Palo Alto – January 20, 2025 – If you search the Google Map of Palo Alto using the term “Magical Playground,” you will come across a children’s park filled with innovative play structures and a natural history museum. Kids love it, and they usually learn something, too.

A more adult-oriented magical playground for the tech community is less than two miles away. Hidden inside an unassuming warehouse is the headquarters of Playground Global, one of Silicon Valley’s most active and promising deep-tech investors. Here the most important investment Due Diligence question is: Does this idea comply with the laws of Physics? If the answer is yes or no, Playground will probably not invest. If the answer is “Possibly and it could change the world,” – then it’s game on!

I was lucky enough to attend a fireside chat event at Playground HQ. The speakers were Peter Barrett, General Partner at Playground; Rene Haas, CEO of ARM; Young Sohn, Founder of Walden Catalyst; and Sasha Ostojic, Venture Partner.

Here are my three key takeaways:

1) Now is a great time to be investing in chips
AI is here, but computing is not nearly good enough for it today. AI software is in danger of consuming the world and all its energy resources. Better hardware is needed to prevent this. This creates huge demand and opportunity for innovation and chip start-ups. For example, AI training will be in the cloud, but inference will be everywhere. Innovation is needed to make this happen.
NextSilicon and d-Matrix are great examples of game-changing chip companies from Playground’s portfolio. NextSilicon has developed a uniquely adaptable architecture that is dynamically optimized at run-time in software. Each workload’s hardware configuration is customized for speed and power efficiency.
Meanwhile, d-Matrix has solved the von Neumann memory bottleneck by integrating memory with compute. In doing so, they have created a blazingly fast inference server that is predicted to be 3x more energy efficient & 3x more cost-effective than today’s best-in-class AI Processor based servers.

2) Quantum Computing will begin performing magic in 2025
And not because the United Nations has designated 2025 as the ‘International Year of Quantum Science.’ But we might just demonstrate Quantum Supremacy. This is when a quantum computer solves a useful problem that a classical computer could not solve in a practical time frame. Note that I deliberately highlighted the word “useful,” recognizing that Google’s Quantum chip, Willow, did manage to solve a non-useful problem in late 2024.
To achieve useful Quantum Supremacy, significant improvements in both quantum software algorithms and quantum computing hardware are needed. Another Playground portfolio company, Phasecraft, is at the forefront of reducing the complexity of quantum algorithms for real-world problems, for example, so that designing next-generation solar cells can be modeled with one millionth of the cost compared to previous algorithms. In parallel, a second Playground portfolio company, PsiQuantum, is building the world’s first useful, tolerant, and fully error-corrected Quantum computer. In April 2024, PsiQuantum announced it would build the first utility-scale quantum computer in Brisbane, Australia.

3) Photonics & Light Innovation Increasingly in Focus
Innovation in deep-tech photonics & light was cited as the discipline most likely to enable step-changes in performance in semi-conductor manufacturing, AI & quantum compute performance. XLight is developing a more powerful light source, piped around a fab as a utility and useable by any of the hundreds of unique process steps in a large high-performance compute fab. xLight’s laser will be 4x brighter, use 5 to 10 times less power, and more importantly, enable the next generation of semiconductor manufacturing equipment.
As Processors in AI data centers become increasingly fast, network interfaces within the data center, including AI Processor-to-AI Processor communication, must move from copper to optical. Incremental advances in pluggable interconnect technology are helping a lot, but new architectures enabled by silicon photonics for native, low-latency, high-speed communication are emerging. Ayar Labs, another Playground portfolio, has strong traction here, with an impressive list of strategic investors and collaboration partners.

Conclusion – How dead is Moore’s law?

The fireside chat opened with the question: How dead is Moore’s law? The phrasing of the question hinted at the answer. In the world of traditional computing, there are only two possible states: dead or alive. In the new quantum world, many more states are possible. Moore’s law is neither dead nor alive but in some form of Cinderella-like stasis waiting to be awoken by the magic of innovation. Some of the magic will, without a doubt, be supplied by Playground!