A little-known US AI startup is looking to steal the bread and butter of pioneering AI companies by allowing corporate clients to create their own custom AI models.
OpenAI and Anthropic’s business models largely revolve around selling AI to enterprise customers, through APIs and enterprise subscriptions. Sarah Friar, OpenAI’s chief financial officer, reportedly told investors in a closed-door meeting in August that while the startup’s revenue had historically been split about 60/40 in favor of individual customers, that ratio had recently reversed: “The majority of our revenue now comes from enterprises,” Friar said: according to CNBC. In February, Anthropic said it had more than 500 enterprise customers spending more than $1 million per year on API usage or Claude subscriptions. A leaked IPO prospectus revealed last month that a quarter of Anthropic’s revenue last year came from just two clients (both unnamed, but almost certainly enterprises).
Several powerful open-source and open-weight models developed by Chinese AI labs have proven a tempting alternative, as they can be modified to fit each client’s needs. needs and, more importantly, are less expensive than US border models. ChatGPT and Claude, on the other hand, offer a confusing approach jumble of pricing plans this can be difficult to follow. The era of token-maxxing, where companies encouraged their employees to use AI tools as much as possible in their daily work, has given way to sober frugality. But Chinese models also pose their own problems: Most large U.S. companies are understandably hesitant to hand over sensitive customer information to an AI developer with direct ties to the Chinese Communist Party.
Reflection AI wants to fill this gap.
Launched in 2024 by two former Google DeepMind researchers, the company is working to build what it calls an “AI factory”: software that lets companies create their own AI models, eliminating the need to rely on subscriptions to closed systems like ChatGPT or risky Chinese alternatives. “Today, the best open models come from China,” the startup’s co-founder and CEO Misha Laskin told CNBC in July. “As a result, global AI adoption is largely being built on Chinese models, including (by) American companies. And that’s great, but we want to build a thriving AI ecosystem with lots of competition and lots of players, and we’re building the counterbalance to that by building big, open models here in America.”
According to a new report from Axios, Reflection is also preparing to launch its own open-weight AI model to directly compete with the largest border labs on both sides of the Pacific. Although the report is light on detail, it does indicate that the model could be expected soon and will have capabilities comparable to China’s most advanced open-weight models. The first Reflection model will be “will be joined by other open-weight models from other Western players this month,” Axios added in its report. Reflection did not immediately respond to Gizmodo’s request for comment on the new model.
Echoing Nvidia’s Jensen Huang and Meta’s Mark Zuckerberg, Reflection believes that a completely open ecosystem is the safest and most equitable path to the future: “We must build open models so good that they become the obvious choice for users and developers everywhere, ensuring that the foundations of intelligence remain open and accessible rather than controlled by a few,” the company wrote in a statement. blog post last year.
Nvidia had invested around $800 million in Reflection in March this year, according to the Wall Street Journal. Reflection also signed a multi-year partnership with SpaceX in June, in which it agreed to pay Elon Musk’s aerospace company $150 million per month for computing generated by the Colossus data center in Tennessee.
Since little is known at this point about Reflection’s next model, it’s too early to say whether it will be capable enough to convince businesses en masse to abandon ChatGPT or Claude. Regardless of how it plays out, the growing enthusiasm around open models is likely to force OpenAI and Anthropic to focus less on creating new, more powerful AI systems and more on making their existing products more affordable and accessible, especially as they both get closer to their stock market debuts.
Gn bussni
