In partnership with
Happy Tuesday, AI & Data Enthusiasts! Baidu is ramping up competition with ERNIE 4.5 & X1, taking over DeepSeek R1 and GPT-4.5. Meanwhile, an OpenAI exec is venturing into AI-driven materials science, and xAI is expanding into video generation. Plus, weāre exploring the growing energy concerns - did you know an AI-powered Google search can consume 23-30 times more energy than a traditional one?
In todayās edition:Ā
š ERNIE 4.5 & X1 take on DeepSeek R1 & GPT-4.5
š¬ OpenAI Exec Launches Materials Science Startup
š xAI Acquires Hotshot for AI Video
š Rising Concerns Over Impact & Transparency
- Naseema Perveen
WHAT CAUGHT OUR ATTENTION MOST
š Baidu Strikes Back: ERNIE 4.5 & X1 Challenge DeepSeek R1 & GPT-4.5
Baiduās stock surged 10.7% after unveiling ERNIE 4.5 and ERNIE X1, two AI models designed to rival DeepSeek R1 and GPT-4.5. Baidu claims its latest models offer better reasoning, multimodal capabilities, and cost efficiency.
Model | Performance | Cost per Million Tokens | Key Features |
|---|---|---|---|
ERNIE 4.5 | Surpasses GPT-4.5 in multimodal tasks | 1% of GPT-4.5's cost | Text, image, audio, video understanding |
ERNIE X1 | Matches DeepSeek R1 in reasoning | $0.28 (half of DeepSeek R1ās cost) | Planning, reflection, evolution |
DeepSeek R1 | Strong in reasoning & math | $0.55 | Competitor to GPT-4 & Claude 3 |
GPT-4.5 | Leading in general AI tasks | Input: $75.00 | Best for coding, reasoning, and creativity |
š Market Reaction
Investors welcomed Baiduās AI push, sending shares up over 10% on Tuesday.
Analysts say this is a sign of renewed confidence in Baidu, but sustained success depends on real-world adoption.
Baiduās ERNIE X1 offers similar reasoning abilities to DeepSeek R1 but at half the price, making it an attractive option for cost-conscious enterprises. Meanwhile, ERNIE 4.5 claims to outperform GPT-4.5 in multimodal understanding.
š Baiduās Open-Source Strategy
Unlike previous releases, Baidu has open-sourced ERNIE X1, aiming to boost adoption.
However, some experts question whether this approach will generate enough revenue to keep Baidu competitive.
āļø Challenges & Drawbacks
DeepSeekās rapid growth: With backing from top Chinese investors, DeepSeek is aggressively expanding and could overshadow Baidu.
Trust issues: Baiduās past AI models havenāt always delivered on their hype, raising concerns about ERNIEās real-world impact.
Regulatory hurdles: Chinaās strict AI regulations could limit ERNIEās capabilities, making it harder to compete globally.
Cloud disadvantage: Unlike OpenAIās partnership with Microsoft Azure, Baidu lacks a dominant cloud infrastructure, potentially restricting enterprise adoption.
As the AI rivalry between Baidu and DeepSeek heats up, the next few months will determine whether ERNIE can truly challenge DeepSeek and GPT-4.5 - or if Baiduās lead is just temporary.
IN PARTNERSHIP WITH THE AI REPORT
Thereās a reason 400,000 professionals read this daily.
Join The AI Report, trusted by 400,000+ professionals at Google, Microsoft, and OpenAI. Get daily insights, tools, and strategies to master practical AI skills that drive results.
WORTH WATCHING
š Baiduās Ernie X1 & 4.5: DESTROYS GPT-4.5 & DeepSeek? š¤Æ
Baidu just dropped Ernie X1 and 4.5, claiming they outperform GPT-4.5 and DeepSeek in key benchmarks. Is this the new AI powerhouse, or just hype? Dive into the latest showdown in AI dominance! š
š Watch here: YouTube Link
KEEP YOUR EYE ON IT
š¬ OpenAI Exec Launches AI-Driven Materials Science Startup
OpenAIās VP of Research for post-training, Liam Fedus, leaves to launch a new AI-powered materials science company. OpenAI has expressed interest in investing and partnering with his venture, signaling the growing importance of AI in scientific research.
