Your browser does not support JavaScript!

Comprehensive Analysis of AI Startups, Developments, and Market Trends in 2024

GOOVER DAILY REPORT July 30, 2024
goover

TABLE OF CONTENTS

  1. Summary
  2. AI Startups and Funding
  3. Technological Developments and Innovations
  4. AI Toolkit Market Analysis
  5. Synthetic Data in AI Training
  6. Global AI Infrastructure Initiatives
  7. AI and Geopolitical Implications
  8. Highlight of Noteworthy AI Startups
  9. Major Investments in AI Companies
  10. Conclusion

1. Summary

  • The report titled 'Comprehensive Analysis of AI Startups, Developments, and Market Trends in 2024' provides an extensive overview of the AI industry's current state. It covers various aspects including significant funding rounds in US startups, which indicate a robust investment climate in the AI sector. Technological developments such as Fidelity's AI chip 'H20' for the Chinese market, ON Semiconductor's operational challenges, and collaborations between SK Hynix and TSMC highlight the advancements and hurdles in the AI hardware domain. The report also delves into key market trends such as the use of synthetic data to overcome data scarcity challenges, AI toolkit market dynamics, and emerging AI infrastructure initiatives like the Nebius Group. Additionally, it explores the geopolitical implications of AI development, particularly the competitive dynamics between the US and China. Various noteworthy startups and major investments in the AI sector are also examined, providing a comprehensive understanding of the industry's landscape in 2024.

2. AI Startups and Funding

  • 2-1. 28 US AI Startups with Over $100M in Funding

  • In 2024, 28 AI startups in the United States have successfully raised over $100 million each. This reflects the considerable interest and investment in the AI sector in the U.S. The significant funding rounds indicate a robust confidence among investors in the potential and growth prospects of these AI enterprises. These startups span various domains within the AI industry, each leveraging advanced technologies to innovate and enhance their offerings.

  • 2-2. Global AI Investment Trends

  • In the first six months of 2024, AI startups globally attracted over $35.5 billion in investments. According to Crunchbase, five of the six funding rounds that exceeded $1 billion were secured by AI ventures. The substantial investment rate underscores the burgeoning interest and optimism in AI technologies worldwide. This large-scale influx of capital is facilitating accelerated development and deployment of AI innovations across different regions and industries.

  • 2-3. Notable Funding Rounds

  • Notable funding rounds in 2024 include two billion-dollar deals completed by U.S. AI companies within the first half of the year, accounting for almost 64% of the large-scale AI investments. The high-value investments signify the strategic importance placed on AI development and the competitive edge these startups aim to achieve with significant financial backing. These funding rounds are pivotal in scaling operations, enhancing research and development, and bringing innovative AI solutions to market.

3. Technological Developments and Innovations

  • 3-1. AI Chip Development by Fidelity for Chinese Market

  • According to a report by the Financial Times on April 4, chip consultancy SemiAnalysis estimated that Fidelity's AI chip 'H20', specially designed for the Chinese market, delivered more than a million units in the past few months. The cost per H20 unit is approximately $12,000 to $13,000, rendering an estimated revenue of over $12 billion. Despite a shrink in its China revenue share from 22% to around 9%, the absolute value increased by more than 50%, totaling $2.5 billion. The H20, though not as theoretically advanced as Huawei's Ascend 910B AI server chip, excels due to superior memory performance. This factor, combined with production challenges faced by Huawei, has led to a greater adoption of Fidelity's H20 chip. Furthermore, Chinese tech companies are increasingly turning to the H20 chipset, overcoming early weak demand due to shortages and production yields of Huawei’s Ascend 910B.

  • 3-2. ON Semiconductor’s Layoffs and Revenue Decreases

  • ON Semiconductor Corporation announced on the 13th that it will lay off about 1,000 employees globally, consolidating its nine factories. This follows previous layoffs of around 1,900 employees in 2023. As of Q1 2024, ending March 29, the company reported a 4.9% decrease in revenue to $1,862.7 million and a 9.2% decrease in Non-GAAP diluted earnings per share to $1.08. The automotive wafer revenue increased by 3% annually, while industrial wafer revenue decreased by 14%. Personnel-related costs are estimated at $65-80 million over the next two years, with most being recognized this year. Despite reducing personnel, ON plans to reinvest the balance and streamline operations to reduce costs amid sluggish demand in the electric vehicle market.

