Harris Cohen

4-decade entrepreneur, former CEO turned AI & emerging tech advisor. Pioneered private-label USB & Bluetooth; championing innovation & transformation

[Harris Cohen @HCohenTech]

@SERobinsonJr So sad

[Harris Cohen @HCohenTech]

https://t.co/McBG3xJn8M

[Patently Apple @PatentlyApple]

https://t.co/SwLX9Fztgw

[Patently Apple @PatentlyApple]

https://t.co/SwLX9Fztgw

[Harris Cohen @HCohenTech]

@TickerSymbolYOU Yes, indeed

[Harris Cohen @HCohenTech]

I'm going! Are you? Sanford Stem Cell Institute Seminar Series with David T. Teachey, MD (CHOP) https://t.co/ypTBrWbaTN #null via @localist

[Harris Cohen @HCohenTech]

Let’s go back in time with Mr. Wizard.

[Harris Cohen @HCohenTech]

@elonmusk Perfect

[Harris Cohen @HCohenTech]

@moondrencht “Wilson”

[Harris Cohen @HCohenTech]

@sama Such a classic bedtime lullaby

[Harris Cohen @HCohenTech]

During today’s Starship flight, the internal payload bay camera showed a storm of white particles floating around. I’m guessing these were ice crystals formed from residual cryogenic propellant (LOX or methane)? @elonmusk, could you or @SpaceX confirm if that’s accurate? Curious about what caused it. #Starship #SpaceX

[Harris Cohen @HCohenTech]

@MatthewBerman That’s hilarious.

[Harris Cohen @HCohenTech]

What's all the fuss about this Jevons Paradox? Is it really a paradox? 🤔 And what about DeepSeek's Game-Changing Tech? 🚀

🔍 The Jevons Paradox is often labeled as a paradox because it seems to defy common sense: when technology increases the efficiency of using a resource, like coal in the 19th century, it can paradoxically lead to an increase in overall consumption rather than a decrease. Named after the English economist William Stanley Jevons, who observed in 1865 that more efficient steam engines resulted in more, not less, coal use, this phenomenon challenges our intuitive understanding of resource conservation.

📚 In economic theory, the Jevons Paradox isn't considered a theorem but rather an observation or phenomenon. It's a situation where efficiency gains lower the cost of using a resource, thereby boosting demand and potentially offsetting the initial savings from efficiency. This effect, known as the "rebound effect," is a topic of considerable debate.

🖥️ Now, fast forward to today, and we're seeing a modern echo of this paradox with the tech world abuzz over DeepSeek's LLM. Analysts are in shock and some in fear 😱, as DeepSeek demonstrates that you can achieve top-tier AI performance with significantly less powerful, and thus less expensive, versions of NVIDIA's chips. Much like Jevons' coal, where efficiency led to increased use, DeepSeek's approach could mean more companies jump into AI, potentially increasing the demand for even these less costly chips. This challenges the current market dynamics where high-end chips were thought to be indispensable for cutting-edge AI.

💡 Despite the paradox, efficiency improvements offer significant benefits, such as economic growth and reduced resource costs. However, Jevons' insight suggests that efficiency alone might not lead to sustainability without additional measures. Similarly, while DeepSeek's tech might democratize AI development, it also raises questions about the sustainability of the tech industry's reliance on ever more advanced, and expensive, hardware.

🔄 So, is the Jevons Paradox really a paradox? In the truest sense, yes, because it presents a scenario where the expected outcome (reduced consumption through efficiency) is subverted by an increase in use due to lower effective costs. And with DeepSeek, we might be witnessing this paradox in action, reshaping our expectations of technological advancement in AI.

[Harris Cohen @HCohenTech]

🚀 MindMed's Breakthrough in Anxiety Treatment

MindMed just dosed the first patient in Panorama, their SECOND Phase 3 study for MM120, a novel LSD treatment for Generalized Anxiety Disorder (GAD)!

What's New: Building on successful Phase 2b results, Panorama aims to change the game for GAD treatment.

Study Design: 12-week double-blind phase, followed by a 40-week extension.

Expectations: Topline data by H2 2026. Could this be the future of mental health care?

Why It Matters:
GAD affects millions, with little innovation since 2007.

