ARK Invest, led by Cathie Wood, released its weekly market report on Aug. 23, laying out three themes focused on artificial intelligence and medical diagnostics: the revenue growth of Anthropic and OpenAI, the lower-cost positioning of Grok 4.6, and the continued expansion of minimal residual disease, or MRD, testing.
Anthropic and OpenAI post rapid ARR growth
ARK said AI agents should keep driving development at Anthropic and OpenAI. The report said Anthropic filed a draft S-1 with the U.S. Securities and Exchange Commission on June 1 as it prepares for a potential initial public offering, and that it has been in contact with investors to test market sentiment.
As of the end of May, Anthropic’s annual recurring revenue, or ARR, had reached $47 billion, up from about $9 billion at the start of 2026, a gain of more than fivefold in five months. Data provider TickerTrends estimated that Anthropic’s current ARR may already exceed $74 billion.
OpenAI’s ARR was put at about $41 billion, roughly double the $20 billion level at the beginning of the year. ARK said the two companies together now generate more than $115 billion in ARR, topping the combined trailing 12-month revenue of SAP, Salesforce, and Adobe, and approaching the roughly $150 billion annualized run rate of Microsoft’s productivity and business processes division.
According to the report, both companies plan to use public-market fundraising to support large-scale compute buildouts.
Grok 4.6 narrows the gap at a lower price
ARK said Grok 4.6 may keep pushing down the cost curve for frontier AI. The report described SpaceXAI’s Grok 4.6 as a model built for coding, agent tasks, and knowledge work, with a 500,000-token context window and pricing of $2 per million input tokens and $6 per million output tokens.
For comparison, OpenAI GPT-5.6 Sol was listed at $5 per million input tokens and $30 per million output tokens, while Anthropic Claude Fable 5 was listed at $10 and $50. ARK said Grok 4.6 is priced far below both models.
Artificial Analysis gave Grok 4.6 an Intelligence Index score of 61, matching GPT-5.6 Sol and trailing Claude Opus 5 and Fable 5 by only about 1 to 2 points. On task cost, ARK said Grok 4.6 comes in at about $0.84 per task, placing it on the intelligence-cost Pareto frontier. In AA-Briefcase’s long-horizon agent knowledge-work Elo rankings, Grok 4.6 scored 1577, essentially in line with Claude Fable 5 at 1574.
The report also said SpaceXAI has launched Grok Bot, extending competition beyond model performance into the agent software layer for long-running tasks. ARK argued that as model capability improves, the more relevant cost metric should shift from price per token to price per completed task. If training and inference costs keep falling by 85% and 99.9% per year, respectively, enterprises are likely to deploy agents across more workflows, accelerating AI adoption.
MRD testing shows clinical utility and commercial scale
On healthcare diagnostics, ARK said MRD testing has demonstrated utility and continues to scale. The report cited the IMvigor010 and IMvigor011 bladder cancer trials as key evidence, saying ctDNA-positive patients who received atezolizumab saw a 50% improvement in median overall survival.
ARK said Natera sold 283,000 Signatera tests last quarter, up by 34,000 from the prior quarter. All other solid-tumor MRD tests combined totaled only 41,000, implying that Natera held about 87% of the solid-tumor MRD market on the figures cited in the report.
The report said Signatera could reach about $1.5 billion in revenue by year five, roughly double Cologuard at the same stage. It also put the consensus market opportunity at about $20 billion, with cancer survivor monitoring, advanced-treatment monitoring, CAR-T, and global expansion seen as sources of additional upside in scale.