If you were to assess OpenAI before Friday afternoon, it seemed to possess everything an enterprise buyer—and even an investor—could desire in a startup: a groundbreaking product in ChatGPT, a charismatic CEO, and significant potential for future revenue. The company appeared as stable as a startup could be—until everything changed. As the situation evolves, it seems that CEO Sam Altman is poised to join Microsoft, potentially taking co-founder Greg Brockman and a majority of his team with him.
When chaos erupted on Friday, the dangers of vendor lock-in became glaringly apparent, leaving startups heavily invested in OpenAI technology grappling with uncertainty over the situation's outcome. Vendor lock-in has long been a concern for enterprise buyers. Although numerous large language models (LLMs) are available from various vendors, OpenAI’s GPT-3.5 and GPT-4 remained prominent leaders in the field.
Many companies I've spoken with emphasize that despite OpenAI’s market dominance—bolstered by Microsoft’s major investment—there's a sense that the current AI landscape is still at an early stage, encouraging a flexible approach when selecting a model. Startups that opted for flexibility over reliance on a single AI vendor must now feel a sense of relief. If there’s one key takeaway from this ongoing drama, it's the importance of avoiding dependence on any one vendor.
Founders Proceed with Caution
Founders who placed all their bets on OpenAI now face an uncomfortable predicament as uncertainty looms over the company. One startup founder, speaking on background for candidness, noted that their organization was on the verge of signing a major contract with OpenAI when the news broke, leaving them in a state of limbo.
“It’s unclear how things will evolve with Sam and several others heading to Microsoft,” the founder shared. “I’m curious to see what new initiatives they announce there... I'm hopeful OpenAI will continue developing an excellent developer platform, but there's a possibility that the pace of updates may slow down, so we’ll be keeping a close watch.”
Effective Accelerationism, Doomers, Decels, and AI Priors
Arthur co-founder and chief scientist John Dickerson, whose startup assists clients in monitoring and evaluating AI models, expressed that while his company greatly values OpenAI, recent developments underscore the necessity for businesses to continually assess alternative options.
“We heavily utilize OpenAI products, both internally and for our clients. However, these last few days reveal the vulnerabilities of relying on a single provider or model to support production systems,” he remarked.
For instance, Box has developed its AI layer to accommodate multiple models. Although the company hadn't anticipated the recent events surrounding OpenAI, CEO Aaron Levie noted that their flexibility is now more valuable than ever.
“Our entire strategy focuses on building an abstraction layer that operates across any AI model and vendor,” he explained. “While we could never have foreseen these specific circumstances as a justification for needing an abstraction layer, we’re designed for maximum flexibility regarding our AI model sources, positioning us well to adapt.”
VCs Advocate for Flexibility
It’s no surprise that executives are seeking diverse options among LLM vendors, a recommendation echoed by investors. Lily Lyman, general partner at Underscore VC, indicated that while they promote open-source models, they understand some startups may not initially embrace this approach.
“We acknowledge that many companies, including some in our portfolio, have relied on OpenAI,” Lyman commented. “While this leadership upheaval is unsettling, the underlying technology remains highly valuable—especially to Microsoft. We hope for a resolution that maintains this value while addressing safety and alignment challenges.”
Despite the turmoil, general partner at Upfront Ventures Kobie Fuller noted that he does not anticipate a slowdown in innovation within the AI arena or among large language models. “The coming weeks should be interesting, but I'm not worried about any of my companies relying on LLMs or the disappearance of the entire LLM infrastructure. That’s definitely not the case,” he stated.
Early-stage companies naturally strive for top-performing models, and until now, OpenAI has held that position, as noted by Jon Lehr, co-founder and general partner at Work-Bench. However, he predicts a rise in open-source alternatives and encourages startups to embrace these options.
“The reality is that businesses need to avoid vendor lock-in with a single supplier. The recent developments around OpenAI will likely stimulate a more diverse ecosystem and promote open-source solutions as a preferred choice among venture-scale founders building with LLMs, enabling greater control over model weights and output,” he remarked.
One thing is certain: companies have always understood the risks of placing all their reliance on one vendor. While no one could have predicted this particular scenario, it serves as a reminder of the inherent dangers of depending solely on one source, even one that appears stable.
If there’s a silver lining to this chaotic situation, perhaps it’s the rediscovery of that crucial lesson.
Ranking
