0

Like the Star Wars tension between the Dark Side and the Jedi side, there’s intrigue in the OpenAI-Anthropic race and the paradoxical surge Anthropic saw when it stuck to its “safety-first” roots.  The “focused vs. scattered” strategies differentiate the two companies, who have a deeply personal, high-stakes rivalry that makes their respective dominance in the lucrative enterprise AI market all the more compelling.

Over the past year, I’ve heard many analysts say, “compute is the ballgame, and chips are the stadium.” And right now, OpenAI and Anthropic are the All-Star players – one the incumbent star, and the other a rising star whose annual revenue run rate has gone from $1 billion in January 2025 to $30 billion in April 2026 – which amounts to 1,420% annualized revenue growth!

As a result, OpenAI is trying to make a stronger case for why investors should hold and not rotate out. Sam Altman is projecting a staggering 30 GW of capacity, for which he plans to spend $600 billion on AI infrastructure, like data centers and chips, over the next five years.  

Currently, OpenAI capacity stands at 1.9 GW, where Anthropic stands at 1.4 GW. Long term, Anthropic’s plans for 7-8 GW of capacity by 2027 are dwarfed by OpenAI’s plans, but some analysts argue Anthropic’s valuation is more “sensible” due to its specialization in coding and lower operational burn. Yet others believe OpenAI’s scale and compute will keep it on top, even if it’s many side ventures slow down or peter out, as with the recent killing of Sora and current pause of Stargate UK.

Other financial sidebars include the pending $134 billion lawsuit against Elon Musk and management-team “turmoil” about its rapid commercialization, with CFO Sarah Friar voicing concern that the company isn’t operationally ready for a Q4 2026 IPO. This is layered on top of the degradation in investor sentiment when former employees, like chief scientist Ilya Sutskever, voiced concerns over a “consistent pattern of deception” by OpenAI about safety protocols.

OpenAI is trying to right its wrong, and improve PR, as with its recent purchase of the very popular Technology Business Programming Network (TBPN), and its recent release of policy proposals like its Public Wealth Fund to distribute AI-generated economic benefits broadly, and its proposal of a 32-Hour Work Week, with full pay, as an “efficiency dividend” for workers displaced by AI automation.

Overall, OpenAI has room for trial and error, as it maintains significant leads in global scale, pricing, and specific technical domains. With hundreds-of-millions of weekly active users, OpenAI dwarfs Anthropic’s user base (which is about 5% that of ChatGPT’s), plus, its $122 billion funding and $852 billion valuation is the largest in history.

In sharp contrast, Anthropic’s founder team is intact and very focused on being a “deep specialist” for the enterprise. All is not golden, however, as the recent “dual use” threat of “Mythos,” which Anthropic has deemed “too powerful and dangerous for public release,” gave it a taste of what bad press feels like. However,  its ability to autonomously detect and exploit software vulnerabilities, at scale, may have a silver lining in relation to enterprise growth and high-value partnerships. If Mythos is indeed a “step change” rather than an iteration (currently winning a 93.9% score on software engineering benchmarks), then Anthropic’s pivot to Project Glasswing – a defensive cybersecurity initiative– could give it a big boost. If it can autonomously find and exploit major software bugs so that defenders can proactively patch vulnerabilities before adversaries use them, it could be a source of premium revenue streams.

Public perception will matter

Like the Star Wars tension between the Dark Side and the Jedi side, there was a palpable feeling of distrust that emerged when OpenAI was perceived to pivot away from its “safety-first roots.” This came to a head when Secretary of Defense Pete Hegseth’s threat to blacklist Anthropic paradoxically strengthened Anthropic’s commercial appeal and contributed to its historic revenue surge. By refusing to relax safety guardrails, Anthropic was catapulted into the Jedi-like “light side” for enterprises and institutions that seem to value a “principled, safety-first” approach.

That said, the damage to Anthropic when first deemed a “supply-chain risk” by the Trump administration did cut Anthropic revenue by multiple-billions-of-dollars, according to court filings by Anthropic executives, who noted the loss of a $100 million revenue pipeline from a customer that switched to a competitor after the blacklisting, as well as $180 million in potential losses due to the legal uncertainty that caused a disruption with financial institutions. Additionally, the critical loss of its $200 million Dept. of War contract last month is nothing to sneeze at.

For all of these reasons, analysts are deeply divided over who might win the OpenAI-Anthropic race. When it comes to revenue, Wall Street is closely monitoring Anthropic’s annualized revenue run rate, with analysts from Jefferies and Morningstar suggesting this “revenue showdown” indicates Anthropic is winning the enterprise battle, while OpenAI remains the “formidable generalist” with a massive consumer base.

In a rare interview, tech influencer Ashutosh Shrivastava told Front Page that companies in the space watch what their competitors are doing, and then work relentlessly to adjust what they do accordingly: “Anthropic has been a trendsetter, working on something that is fabulous and sticking to their guns…they break the mold and disrupt.”

But, Shrivastava also pointed out that OpenAI’s funding is historic, and that its underlying research has been far ahead of the competition over the past few years, not to mention OpenAI’s Codex, which is a beloved code-generation tool for developers. He has also, in the past, predicted that OpenAI’s AgentBuilder and AgentKit would be a boost, allowing users to monetize their own custom agents.

Coexistence of AI and Ads

As of April 2026, a major strategic divide has emerged regarding advertising: OpenAI is aggressively pursuing ad revenue to offset its massive compute costs, while Anthropic has explicitly rejected ads to brand itself as the premium, “trust-first” alternative.  

For OpenAI, advertising is a critical pillar for long-term profitability, with projections that it will have about $100 billion in revenue by 2030, according to Axios reporting. Into 2026, OpenAI expects ads to generate approximately $2.5 billion, quadrupling to $11 billion by 2027. While Google remains “king” in the ad realm, with 100x the ad revenue of OpenAI, Altman’s aggressive push could  lead the company to capture nearly one- third of Google’s market share by 2030. 

The “focused vs. scattered” strategies differentiate Anthropic from OpenAI, with Anthropic focused on its core Claude reasoning models and tools like Claude Code, and OpenAI pursuing “side quests” in ads and chatbots.  Whether the leaner R&D-to-revenue ratio for Anthropic will beat out OpenAI’s diversification cannot be predicted, as so much changes so fast in this industry. In myh opinion, it’s going to come down to compute, and as of now, OpenAI holds a commanding lead in total raw compute capacity. But, Anthropic has established a lead in compute efficiency—generating significantly more revenue for every GW spent. 

Whether success comes down to raw power or efficiency is currently the most debated topic in the industry.