NewScientist)
IBM plans to launch the world's largest quantum computer in 2025 by connecting multiple smaller processors into a modular system, aiming to surpass its current 1,121-qubit Condor chip and reclaim leadership in quantum computing. This approach addresses engineering challenges associated with scaling qubit numbers and is expected to significantly advance computational power in the field.
🦾 LLNL selected to lead next-gen extreme ultraviolet lithography research (EENews Europe)
Lawrence Livermore National Laboratory (LLNL) has been selected to lead a research partnership focused on advancing extreme ultraviolet (EUV) lithography technology. The project aims to enhance EUV source efficiency by approximately tenfold compared to current carbon dioxide laser systems, potentially enabling the production of smaller, more powerful, and energy-efficient semiconductor chips.
Headlines
🤖 OpenAI teases new reasoning model—but don’t expect to try it soon (The Verge)
🤖 Microsoft makes powerful Phi-4 model fully open-source on Hugging Face: You can download it here (VentureBeat)
🤖🦾 Nvidia finalizes acquisition of AI software firm Run:ai: Run:ai offers GPU orchestration software to organize groups of GPUs more efficiently (Tom’s Hardware)
🦾 MI300X vs H100 vs H200 Benchmark Part 1: Training – CUDA Moat Still Alive (Semianalysis)
🦾 M5 Pro chip could separate CPU and GPU in ‘server-grade’ chips: Apple will be taking advantage of TSMC’s very latest chip packaging process (9To5Mac)
🦾 ByteDance could spend $7bn to access Nvidia Blackwell chips outside China: The Information highlights the ways in which Chinese companies exploit loopholes in US sanctions (DCD)
🦾 TSMC installs equipment at the first 2-nanometer chip plant (TechInAsia)
🦾 This Year, RISC-V Laptops Really Arrive: DeepComputing PCs aim to elbow out Arm and x86 architectures (IEEE Spectrum)
🦾 Chinese RISC-V project teases 2025 debut of freely licensed advanced chip design: Third-gen Xiangshan may be close to the performance of Arm’s made-for-HPC Neoverse 2 (The Register)
🛰️ NASA to Test Solution for Radiation-Tolerant Computing in Space (NASA)
🧠 A New Approach to Neuromorphic Computing: Research demonstrated a MOSCap device using Hafnium diselenide that mimics neuron-like adaptive behavior and memory retention (AZO Optics)
🧠 Innatera Unveils Neuromorphic Processor for Ambient Intelligence at CES 2025 (BISinfotech)
⚛️ Quantum computing stocks take a hit as Nvidia CEO predicts long road ahead (Reuters)
⚛️ What Nvidia’s CEO Missed About Quantum Computing (Forbes)
⚛️ Eurofiber, Quantum Bridge & Juniper Networks Unite for Quantum-Safe Encryption: “to integrate advanced post-quantum cryptography into its extensive network infrastructure” (TheFastMode)
⚛️ Development of a new European quantum technology begins: The EQUSPACE consortium has received 3.2 million euros from the European Innovation Council’s Pathfinder Open funding program (Lightsources)
⚛️ First Multiuse Optical Quantum Computer Comes to Japan: Riken's new machine has computing power equivalent to 1,000 qubits (Spectrum IEEE)
Funding News
🌐 $2M Pre-Seed – Simulation Theory: reducing waste by optimizing compute resources (Finsmes)
⚡️ $5M Seed – Opticore: optical GPUs for lower energy AI (Optics)
🦾 €10M Series B – Wooptix: semiconductor metrology equipment incorporating wavefront phase imaging (Wooptix)
⚛️ $100M Round in progress – Quantum Machines: control technologies for quantum computers (CTech)
⚛️ $100M Private Placement – Quantum Computing Inc.: integrated photonics and quantum optics (Quantum Computing Report)
🤖 $2B Round in progress – Anthropic: AI safety and research (Reuters)
Special: 2024 Reviews & 2025 Predictions
🤖 2024 in AI predictions (LessWrong)
🤖 Things we learned about LLMs in 2024 (Simon Willison)
⚛️ 10 quantum computing milestones of 2024 (NetworkWorld)
⚛️ 2024: The Year of Quantum Computing Roadmaps (Quantum Computing Report)
⚛️ TQI’s 2025 Predictions For The Quantum Industry (Quantum Insider)
🦾 2024 in Review: Readers’ Top 10 Articles (EETimes Europe)
🦾 The Year in Computer Science (Quanta Magazine)
🦾 Data Center Trends to Watch in 2025 (DataCenter Knowledge)
Deep Dive: Nvidia’s Project DIGITS
At CES, Nvidia unveiled Project DIGITS, a $3,000 AI-focused desktop computer powered by its latest Blackwell AI chip and a new ARM-based CPU co-developed with MediaTek, marking MediaTek’s entry into the desktop CPU market. Set to launch in May, DIGITS can handle AI models with up to 200 billion parameters—two systems can even be linked to support models with up to 405 billion parameters—and it runs on Nvidia's DGX OS, a customized Linux distribution based on Ubuntu 22.04 (see coverage by The Verge and ZDNET).
Nvidia building a consumer product like this seems unusual—after all, the production numbers will likely remain manageable, and it wouldn’t significantly impact the company’s bottom line even if demand from hyperscalers slows down. But this move by Nvidia will get even more developers to familiarize themselves with Nvidia’s AI ecosystem and tooling. As Jensen Huang put it: "Placing an AI supercomputer on the desks of every data scientist, AI researcher, and student empowers them to engage and shape the age of AI.”
💡 Love these insights? Forward this newsletter to a friend or two who can subscribe here