MM120 offers rapid response & lasting remission.
#MentalHealthMatters #InnovationInPsychiatry #MindMed

[Harris Cohen @HCohenTech]

🚀 Disruptive AI Underdog: DeepSeek's R1 Model Shakes Up the Industry with Only $5.6M

You won't believe this, but a scrappy Chinese AI startup, DeepSeek, has just thrown down the gauntlet to giants like @OpenAI and @GoogleDeepMind with their groundbreaking R1 model. Here's why everyone's talking about it:

David vs. Goliath: With just a $5.6M budget, DeepSeek, led by billionaire Liang Wenfeng, has crafted a 671 billion-parameter model using a modest 10,000-GPU cluster.

Transparency Triumph: They’ve not only built this beast but also shared their entire playbook with the world, fostering an open-source revolution in AI development.

National Pride: This move has ignited discussions across China and sparked a global conversation about the future of AI.

Why Should You Care?

Unmatched Efficiency: DeepSeek managed to train their massive model with only 2,048 Nvidia H800s, proving that you don't need deep pockets to make deep impacts in AI.

Nvidia H800 GPUs, are a version of the H100 specifically adapted for the Chinese market due to U.S. export restrictions. These chips have lower inter-chip communication bandwidth compared to the H100s.

However, there are allegations and discussions suggesting that DeepSeek might have access to Nvidia H100 GPUs, which are more advanced and not supposed to be available in China due to export controls. Some industry insiders, including Scale AI CEO Alexandr Wang, have speculated that DeepSeek could possess around 50,000 H100 GPUs, although this is not officially confirmed by DeepSeek.

Challenging the Status Quo: This development questions the 'moat' around established AI companies, showing that innovation can come from anywhere.

Intensifying Global Race: The AI landscape is heating up with DeepSeek challenging the likes of @Meta, @AnthropicAI, and even Elon Musk's xAI in both tech and infrastructure.

Liang’s team at @HighFlyer has purported to have mastered the art of maximizing Nvidia chip performance, navigating around US export restrictions with flair. While it's up for debate whether DeepSeek can keep pace with the industry's behemoths, one thing's for sure: they've just added a thrilling twist to the AI saga.

🤖 What's Your Take? Is DeepSeek the dark horse we didn't see coming in the AI race, or will the established players maintain their lead? Drop your thoughts below!

[Harris Cohen @HCohenTech]

🚀 NVIDIA’s Next Frontier: Quantum Computing

Absolutely fascinated by this deep dive on how $NVDA (@NVIDIA) is quietly shaping the quantum computing landscape. By leveraging its GPU dominance, NVIDIA’s CUDA Q stack is not merely supporting quantum—it’s actively demystifying and accelerating it.

With hardware still nascent, expensive, and specialized, enterprise demand for next-gen computational tools continues to grow. A robust GPU-based simulator shortens the proof-of-concept cycle and builds confidence in quantum’s real-world potential.

Ultimately, @NVIDIA isn’t just dipping a toe in quantum—it’s building the bridge that countless developers, researchers, and enterprises will cross to get there.

#QuantumComputing #InnovationUnleashed

[Harris Cohen @HCohenTech]

I asked Grok ..
https://t.co/xgSLUyjoT7

[Harris Cohen @HCohenTech]

Groundbreaking discovery! Scientists at the Max Planck Institute (@maxplanckpress) and MIT (@MIT) have created a light-controlled magnetic state in an antiferromagnetic material. By using terahertz lasers tuned to the material’s natural vibrations, they triggered a shift in atomic alignment—creating a magnetic state that lasts long enough to study.

This breakthrough could pave the way for more reliable, stable, and energy-efficient memory technology, potentially revolutionizing data storage with chips that are both durable and interference-resistant.

Why does it matter? Antiferromagnets naturally resist external magnetic fields, making them ideal for next-gen memory chips that can store more data while consuming less energy.

Full details: https://t.co/EZnK0L5mFn

[Harris Cohen @HCohenTech]

Quantum Computers Just Got Smarter With Dual-Code Error Correction https://t.co/ytUQjO8SA8

[Harris Cohen @HCohenTech]

President Trump, Masayoshi Son, Larry Ellison, and Sam Altman have just unveiled “Stargate” — a massive $500B initiative involving Microsoft, OpenAI, SoftBank, Nvidia, and Oracle. This landmark collaboration aims to create 100,000 jobs and solidify the U.S. as a global AI powerhouse. Stay tuned for more cutting-edge updates! #AI #Stargate #TechNews #Innovation