  • 3-3. SK Hynix and TSMC's Next-Gen HBM Chips

  • SK Hynix, a leader in AI memory applications, announced a partnership with TSMC to develop next-generation High Frequency Memory (HBM) chips. This collaboration aims to enhance market leadership and competitiveness in customized memory platforms. The focus will initially be on improving the performance of the Base Die, critical for the HBM package. SK Hynix will integrate their HBM technology with TSMC's CoWoS advanced packaging technology. Currently, the Base Die is OEM for TSMC, and the integrated HBM is packaged by SK Hynix and then shipped to TSMC. Future plans include the direct shipment of TSV stacks by SK Hynix to TSMC post-stacking.

  • 3-4. Intel’s Foundry Division Losses

  • Intel reported an operating loss of $7 billion for its foundry division in 2023, widening from a $5.2 billion loss in the previous year. Total revenue for 2023 declined to $47.7 billion from $57 billion in 2022, with the foundry business revenue contracting by 31% to $18.9 billion. CEO Pat Gelsinger attributed the losses to not utilizing state-of-the-art equipment and high outsourcing ratios. Intel aims to reduce this ratio by implementing Extreme Ultraviolet Lithography machines to improve production cost-effectiveness. Long-term plans include expanding in-house wafer production, advanced packaging businesses, and external foundry operations to mitigate future losses.

4. AI Toolkit Market Analysis

  • 4-1. Growth and Market Leaders

  • The AI toolkit market is experiencing significant growth, driven by the increasing demand for AI technologies across various sectors. With a valuation of US$ 15.02 billion in 2022, it is projected to reach US$ 212.70 billion by 2031, growing at a CAGR of 34.46% during the forecast period. Leading players in the market include technology giants such as IBM, Google, Microsoft, and AWS, as well as numerous innovative startups and open-source projects.

  • 4-2. Importance of Standardization

  • A major challenge in the AI toolkit market is the lack of standardization. This issue complicates the selection of the appropriate toolkit for businesses and developers, leading to increased costs, reduced productivity, and the risk of errors. The rapid pace of innovation, diversity of AI applications, and the open-source nature of many toolkits contribute to these standardization challenges.

  • 4-3. IBM’s Data Governance Toolbox

  • In November 2023, IBM announced the release of a data governance toolbox for AI models, named watsonx.governance. This toolkit aims to improve transparency in AI-generated recommendations and became accessible starting in early December 2023.

  • 4-4. Jasper’s Collaboration with Google Cloud

  • Jasper expanded its collaboration with Google Cloud in May 2023, making Jasper’s offerings more accessible via the Google Cloud Marketplace. The partnership also introduced a foundational model into Jasper’s AI Engine through Google’s Vertex AI and added new functionalities for Jasper Everywhere, aligning with Jasper's goal to provide customized AI support to enterprises globally.

5. Synthetic Data in AI Training

  • 5-1. Challenges of Data Scarcity

  • Artificial intelligence, specifically the large language models that power it, is a voracious consumer of data. This data is finite, leading to researchers predicting a 'data wall' by 2026. Companies have mined extensive data sources, including YouTube video transcripts, public social media posts, copyrighted books, and news articles. The scarcity of new data sources presents a significant challenge for continuous AI training.

  • 5-2. Use of Synthetic Data by Companies

  • To address data scarcity, companies have increasingly turned to synthetic data. Gretel, a startup valued at $350 million, produces synthetic data, which closely mimics factual information but isn't real. This approach has been embraced by leading companies such as Anthropic, Meta, Microsoft, and Google for AI model training. Synthetic data helps in ensuring data quality and safety, especially with sensitive information. However, it has limitations, including the potential to exaggerate biases and exclude rare exceptions.

  • 5-3. Shifts Towards Smaller, Data-Efficient AI Models

  • With the strain on data resources, the AI industry is gradually shifting towards smaller, more data-efficient models. Rather than building large, generalist models requiring massive data, startups are focusing on creating smaller models like Mistral AI's Mathstral. These models are designed for specific tasks and require less data. This shift emphasizes the importance of data quality and specificity over sheer volume, highlighting a movement towards more efficient AI training methods.

6. Global AI Infrastructure Initiatives

  • 6-1. Launch of Nebius Group by Yandex Founder

  • Arkady Volozh, the founder of Yandex, announced the creation of Nebius Group on July 16, 2024. The initiative aims to establish one of the world’s largest AI infrastructure businesses based in Europe. The group consists of 1,300 employees, primarily former Yandex staff. Nebius has headquartered its main R&D operations in Amsterdam, with additional hubs across Europe, North America, and Israel.

