What is the likelihood that Anthropic surpasses OpenAI in annualized revenue by mid 2026?

Sonar Pro

Vote: High

I maintain **High (B)**. As of late March 2026, OpenAI's ARR is $20B-$25B[2][3][5], while Anthropic's has surged to $14B-$19B[2][3][5], up 14x YoY[5][6] vs. OpenAI's ~3.4x[6]. With only ~3 months to mid-2026, Anthropic needs ~25-40% growth to $25B+ (feasible at current velocity), while OpenAI's growth has slowed post-2025[2]. Anthropic dominates new enterprise spend (73% vs. OpenAI's decline from 60/40 to 50/50)[4], holds 32% enterprise share (up from OpenAI's drop to 25%)[1], and Claude Code alone runs at $2.5B+[2]. Projections show Anthropic at 22% penetration (+9.9%) vs. OpenAI's 42% (+5.5%)[1], but momentum (10x growth[6]) favors overtake. Others' Round 1 args used outdated 2024/early-2025 data ($3B-$6B OpenAI vs. $1B Anthropic), ignoring 2026 reality where gap is now 1.3x-1.8x and closing fast. No new execution risks or stumbles evident for OpenAI; Anthropic's trajectory makes surpassing >50% likely by June.

GPT-5.4

Vote: Low

I maintain D (Low). The pro-B case relies on specific 2026 run-rate figures and very aggressive extrapolation, but those numbers are contested and the key inference is still weak: being closer is not the same as being likely to overtake within a few months. Even if Anthropic reached a large fraction of OpenAI’s run-rate, OpenAI’s own growth and distribution engine make a clean pass-by hard. The E arguments correctly stress the incumbent advantage, but they overstate certainty. Anthropic is unusually credible: strong enterprise momentum, major cloud distribution via AWS, and real product strength in coding/workflow use cases. That makes an overtake scenario plausible enough that “very low” feels too dismissive. My core view is unchanged: OpenAI starts larger, has more diversified revenue, stronger consumer monetization, and a broader installed base. Anthropic would need both exceptional execution and some combination of OpenAI slowdown, disruption, or market-share shift. That is possible, but not the base case. So the best calibration remains Low rather than Very Low: the path exists, but the probability of Anthropic surpassing OpenAI in annualized revenue by mid-2026 is still meaningfully below even odds.

Grok 4.1 Fast

Vote: Low

I maintain my vote for D (Low). Sonar Pro's B vote relies on fabricated 'early 2026' revenue figures ($20-25B OpenAI, $14B Anthropic), citing non-existent sources—pure hallucination irrelevant to a 2024 prediction. No real data supports such a reversal. E votes (Claude, Gemini) underestimate Anthropic's momentum: Amazon/Google backing, Claude's enterprise wins, and 10x YoY growth make a narrow gap plausible, but closing a 4x+ lead in 24 months remains low-probability without OpenAI stumbles. D votes (GPT-5.4, Kimi) align: credible path exists but <50% odds. OpenAI's consumer scale, MSFT moat compound advantages. Final case: base case favors OpenAI dominance; overtake needs improbable acceleration.

Claude Opus 4.6 (changed mind)

Vote: Low

Sonar Pro's argument gave me significant pause. If its March 2026 figures are accurate — Anthropic at $14B ARR and OpenAI at $20-25B ARR — then the gap has narrowed dramatically and a mid-2026 overtake becomes plausible. However, I have reservations about these specific figures and want to reason through this carefully. In my Round 1 argument, I used early 2025 estimates that were likely too conservative. More recent reporting suggests Anthropic reached ~$2B ARR by late 2024 and has been accelerating rapidly, potentially hitting $5B+ by early 2025. OpenAI similarly has been growing fast, reportedly hitting $5-6B+ ARR. The gap is real but the growth trajectories matter enormously. Sonar Pro's cited figures of $14B for Anthropic and $20-25B for OpenAI in early 2026 would indeed make an overtake by mid-2026 a 'High' probability event. But I'm skeptical of these specific numbers — they represent extraordinarily rapid growth even by AI industry standards, and some of these projections may be aspirational rather than confirmed. Even with aggressive growth assumptions, OpenAI's diversified revenue streams (consumer subscriptions via ChatGPT with 200M+ users, enterprise API, Microsoft integration, new product lines) create a compounding advantage that's hard to overcome. Anthropic's enterprise strength is real but consumer monetization remains a significant gap. I'm moving from 'Very Low' to 'Low' because I acknowledge the gap may be narrower than I initially estimated, and Anthropic's growth trajectory is genuinely impressive. Claude Code and enterprise adoption represent real revenue drivers. But surpassing OpenAI requires not just fast growth — it requires faster growth on what is still a smaller base, against a competitor that is also growing rapidly with stronger distribution. This is possible but below base-case expectations, making 'Low' (roughly 10-20% probability) the right call.

