💥🦾 Are Photonic GPUs the Future?

Happy Monday! This week’s deep dive explores what photonic GPUs are and why they matter, in light of Arago’s $26M funding round.

We also cover major headlines across AI, semiconductors, quantum, photonics, neuromorphic, and data centers, alongside curated readings on compute hardware, optical innovation, and sustainability. And of course, we’ve gathered all the relevant funding news.

In our bonus section, we break down key insights from last week’s Semiconductor Industry Association’s 2025 report on the state of the U.S. chip industry.

Deep Dive: Arago’s $26M bet—Are Photonic GPUs the Future?

Paris and Silicon Valley-based startup Arago recently raised $26M to bring its photonic AI chip to market, claiming it can match top-tier GPU performance while using 10× less energy.

In this context, photonic GPUs, which compute with light instead of electricity, are gaining attention as a potential breakthrough for energy-efficient, high-throughput inference.

Below, you can find what they are, why they matter, and how they compare to traditional silicon-based GPUs.

To note: Photonic Computing is also called Optical Computing. Light is made up of photons, so the terms can be used interchangeably.

What Photonic GPUs are

• Use light (photons) instead of electricity (electrons) to perform computations

• Process data through integrated photonic circuits using waveguides and modulators

• Optimized for massively parallel operations like matrix multiplication

• Typically designed as hybrid systems combining photonic compute with electronic control and memory

Why Photonic GPUs Matter

• Photonic chips offer much lower power consumption and generate minimal heat

• Enable faster data movement and higher throughput

• Particularly suited for large-scale inference in data center or edge environments

How Photonic GPUs Compare to Traditional GPUs

Key Advantages of Photonic GPUs

• Speed: Light moves faster than electricity, allowing higher data throughput

• Efficiency: Lower power consumption and minimal heat, ideal for energy-constrained environments

• AI inference: Well-suited for parallel tasks like matrix multiplication

Current Challenges of Photonic GPUs

• Flexibility: Not yet suitable for general-purpose workloads or training large models

• Maturity: Hardware, software, and developer tools are still early-stage

• Integration: Combining optical and electronic components adds design complexity

Arago joins other players like Lightmatter and Optalysys, working to commercialize photonic compute, signaling a shift from research to deployment across the space.

Spotlights

🦾 Nvidia's stock market value hits $4 trillion on AI dominance (Reuters)

“Nvidia, opens new tab briefly reached a market capitalization of $4 trillion on Wednesday, making it the first company in the world to reach the milestone and solidifying its position as one of Wall Street's most-favored stocks.”

Meta Superintelligence – Leadership Compute, Talent, and Data (SemiAnalysis) (12 mins)

“The real wake-up call came when Meta lost its lead in open-weight models to DeepSeek. That stirred the sleeping giant. Now in full Founder Mode, Mark Zuckerberg is personally leading Meta’s charge, identifying Meta’s two core shortcomings: Talent and Compute.”

We’ve read enough about Meta’s talent grabs, but this is a sharp summary from one of our favorite high-quality sources, also covering how Meta is backing the new team efforts with massive compute bets.

Headlines

This week’s headlines highlight AI funding, semiconductor growth, breakthroughs in photonics and neuromorphic computing, and big moves in data center infrastructure.

 🤖 AI

LangChain is about to become a unicorn, sources say (TechCrunch)

Mistral is reportedly in talks to raise $1B (TechCrunch)

Elon Musk’s xAI launches Grok 4 alongside a $300 monthly subscription (TechCrunch)

Amazon considers another multibillion-dollar investment in Anthropic, FT reports (Reuters)

Grok 4 benchmarks leak with 45% score on Humanity Last Exam (Testing Catalog)

🦾 Semiconductors

AI chip startup Groq discusses $6 billion valuation, The Information reports (Reuters)

Tenstorrent acquires Blue Cheetah Analog Design for advanced interconnect development (Data Center Dynamics)

Nvidia’s deal to buy Canadian AI startup CentML could top US$400M (The Logic)

GenAI’s Breakneck Pace Is Reshaping The Semiconductor Industry (Semiconductor Engineering)

Researchers develop eco-friendly method of creating semiconductor materials for electronics (Tech Xplore)

NJIT Engineers Reduce Power Consumption in Future Computer Memory (New Jersey Institute of Technology)

MIT Scientist Shrinks Active Memory Footprint by 50% (AInvest)