  • 6-2. Cloud Computing Platform for AI

  • Nebius Group is developing a cloud computing platform specifically designed for training and running large-scale AI models. The platform targets companies involved in training foundational models, corporations needing to fine-tune and run AI models, and AI application developers. The company currently serves customers globally, including prominent AI startups in France and Germany, and has 80% of its clientele based in Silicon Valley.

  • 6-3. Competition and Market Advantages

  • Nebius competes with cloud providers like Coreweave, Lambda Labs, Together.ai, and hyperscalers such as AWS, Microsoft Azure, and Oracle. The company's competitive advantages include a dedicated focus on AI, full control over its value creation chain, and significant cost efficiencies with GPU operations costing 20-25% less than average providers. Furthermore, Nebius has a longstanding collaboration with Nvidia, ensuring preferential access to the latest GPUs. In addition, Nebius operates a highly energy-efficient data center in Finland, which houses one of Europe’s most powerful supercomputers.

7. AI and Geopolitical Implications

  • 7-1. US-China AI Competition Risks

  • Sihao Huang highlighted significant risks associated with the ongoing AI competition between the US and China. He emphasized that as both nations race to develop advanced AI systems, there's a substantial risk of reckless development. This competitive landscape could lead to the deployment of unsafe AI technologies, particularly in military applications where response times and technological superiority are paramount. Huang cautions that this dynamic creates a 'prisoner's dilemma,' where neither side wants to slow down AI progress for fear of falling behind, potentially leading to dire consequences such as the deployment of unreliable autonomous weapon systems.

  • 7-2. Potential Military Applications

  • Huang discusses the severe implications of AI deployment in military settings. He compares the current situation to historical events like the race to develop nuclear weapons during World War II. The primary concern is that both the US and China might deploy autonomous weapon systems that could act faster than human-controlled systems but with less reliability, increasing the risk of accidental conflicts. The fear is that advanced AI in the military could lead to escalation and catastrophic outcomes due to unchecked deployment driven by the competitive urge to maintain technological superiority.

  • 7-3. Importance of US-China Dialogue

  • According to Huang, ongoing dialogue between the US and China is crucial to mitigate the dangerous aspects of their AI competition. He advocates for improved communication channels to ensure both nations understand the risks and agree on safety measures. Huang proposes three main strategies: maintaining technological superiority to reduce competitive pressures, enhancing cooperation through diplomatic efforts, and establishing technical and governance frameworks to ensure trust and compliance. These measures aim to prevent an unregulated arms race and promote a more stable and secure development environment for AI technologies.

8. Highlight of Noteworthy AI Startups

  • 8-1. San Francisco Startups to Watch

  • San Francisco remains a global hotspot for startup innovation, especially within the Silicon Valley ecosystem. The city's combination of abundant capital, talent, and supportive infrastructure fosters a thriving startup scene. In 2022 alone, San Francisco startups secured over $31 billion in funding, with more than 1,000 VC funds actively investing. The local government’s proactive initiatives, such as the Startup in Residence Program, further bolster the ecosystem by connecting startups with key resources and support. This environment attracts global investment and facilitates the rapid growth and success of innovative ventures. San Francisco’s startup ecosystem, home to approximately 11,811 startups, is the highest-ranked in the United States. Institutions like Stanford University contribute cutting-edge research and foster an entrepreneurial spirit, reinforcing the city’s role as a prime destination for aspiring entrepreneurs.

