VCs Target AI Data Centers, Local LLMs, and Domain-Specific Models for Future Growth

Venture Capital Firms Embrace Emerging AI Opportunities Amid Slowdown

In response to declining deal activity and exit values, venture capital (VC) firms are increasingly focusing on promising AI opportunities poised for long-term growth. The latest Pitchbook Artificial Intelligence and Machine Learning Report highlights the ongoing challenges faced by VCs, including a significant drop in deal activity and exit values. Pitchbook identifies AI data centers, local large language models (LLMs), and domain-specific foundation models as key growth catalysts for VCs.

Market Challenges for VCs

AI and machine learning (ML) deal activity has decreased by 19% over the past year, plummeting from 8,968 deals in 2022 to 7,238 in 2023. The value of deals has also declined, with Pitchbook reporting only $2.7 billion in disclosed deal value during Q4 2023—the lowest seen since Q1 2019. M&A activity is down as leading tech firms opt for partnerships with LLM startups instead of acquisition.

Notable exceptions include AMD's acquisition of Nod.AI in MLOps, IBM’s purchase of Manta in database management, and ServiceNow's acquisition of UltimateSuite in predictive analytics. There are hopes that the anticipated IPO of semiconductor startup Astera Labs will revitalize deal values in the upcoming quarters.

Despite these setbacks, positive signs of long-term growth are emerging. Generative AI leaders secured $6 billion across 194 deals in Q4 2023, largely funded by major tech players like Microsoft and Google, eager to access the latest LLM technologies. Additionally, investment in horizontal platforms reached a record $33 billion in 2023, while funding for vertical applications dropped to levels not seen since 2020.

Identifying New Investment Opportunities

VCs are keen to build organizational structures and product strategies that leverage Nvidia’s rapid innovations, particularly in GPU technology. Pitchbook identifies AI data centers, local LLMs, and domain-specific foundation models as key areas set to thrive on Nvidia’s momentum as a market leader.

Nvidia reported impressive fourth-quarter FY 2024 revenue of $22.1 billion, a 265% year-over-year increase and 22% sequential growth. The data center segment experienced remarkable growth, surging 409% from the previous year to $18.4 billion. CEO Jensen Huang emphasized the diverse demand for data processing and AI solutions across various industries, including automotive, finance, and healthcare.

AI Data Centers: A Catalyst for Growth

AI data centers, optimized for AI workloads, focus on maximizing performance while minimizing power consumption and heat output. Designed to support high-performance servers, storage, networking, and specialized accelerators, these data centers prioritize sustainability.

IDC estimates that $8 billion has been invested in generative AI infrastructure, producing $2.1 billion in cloud revenue and $4.5 billion in application sales. Pitchbook forecasts that AI data centers will not achieve software-as-a-service (SaaS) level margins until 2027, prompting startups to seek cost-effective solutions that promise significant GPU hour savings.

Pitchbook notes that startups are currently offering 50% to 70% savings on GPU hours with advanced Nvidia A100s and exclusive access to the latest H100 chips. Lambda, a leading startup in GPU cloud services, has developed the largest cluster of H100 chips among public cloud providers, surpassing Google and Oracle.

VCs are exploring opportunities to create and collaborate with colocation provider ecosystems. Specialty cloud providers, carving out a $4.6 billion market from the nearly $150 billion internet-as-a-service sector, distinguish themselves by offering AI chip availability, local presence, and support for multicloud environments and various legacy hardware systems.