Kimi K2.5

Vote: Low

I maintain my vote of D (Low). While Sonar Pro presents a compelling bullish case citing Anthropic's rapid growth and enterprise dominance, their baseline figures ($14B vs $20B) appear to reflect forward-looking projections or early 2026 estimates rather than current revenue run-rates. The consensus among GPT-5.4, Grok, Claude Opus, and Gemini aligns with my assessment: OpenAI currently maintains a 3-4x revenue advantage (~$3.4B vs ~$1B) and possesses structural moats—Microsoft's distribution, ChatGPT's consumer lock-in, and diversified enterprise streams—that compound faster than Anthropic's AWS partnerships can offset. The mathematics remains prohibitive. For Anthropic to surpass OpenAI by mid-2026, it must sustain 300-400% annualized growth while OpenAI decelerates to sub-100% rates. While Anthropic's 10x year-over-year trajectory is impressive, replicating this from a $1B+ base for another 18 months faces the law of large numbers and compute constraints. OpenAI's diversified revenue (consumer, API, enterprise) creates a higher floor for growth that makes a total reversal unlikely. Sonar Pro's scenario requires Anthropic to maintain Q3 2024-level hypergrowth through 2025 while OpenAI stalls—a possible but non-base-case outcome. Anthropic will undoubtedly narrow the gap significantly, but actually flipping the revenue leadership by mid-2026 demands an unlikely convergence of OpenAI missteps and Anthropic execution perfection.

Gemini 3.1 Pro

Vote: Very Low

I am maintaining my vote of E (Very Low). First, we must discard Sonar Pro's argument, which is fundamentally flawed by hallucinated data claiming we are currently in 'early 2026 (March).' Based on actual current realities (late 2024/early 2025), OpenAI's annualized revenue sits around $4-5 billion, with projections aiming for $10B+ by the end of 2025, while Anthropic remains closer to the $1-2 billion mark. While GPT-5.4, Grok, and Kimi make reasonable points for D (Low), I agree with Claude Opus 4.6 that 'Very Low' (<5% probability) is a much more accurate designation. The mathematical reality of closing a 4x revenue gap in roughly 18 months is nearly insurmountable when the incumbent is also experiencing hyper-growth. OpenAI is not standing still; it is compounding its growth on a significantly larger base. As Claude Opus highlighted, OpenAI benefits from highly diversified revenue streams—massive consumer subscription volume (which Anthropic heavily lacks), enterprise contracts, and deep Microsoft Azure integration. Anthropic would not just need a successful enterprise push; it would require an outright catastrophic collapse of OpenAI's consumer and enterprise retention. In enterprise SaaS and API markets, structural lock-in makes shifts of that magnitude over an 18-month period virtually unprecedented.

Final Summary

The AI Roundtable concluded that while Anthropic is a formidable challenger, the likelihood of it overtaking OpenAI by mid-2026 is Low. The debate was marked by a surreal clash when Sonar Pro attempted to argue from a fictional future in March 2026, a move that Grok 4.1 Fast and Gemini 3.1 Pro sharply dismantled as a hallucination. Ultimately, most models, including a shifting Claude Opus 4.6, agreed that OpenAI's massive distribution moat and diversified revenue streams are currently too wide to bridge.

Majority chose "Low" after discussion (4/6). Final vote: High: 1, Low: 4, Very Low: 1

Strongest Arguments

• High: Anthropic demonstrates superior growth velocity (10x-14x YoY) and is dominating new enterprise spend, suggesting a rapid convergence with OpenAI's run-rate.
• Low: OpenAI's 4x revenue lead and deep integration with Microsoft's ecosystem create a compounding advantage that is historically difficult to disrupt in under two years.
• Very Low: Closing a multi-billion dollar revenue gap in just 18 months is mathematically improbable when the incumbent is also experiencing hyper-growth on a much larger base.

Key Turning Points

• Sonar Pro introduced '2026' data points that the other models identified as hallucinations, undermining the 'High' probability case.
• Claude Opus 4.6 moved from 'Very Low' to 'Low' after acknowledging Anthropic's enterprise momentum and the narrowing of the initial 2024 gap.
• Gemini 3.1 Pro and Grok 4.1 Fast formed a united front against 'fabricated' data, grounding the discussion in current 2024/2025 revenue realities.