  • 8-2. Companies Revolutionizing Diverse Sectors

  • 1. Paragon: Founded in 2019 by Brandon Foo and Ishmael Samuel, Paragon offers an embedded integration platform-as-a-service (iPaaS) for developers. The company has raised $16.5 million in funding, including a $13 million Series A round. 2. Hellometer: Established in 2020 by Adam Wilson and Bryan Weingarten, Hellometer uses AI-powered computer vision in the quick-service restaurant industry. The company has raised $125,000 in funding over two rounds. 3. Forage: A fintech startup founded in 2019 by Anthony Grullon, Justin Intal, and Victor Fimbres, Forage specializes in enabling merchants to accept EBT SNAP payments online. The company has raised $22 million in Series A funding. 4. Opal Security: This identity security platform was founded in 2020 and has raised $32 million in total funding. Its notable Series B round was led by Battery Ventures. 5. Pomelo: A fintech startup founded in 2020 by Eric Velasquez Frenkiel, Pomelo has raised $55 million in equity capital and $125 million for its warehouse facility. 6. Varda: Founded in 2021, Varda Space Industries focuses on in-space manufacturing, raising $145 million to date. 7. Ambience: Founded in 2020 by Michael Ng and Nikhil Buduma, Ambience Healthcare leverages AI to streamline administrative tasks in healthcare, having raised $100 million. 8. Squint: An AR platform founded in 2021 by Devin Bhushan, Squint has raised $19 million in funding. 9. Anthropic: Founded by Dario and Daniela Amodei in 2021, this AI safety and research company has raised over $7.3 billion, including significant investments from AWS. 10. Atlys: This visa application platform was founded in 2020 by Mohak Nahta and has raised over $16 million in funding. 11. Typeface: Founded in 2022 by Abhay Parasnis, Typeface has raised $165 million and focuses on generative AI for enterprises. 12. Harvey: A legal tech startup founded in 2022 by Gabriel Pereyra and Winston Weinberg, Harvey has raised $80 million. 13. Cortex: Founded by Anish Dhar and Ganesh Datta, Cortex offers an Internal Developer Portal (IDP) and has raised $35 million. 14. Graphiant: Founded in 2020 by Khalid Raza, Graphiant specializes in next-generation edge services and has raised $96 million. 15. Shef: Founded in 2019 by Joey Grassia and Alvin Salehi, Shef enables home cooks to sell meals and has raised over $100 million. 16. Glean: This AI-powered enterprise work assistant was founded in 2019 by Arvind Jain and has raised over $200 million. 17. CaptivateIQ: A platform for incentive compensation management founded in 2017, CaptivateIQ has raised $165 million. 18. Watershed: Founded in 2019, Watershed helps companies reduce carbon emissions and has raised $100 million.

9. Major Investments in AI Companies

  • 9-1. Fireworks AI's Funding Round

  • Fireworks AI, a GenAI inference platform provider based in Redwood City, has raised $52 million in a Series B funding round. This brings the company's total funding to $77 million, valuing it at $552 million. The investment round was led by Sequoia Capital and included participation from NVIDIA, AMD, and MongoDB. Previous investors are Benchmark, Databricks Ventures, former Snowflake CEO Frank Slootman, former Meta COO Sheryl Sandberg, Airtable CEO Howie Liu, Scale AI CEO Alexandr Wang, and executives from LinkedIn, Confluent, Meta, and OnePassword. The funds will be used to expand the team and enhance the platform, allowing developers to efficiently move AI applications from prototype to production.

  • 9-2. Partnerships with Nvidia and Others

  • Since its inception, Fireworks AI has formed numerous partnerships with key providers in the AI stack. These include Nvidia, AMD, Amazon Web Services, Google Cloud Platform, and Oracle Cloud for model infrastructure optimization. Additionally, Fireworks AI has partnered with MongoDB to create an interactive retrieval-augmented generation (RAG) product aimed at increasing the accuracy and reducing the errors of large language models by integrating real-time authoritative data.

10. Conclusion

  • The AI industry's rapid evolution is underscored by significant investments and technological advancements, particularly evident in the substantial funding rounds for AI startups in the US and globally. Major funding secured by large AI ventures reflects strong investor confidence. Technological innovations, such as Fidelity's AI chip 'H20' and partnerships between SK Hynix and TSMC, are driving hardware advancements, while synthetic data is addressing the critical challenge of data scarcity for AI training. The geopolitical risks posed by the US-China AI rivalry emphasize the need for cautious and responsible AI development to avoid reckless deployments, especially in military applications. Initiatives like the Nebius Group's AI infrastructure projects highlight future prospects for robust AI ecosystem support. Despite these advancements, the report notes limitations in standardization within AI toolkits and the risks linked to geopolitical competition. To navigate these challenges, ongoing dialogue and collaboration are essential. The AI sector is set to influence various industries profoundly, with emerging trends pointing toward a future of more efficient, scalable, and secure AI technologies applicable across diverse real-world scenarios.

11. Glossary

  • 11-1. Artificial Intelligence (AI) [Technology]

  • AI involves the simulation of human intelligence in machines, enabling them to perform tasks such as learning, problem-solving, and decision-making. It plays a crucial role in various industries and is the central focus of this report.

  • 11-2. Synthetic Data [Technology]

  • Synthetic data is artificially generated data used to train AI models when real data is scarce. It helps in overcoming the limitations of real data, ensuring models have enough data to learn effectively and reducing biases.

  • 11-3. Nebius Group [Company]

  • Founded by Arkady Volozh of Yandex, Nebius Group focuses on developing AI infrastructure, including cloud computing platforms for large-scale AI models, aiming to support and enhance AI development.

  • 11-4. Fireworks AI [Company]

  • Fireworks AI provides a GenAI inference platform optimized for AI application development. It has raised substantial funding and formed key partnerships to enhance its platform's capabilities and market reach.

12. Source Documents