š A New AI Venture in Materials Science
Fedus aims to leverage AI for scientific advancements. His startup will focus on AI-driven materials discovery, competing with major players like Google DeepMind and Microsoft. Despite his departure, Fedus emphasized that OpenAI remains committed to AI for science and may collaborate with his company.
š Competition in AI-Powered Materials Discovery
The AI materials science space is heating up, with DeepMindās Gnome system already identifying potential new materials and Microsoft introducing MatterGen and MatterSim. Fedusā company will join this growing race.
š The Challenges of AI in Scientific Discovery
While AI holds promise for materials science, some experts remain cautious, questioning whether todayās AI models can truly uncover groundbreaking scientific innovations.
As the industry evolves, Fedusā new venture will be one to watch in the race to revolutionize materials discovery with AI.
š xAI Acquires Hotshot to Expand AI Video Capabilities
xAI, has acquired Hotshot, a San Francisco-based startup specializing in AI-powered video generation. The move signals xAIās growing ambitions in the generative AI space, positioning it against competitors like OpenAIās Sora and Googleās Veo 2.
Hotshotās AI Video Models
Hotshot, co-founded by Aakash Sastry and John Mullan, developed three video foundation models:
Hotshot-XL
Hotshot Act One
Hotshot
Originally focused on AI-powered photo editing, the company later pivoted to text-to-video generation, attracting investments from major VCs, including Lachy Groom, Reddit co-founder Alexis Ohanian, and SV Angel.
xAIās Plans for AI Video
During a livestream in January, he mentioned that a āGrok Videoā model could launch in the coming months. The acquisition of Hotshot suggests that xAI is moving quickly to develop its video generation technology.
Transition & Next Steps
Hotshot has begun shutting down new video creation services, with existing customers able to download their content until March 30. It remains unclear if Hotshotās entire team will join xAI, as Sastry declined to comment on staffing changes.
With xAI expanding its AI portfolio, this acquisition marks another step in Muskās broader vision for advanced AI-driven content creation.
š The Real Energy Cost of AI: Transparency and Impact
How much electricity does AI really use? That question is growing more urgent as AI adoption accelerates. While firms invest heavily in new data centers, researchers warn that the true energy footprint of AI remains unclear due to a lack of transparency.
š AIās Growing Energy Appetite
Data centers powering AI require vast amounts of electricity - often equivalent to tens of thousands of homes. In states like Virginia, AI-driven data center expansion is projected to double electricity demand in a decade.
Clusters of data centers put intense pressure on local power grids, raising costs and sustainability concerns.
š Measuring AIās Power Consumption
Researchers estimate AIās energy use through two main methods: supply-chain analysis and direct measurement of model requests.
A single AI-powered Google search could use 23-30 times more energy than a traditional search.
Transparency is limited - tech firms rarely disclose energy consumption for AI workloads, making precise estimates difficult.
š± Sustainability vs. Innovation
AI efficiency is improving, but growing usage may offset these gains.
Companies are increasing data center investments, yet reporting fewer details about their environmental impact.
Some initiatives, like the AI Energy Score project, aim to bring clarity to AIās electricity consumption.
š The Call for Transparency
Without reliable data, researchers struggle to assess AIās long-term energy impact.
Policymakers and environmental groups advocate for clearer reporting standards.
As AI advances, balancing innovation with sustainability will be crucial.
š” AIās energy use is an evolving issue - one that requires accountability from the companies driving this revolution. The more we understand, the better we can prepare for a future where AI and sustainability go hand in hand.
ICYMI
$$$ MONEY MATTERS
SoftBank buys $676M old Sharp plant for its OpenAI collab in Japan.
Salesforce to Invest USD $1B in Singapore over 5 Years.
AvatarOS snags a $7M seed round from M13 to build an AI-powered virtual influencer platform.
Googleās Anthropic Investment Tops $3B.
ServiceNow to buy Moveworks for $2.85B to grow its AI portfolio.
Thatās all for now. And, thanks for staying with us. If you have specific feedback, please let us know by leaving a comment or emailing us. We are here to serve you!
Join 130k+ AI and Data enthusiasts by subscribing to our LinkedIn page.
Become a sponsor of our next newsletter and connect with industry leaders and innovators.
