Deep Tech Opportunities Report 2024 (DCVC (Data Collective))

Summary

• In 2021, U.S. venture capital firms invested a record $348 billion before overall investment activity declined.p.7
• Data centers running generative AI models are projected to consume 7.5 percent of all U.S. electricity by 2030.p.14
• Training the GPT-3 AI model required almost 1,300 MWh of electricity, equivalent to the annual energy consumption of 130 average U.S. homes.p.15
• Five major industrial sectors—aviation, steel, cement, fertilizer, and livestock—accounted for roughly 21.4 percent of global greenhouse gas emissions in 2020.p.26
• China approved 114 gigawatts of new coal power plants in 2023, representing a 10 percent increase compared to 2022.p.28
• Water-driven weather events are forecasted to cause $1.8 trillion in damage over the next 10 years, making up 69 percent of expected climate change losses.p.30

Key insights

1. Addressing global existential threats requires sustained economic growth and surplus capital to fuel deep tech innovations.p.4
2. Proprietary, high-quality domain datasets are emerging as the primary critical asset determining AI model performance.p.10
3. Integrating site-specific context and real-world operational data with trainable expert systems is essential for effective industrial robotics.p.13
4. Enhanced geothermal energy systems leverage horizontal drilling and fracturing techniques to deliver scalable clean firm power.p.24
5. Low-temperature CO2 electrolysis can generate drop-in sustainable aviation fuel without requiring changes to existing aircraft infrastructure.p.26
6. Extracting magnesium from seawater offers a zero-carbon alternative to traditional energy-intensive mining and thermal refining.p.31
7. Dual-use AI models and small autonomous hardware platforms enable defense sectors to counter complex asymmetric threats.p.34
8. Real-time 3D light-field sensing and computer vision augment surgical precision without exposing patients to extra radiation.p.38