Dean Takahashi

Despite its beautiful Ori games, Moon Studios is called an ‘oppressive’ place to work #Ori @GamesBeat #MoonStudios venturebeat.com/2022/03/18/des… via @VentureBeat

Sad for the game economy. gamesbeat.com/sony-and-nex-i…

Sony and Nex increase PS5, Playground prices - gamesbeat.com/sony-and-nex-i…

We're podcasting Boss Rush on the news in games this week in a couple of minutes. linkedin.com/events/7442676…

Honor of Kings and Arena of Valor will show up at Esports World Cup 2026. gamesbeat.com/honor-of-kings…

Haha awesome

Congratulations. Reggie Fils-Aimé (founder, Brentwood Growth Partners; former president & COO, Nintendo of America) will be among those inducted into the Consumer Technology Hall of Fame on April 21.

Sandbox VR is making location-based VR work. gamesbeat.com/sandbox-vr-hit…

In a metaverse move, Shield AI acquires Aechelon and raises $1.5B. gamesbeat.com/shield-ai-acqu…

Colossal Biosciences spins out Astromech at $2B valuation. gamesbeat.com/colossal-biosc…

Oof. But not a surprise. Blockchain is having a tough time in games. gamesbeat.com/blockchain-gam…

Agentic AI startups dominate the roster for Berkeley Skydeck's Batch 21 roster for April 2. gamesbeat.com/berkeley-skyde…

Community. Connection. The leaders who've built careers around these things are already inside GB MAX. Join the ones building the network that moves this industry forward. GB MAX Premium includes a GamesBeat Summit VIP ticket. Prices go up April 1. go.gamesbeat.com/gb-max/

What a lucky group of young innovators, to sit down with Ted Hoff! Our nation’s young best and brightest. Love the #RegeneronSTS so much.

Thank you @HowardLutnick for joining the Semiconductor Industry Association Board @SIAAmerica for an important discussion on strengthening America’s leadership in semiconductors and expanding our domestic manufacturing footprint. We deeply appreciate the open dialogue and partnership with @CommerceGov as we work together to accelerate America’s tech stack and expand opportunities across the U.S. workforce.

Mike DeLaet — President, Ares Interactive — joins GamesBeat Summit 2026.   May 18-19 | Los Angeles | Register → events.gamesbeat.com/summit26/   #GamesBeatSummit2026 #GamesBeat

Blockchain gaming's market value plunged 67% in 2025 | Chainplay - gamesbeat.com/blockchain-gam…

And taking names?

@GenePark great to hear, Gene.

got good health news today. post op is looking great, and i plan to stick around for a while. it’s thanks to everyone’s support that i can do this. i can’t even begin to express my gratitude. here’s a vending machine for what i hoped was generic snacks but its just scrubs

I met with an interesting company at GTC, @velaura_ai. With their Titan Core Platform, they say they can reduce power by 2-4x on MATMUL and other arithmetic operations (which consume 40-70% of total XPU power in the DC). Think about what that could mean for XPUs. And they're working with hyperscalers, said it in the press release. The last three years in AI silicon have been all about chasing higher TFLOPS. That era is ending. The AI datacenter workflow is disaggregating. Power draw is limiting the AI future. Think Vera CPX for pre-fill, Groq LPU, low-power AMD GPU with Meta, TPU light, and other inference-only accelerators. Over the next three years, the real test for AI chip developers will be how much they can reduce power consumption per TFLOP. Efficiency, not just raw performance, is now the focus. The company recently recast itself as Velaura AI, an IP platform company with $275M in VC funding, which you might remember as Auradine, which cranked out crypto accelerators. TSMC has already shipped 26M+ of its 3nm ASICs. This isn’t just a theoretical concept; it’s a real-world product that has been mass-produced. Low-voltage design techniques used in Bitcoin ASICs that use SHA hash functions should be applied more or less directly to an AI matrix engine: the mathematical operations performed on the matrices are largely unchanged between encryption and deep learning. Velaura takes a customer's MATMUL RTL (real-time logic), applies proprietary low-voltage libraries and EDA flows, and delivers optimized GDS. No software changes, no SoC redesign. The DIY alternative is 18-24 months with yield risk. Based on who's on the board, investors, and ASIC chip experience, I believe the company has proven itself to a few hyperscalers. The release mentioned two. I'm looking forward to more details.