Welcome to Discover Daily by Perplexity , an AI-generated show on tech , science and culture . I'm Alex . Today we're exploring how a new AI model is shaking up the energy industry . But first let's look at what else is happening . Google has just made its Gemini 2.0 AI models widely available .
The lineup includes Flash Pro and Flashlight , each designed for different use cases . Flash , now generally available , offers low latency and enhanced performance with a 1 million token context window . Pro , an experimental version excelling in coding tasks , is available to Gemini advanced users .
Flashlight is introduced as the most cost-efficient option currently in public preview . Users can access Gemini 2.0 Flash through the Gemini app on both desktop and mobile platforms For those seeking advanced capabilities . The experimental Gemini app on both desktop and mobile platforms For those seeking advanced capabilities .
The experimental Gemini 2.0 Pro is available to Gemini Advanced subscribers . The app now offers a model drop-down feature allowing users to select the 2.0 Flash Thinking Experimental version . Developers can now use Gemini 2.0 Flash for building production applications through the Gemini API in Google AI Studio and Vertex AI .
This comes with higher rate limits and simplified pricing . The new release represents a significant step forward in Google's AI capabilities , with improved performance across various tasks and use cases . Google plans to introduce several new features to Gemini 2.0 .
Bi-directional voice and video interactions and advanced spatial understanding for generating dynamic representations of 2D and 3D layouts . The company is also focusing on deeper integration with its ecosystem and sustainability initiatives to reduce energy consumption .
Moving on to our next story , the release of Hot Tub , the first native porn app for iPhones in the European Union , has sparked controversy . This app challenges Apple's strict content policies under the EU's Digital Markets Act , which requires the allowance of third-party app stores like Alt Store .
Hot Tub functions as an aggregator , allowing users to search and stream videos from popular adult websites . It offers features like ad-free browsing , built-in search functionality and customizable controls . The app's independent developer , known as C1Day3R , designed it to provide a more secure and user-friendly experience for accessing adult content on iOS devices .
The EU Digital Markets Act , which came into effect in 2024 , has played a crucial role in this controversy . It compels Apple to allow alternative app stores on iOS devices in the European Union , aiming to promote competition and reduce the dominance of large tech companies .
This has created tension between Apple's content standards and the EU's push for open digital markets . Apple has strongly opposed the release of Hot Tub , expressing deep concern over safety risks , particularly for children . The company emphasized that it does not approve of the app and would never offer it in its app store .
Apple maintains that it's being forced by EU regulations to allow the distribution of apps like Hot Tub , despite concerns about user safety . This development marks a significant shift in iOS app distribution , particularly in the EU . It challenges Apple's long-standing content restrictions and raises questions about user safety and parental controls .
The situation could lead to fragmentation in the iOS app market , with different content available in the EU compared to other regions . Now let's dive into our main story of the day . Deepseek , a Chinese AI startup , has introduced its R1 model , which claims significant improvements in energy efficiency compared to existing systems . To existing systems .
This breakthrough has triggered a significant sell-off in energy stocks and is challenging assumptions about AI's growing energy demands . The DeepSeek R1 model represents a significant leap in AI energy efficiency . It requires only 2,000 NVIDIA chips and less than $6 million worth of computing power .
While initial reports suggest substantial energy savings compared to similar USAI systems , it's important to note that these efficiency gains might be partially offset by higher energy intensity during response generation due to the model's chain-of-thought approach . Response generation due to the model's chain-of-thought approach .
The model's efficiency challenges the prevailing narrative of AI's ever-increasing energy demands , potentially reshaping the landscape of AI development and its impact on global energy consumption . The announcement of DeepSeek's energy-efficient AI model triggered a significant sell-off in energy-related stocks .
Texas Pacific Land saw a 7.7% drop , while EQT and Antero Resources declined by over 9% . Independent power producers were hit particularly hard , with Constellation Energy plunging 21% and Vistra dropping 28% . This market reaction demonstrates how heavily the energy industry had been banking on AI-driven demand growth .
Deepseek's innovation has disrupted plans for energy infrastructure expansion . The need for expanding data center capacity and associated energy infrastructure may be significantly reduced . Pressure on power grids could ease , potentially reducing the need for grid expansion projects .
Demand projections for natural gas , previously expected to be a primary fuel source for AI-related power generation , are now uncertain . The investment case for nuclear power , including next-generation small modular reactors , may be weakened as AI's energy demands are reassessed .
On the flip side , lower energy requirements could accelerate the transition to fully renewable energy sources . For data centres , requirements could accelerate the transition to fully renewable energy sources for data centers . Companies involved in natural gas pipeline operations face uncertainty as projected demand growth is questioned .
The future of AI energy consumption remains uncertain . Current estimates suggest AI-driven data centers could account for 12% of US electricity demand by 2028 , though this projection may need revision if more efficient AI models become widespread . However , the realization of these energy savings depends on widespread adoption of more efficient AI models .
Mass switching from existing AI systems to more efficient ones like deep-seeks is necessary for substantial changes in electricity demand . The integration of AI in energy management could lead to further efficiencies , with AI optimizing energy distribution and consumption across grids .
Renewable energy adoption in data centres may accelerate as lower energy requirements make it more feasible to operate entirely on solar and wind power . The impact on energy infrastructure planning remains uncertain , potentially slowing the development of new power plants and transmission lines if AI energy demand growth is less than anticipated .
We'll see how other AI companies respond to DeepSeek's innovation and how energy companies adjust their strategies in light of these developments . That's it for today . Thanks for tuning in . Don't forget to subscribe on your favorite platform For more info on anything we covered today . Check out the links in our episode description and don't forget .
You can now access Perplexity's AI-powered knowledge base on the go with the mobile app available for both Android and iOS . There's also the Perplexity desktop app for macOS . In other Perplexity news , the company has launched Sonar and Sonar Pro APIs enabling developers to integrate real-time AI search capabilities into their applications with unparalleled factual accuracy .
The new offering stands out for its affordability and advanced features , including double the citations of standard search and the ability to customize trusted sources , with companies like Zoom already leveraging the technology to enhance their video conferencing platform . We'll be back with more stories that matter . Until then , stay curious .