By Jon Harms, Senior Director, with Noah Yago, Vice President
The AI market has come a long way in the last 24 months. The release of ChatGPT in late 2022 marked a significant milestone, making AI technologies more accessible and sparking a surge of interest and investment. Anthropic, Cohere, Mistral and others have followed suit with their own competing foundation models, and investors have flocked to the space. According to Pitchbook, over $40 billion was invested in foundation models and other Core AI startups in 2024, approximately double the volume of the previous year.
Trends Shaping the AI Market Today
The impact of AI is already being felt. AI generated content appears at the top of every search result. Employees use AI to craft emails, summarize documents, perform research, write code, and analyze data. AI turned NVIDIA into a $3 trillion company, and announcements by China’s DeepSeek erased roughly $1 trillion of total stock market value overnight. AI considerations (and fears) are driving new regulations and beginning to shape the geopolitical arena.
But the market is still relatively nascent and just starting to reach the inflection point where enterprises are gradually shifting from a “proof of concept” phase exploring how to use AI to a production phase where GenAI capabilities are actually being rolled out. In fact, Cisco’s 2024 AI Readiness Index reveals that 98% of organizations feel increased urgency compared to a year ago to deploy AI strategies, and 85% believe they have less than 18 months to do so or they will experience negative effects to their businesses. As organizations take their first steps in the AI journey, it’s forcing technologies to become more cost effective, performant, and efficient. Training AI models, a costly and resource-intensive process, is becoming more affordable due to advancements in technology and optimization techniques. There's a growing trend towards smaller models, which can be faster, cost less to deploy, and can run at the edge. AI Assistants and natural language interfaces are quickly becoming table stakes, while the focus shifts towards AI Agents capable of more dynamic and autonomous actions. As these trends continue to materialize and evolve, enterprise spend will move down the value chain from hardware infrastructure where the lion’s share of monetization happens today to software implementations that put the technology to use in the real-world.
How Lightning AI Can Help
To unlock the hardware-to-software value shift, AI developers need to be enabled to build powerful, usable, safe software. And they need to be able to do so without being bogged down by underlying infrastructure, security, and governance requirements. It’s like a sequel to a movie we’ve seen before. The original movie was the Cloud, and organizations acutely remember (or are still in the middle of) the challenges faced in their cloud transformation journeys. As with the Cloud, the complexity of deploying, maintaining, securing, and monitoring robust AI environments can be overwhelming and requires specialized knowledge, diverting developers' attention away from their primary focus – writing code and building innovative applications. These challenges are exacerbated in multi- and hybrid-cloud deployments, and enterprises need the help of independent, cloud-agnostic vendors with deep expertise in AI hardware orchestration. That’s where Lightning AI comes in. By alleviating the AI management bottleneck, Lightning allows developers to focus on code, fail fast, deploy often, share models, and generally speed up the software development cycle. Will Falcon, Lightning AI's Founder and CEO, explains it well. "When you buy a car, you don't go learn about the mechanics of an engine. You just drive the car. So why should I have to learn about Kubernetes or Docker or any of this to actually go build an AI product? You shouldn't."
With Lightning AI, developers can access GPUs, seamlessly scaling up and down and even switching between CPUs and GPUs. They can train new models and deploy those models with zero setup of underlying infrastructure. They can write code in their browser or in their local IDE. They can choose their own AWS or GCP environment, or they can use Lightning’s cloud. Lightning also gives enterprises the flexibility to deploy production-ready AI models, APIs, and services – eliminating integration complexity with a marketplace of pre-built, scalable solutions. And they can do it all securely and in compliance with HIPAA and SOC2 standards.
Lightning’s mission is simple: to get customers out of “POC mode” and into “production mode” by giving developers everything they need in one place in an interface they are familiar with, so they can write code faster.
Why Cisco Invested in Lightning AI
Last June, Cisco announced its $1 billion AI Fund. Along with the fund, Cisco announced three foundational investments in Cohere, Mistral AI and Scale AI and we have made over a dozen AI investments since, including most recently our investment in Anthropic’s $3.5 billion Series E round. We have been taking a broad approach investing up and down the AI stack, and our investments to date span foundation models, infrastructure, security – and with Lightning AI we have now invested in AI application development, deployment, orchestration and management. The breadth of these investments demonstrates our commitment to incubating secure, reliable and trustworthy AI solutions. We are also happy, active, paying customers of Lightning AI. Today, we use the platform to train models for security and networking, which will become the foundation for customer AI products. Based on our firsthand experience, we believe in the power of the platform and envision a “better together” story as partners and investors. More to come.