Anthropic is pulling ahead of OpenAI in the frontier model race, laying the foundations for a new hierarchy of dominance within the rapidly developing artificial intelligence sector. From the first to the second quarter, Anthropic more than doubled its revenue while OpenAI's increased by just 18%, the Wall Street Journal reported Tuesday. OpenAI's operating margins, which were already in the red, also sank lower – a bad sign ahead of a hotly anticipated initial public offering. Meanwhile, Anthropic is projecting as much as $200 billion in 2028 revenue, far exceeding the $47 billion revenue run rate that the company publicized in May, Reuters reported last week. Investors say it's much too early to demote OpenAI, but they have been expecting stratification and consolidation at the frontier level for some time, which will have consequences for frontier models' partner companies, customers and suppliers. "Anthropic's partners, like HUT that I initiated coverage on yesterday, potentially win. OpenAI's potentially lose," Paul Meeks, head of technology research at Freedom Capital Markets, told CNBC Wednesday. " ORCL is probably the most threatened here because its [remaining performance obligations are] dominated by OpenAI." "That being said," Meeks cautioned, "I don't think we should throw in the towel yet on Team OpenAI. It's way too early for that." Back at the end of April, investment bank Jefferies put out a basket of names levered to OpenAI, including Oracle, CoreWeave , SoftBank and Broadcom , among others. Since the last trading day in April, Oracle stock is down about 12%; CoreWeave is down about 17%; and Broadcom is down about 13%. Nvidia is up about 10% and Microsoft is up about 18% over the same period. While webs of interconnected, circular investments across computing supply chains make it unlikely that OpenAI and its suppliers will be cleaved away from other segments of the industry, analysts do see some partitioning within these webs. Google parent Alphabet and Amazon stand to benefit from Anthropic's ascendence both at the level of cloud computing and semiconductors. "Since Anthropic gets most of its compute from Google and Amazon, it gets most of its compute on Trainium chips and TPUs, not Nvidia chips," Gil Luria, head of technology research at DA Davidson, said. "And that feeds into a whole ecosystem, including Broadcom." While Anthropic looks to be overtaking OpenAI at the moment, OpenAI is still solidly stitched into the fabric of the AI buildout, and analysts don't think the company is going anywhere. "It's Coke and Pepsi," Yi Fu Lee, managing director at Benchmark, said. "At the end of the day, the models are interchangeable." This interoperability is getting an additional boost from open-weight models – both from American and Chinese companies, such as Moonshot AI and DeepSeek. These algorithms are cheaper than their strictly proprietary rivals and often do just as good a job. Plus, multiple models can be coordinated to work together. Most on Wall Street believe that AI is on a long march toward becoming a general-purpose commodity, which means that it will get cheaper as it becomes more established and more widely adopted. In the short term, however, it could be a bumpy ride for companies that fall behind. Dan Nathan, principal of RiskReversal Advisors, said Wednesday that he's worried about Oracle, which signed a $300 billion cloud computing deal with OpenAI. "Oracle is one of those [companies] where you'd say, yes – this is an OpenAI situation," he said. "They gave them a $300 billion contract to build out compute, but if they can't do that, which they might not be able to, to the speed that OpenAI wants it and needs it, then you're going to see Oracle just absolutely collapse."