PwC: Climate change threatens one-third of global semiconductor supply (TechInformed)

⚛️ Quantum Computing

Chip-to-Chip Photonic Quantum Teleportation Over Optical Fibers of 12.3 Km (Newswise) Research Paper: here

⚡️ Photonic / Optical Computing

Scientists Shine Light on Multidimensional Photonic Computing (The Quantum Insider)

Photonic Devices Deliver Power and Precision to Next-Gen On-Chip Sensors (Photonics Spectra)

Hybrid Photonic Platform Enables Ultra-Sensitive Biosensing (AzoOptics)

🧠 Neuromorphic Computing

New memristor-based system by Chinese scientists boosts AI data sorting efficiency (Interesting Engineering)

💥 Data Centers

Nvidia challenger Groq expands with first European data center (CNBC)

Oracle emerges as key AI cloud partner in hyperscaler data center boom (DIGITIMES Asia)

CoreWeave to acquire Core Scientific in $9 billion all-stock deal (CNBC)

Readings

This week’s reading list spans semiconductor strategies, photonic innovation, and the environmental impact of data centers.

 🤖 AI

Why I don’t think AGI is right around the corner (Dwarkesh Podcast) (13 mins)

The American DeepSeek Project (Interconnects) (13 mins)

xAI gave us early access to Grok 4 - and the results are in (x - Artificial Analysis) (4 mins)

🦾 Semiconductors

The Data Dilemma In Semiconductor Testing And Why It Matters: Part 1 (Semiconductor Engineering) (4 mins)

The Data Dilemma In Semiconductor Testing And Why It Matters: Part 2 (Semiconductor Engineering) (3 mins)

Three Questions to Cheng Ting-Fang: Semiconductors as the Hardware of AI - The Case of Taiwan (Institut Montaigne) (7 mins)

Transforming the Semiconductor Industry With Advanced Analytics and Sustainability (Technology Networks) (5 mins)

⚛️ Quantum Computing

Deterministic Entanglement for Photonic Quantum Computing: Atom as Interface (The Quantum Insider) (8 mins)

Silicon photonics enable compact single-qubit machine learning classifier (Quantum Zeitgeist) (10 mins)

⚡️ Photonic / Optical Computing

Testing At The Speed Of Light: Enabling Scalable Optical Testing For Silicon Photonics And CPO (Semiconductor Engineering) (5 mins)

Photonic Chips Boost AI Image Recognition, Lower Energy Use (Quantum Zeitgeist) (7 mins)

💥 Data Centers

There aren’t enough AI chips to support data center projections, report says (Utility Dive) (4 mins)

Guiding Light: How Photonics is Revolutionising Data Centres (Sify) (5 mins)

Funding News

From energy-efficient photonic chips and intelligent AI retrieval to sovereign quantum hardware and data center cooling innovations, these startups are tackling key bottlenecks in performance, cost, and energy consumption.

Bonus: Current State of the U.S. Semiconductor Industry

Report: “State of the Industry Report 2025” (Semiconductor Industry Association, July 2025)

This bonus section highlights key insights from the Semiconductor Industry Association’s 2025 report, published last week. It covers the current state of the U.S. chip industry.

Global Market Growth: Global chip sales reached $630.5B in 2024, up 19.7%, with strong performance in logic and memory segments.

U.S. 2024 Growth: Semiconductor sales grew 19.6% year-over-year in 2024, driven by demand from AI, automotive, and industrial sectors.

Global Competition: China, Korea, Japan, Taiwan, and the EU offer more generous tax and grant packages for semiconductor R&D and production.

Market Share: The U.S. holds 50.4% of global semiconductor sales ($318B in 2024), maintaining global leadership across design, R&D, and manufacturing equipment.

AI Demand Surge: “Computer/AI” became the largest chip demand category in 2024 (34.9%), overtaking communications and consumer electronics.

R&D Investment: U.S. firms invested $62.7B in R&D in 2024, representing 17.7% of revenue, second only to the pharmaceutical sector.

Supply Chain Risk: The U.S. leads in design and equipment but relies on Asia for critical raw materials, highlighting supply chain vulnerabilities.

Manufacturing Resurgence: Over $500B in private investment since 2020 is expected to triple domestic chipmaking capacity by 2032.

Trade & Regulation: U.S. policy must secure global market access, manage export controls carefully, and align environmental regulation with industrial growth.

Love these insights? Forward this newsletter to a friend or two. They can subscribe here.