In 2023, analysts estimated Microsoft’s Copilot AI chatbot would garner $30 billion in revenue. The software giant did not reveal Copilot revenue in its latest earnings call even as OpenAI reports $10 billion in annual recurring revenue.

In the last two years, Microsoft — with $79.6 billion in cash and a $3.65 trillion market capitalization, according to Google Finance — seems to have squandered a huge growth opportunity in the market for AI chatbots.

Meanwhile startups like OpenAI and Anthropic are scaling their AI chatbot revenue at astonishing rates.

Can Microsoft — which in 2023 was expected to achieve $30 billion in Copilot revenue, noted my October 2023 Forbes post — catch up and overtake these startups to dominate the AI chatbot market?

Although Microsoft has considerable resources — such as a huge cloud services business and a massive customer base — I think the answer is probably no.

Here are my reasons:

To be fair, Microsoft’s business is performing well — even as the company cut over 9,000 employees, according to CNN. Shares of Microsoft — which reported $70 billion in March quarter revenue and a 37% net margin, noted CNBC — have risen 16.5% in 2025.

Moreover, the company is excited about Copilot’s performance and potential. By late 2024, “nearly 70% of Fortune 500 companies” had adopted the tool, reported The Globe And Mail.

Companies implementing Copilot and other generative AI technologies experienced an “average return on investment of $3.70 for every dollar spent, with leading organizations achieving up to a 10.3x ROI,” according to a Microsoft-sponsored IDC report.

Yet many companies are hesitating. Although 80% of organizations were testing Copilot, “only 16% had transitioned to full production, citing concerns over security, ROI, and implementation complexities,” according to a Gartner survey featured by The Globe And Mail.

I have requested comment from Microsoft and will update this post if I receive a response.

Not so long ago, Microsoft appeared poised to take a large share of the market for AI chatbots.

How so? Aided by the software giant’s 49% stake in OpenAI and the company’s dominance in the highly profitable software industry, I estimated Microsoft’s Copilot revenue could hit $30 billion, noted my October 2023 Forbes post.

My estimate was based on my own twist on two analyst reports. A March 2023 Credit Suisse report estimated Microsoft would achieve $40 billion in Copilot revenue by 2028, according to Business Insider.

A July 2023 Macquarie Equity Research report offered a lower estimate of $14 billion — based on the assumption that 10% of Microsoft’s 382 million customers would pay Copilot’s $30 per user monthly fee.

My estimate of $30 billion was based on what now seems to have been an unrealistic assumption — that Microsoft would raise its monthly fee to $60 a month. How so? I thought Microsoft would want to cover more of its $80 per month cost to deliver the Copilot service, which the Wall Street Journal reported.

Perhaps these high costs are prompting rounds of layoffs at Microsoft. The company — which cut 10,000 employees in 2023 — announced it would reduce its workforce by about 9,000, less than 4% of the total, noted CNN.

Microsoft sees the cuts as helping the company adapt to change. “We continue to implement organizational changes necessary to best position the company and teams for success in a dynamic marketplace,” a Microsoft spokesperson told CNN.

It is unclear how much revenue Copilot contributes to the company. In Microsoft’s most recent quarterly report, there was no Copilot revenue figure. I am guessing if the figure was in the tens of billions of dollars and growing rapidly, the company might want investors to know.

While Microsoft supplies revenue growth for many of its business lines — ranging from LinkedIn to Azure Cloud Services — the company studiously avoids reporting revenue for all but the broadest categories of products — such as More Personal Computing.

Microsoft did not respond to my request for comment on this topic.

Business customers — including me — prefer ChatGPT and other AI chatbots to Copilot. Two years ago, business customers were assessing Copilot and questioning whether the additional cost was offset by the tool’s business benefits, according to my book Brain Rush.

Last August, a pharmaceutical company stopped used Copilot citing high cost and low value. The pharma executive canceled a $180,000 trial for 500 employees after he compared Copilot’s slide generation skills to that of a “middle school student,” according to a Morgan Stanley research note featured by Business Insider.

As a long-time user of Copilot I view it is a more intrusive version of Microsoft’s Clippy. Copilot appears unhelpfully as I read and respond to emails and sometimes creepily uses content residing in my Microsoft Outlook as sources for responses to Copilot queries.

Copilot continued to struggle to win users because corporate employees prefer ChatGPT. Even businesses which have purchased Copilot struggle to convince users to switch from ChatGPT — their at home AI chatbot, noted a Bloomberg report featured by TechRadar.

Although Copilot uses ChatGPT’s technology, OpenAI’s momentum and existing user base seems to be giving ChatGPT the edge. As of June 2025, ChatGPT had about 800 million weekly active users and three million paying business users, “whereas Copilot has stagnated a little, with 20 million weekly users for the past year,” reported TechRadar.

While Microsoft has declined to report its Copilot revenue, privately-held rivals are boasting about their revenue and growth.

For example, Open AI reported annual recurring revenue of $10 billion last month. After launching ChatGPT about two and a half years ago, the $10 billion figure — 82% more than 2024’s $5.5 billion ARR — includes sales from the company’s consumer products; ChatGPT business products; and its application programming interface — while excluding licensing revenue from Microsoft, noted CNBC.

OpenAI’s growth potential contributes to a very high valuation of about 30 times revenue. In March, the company said it had 500 million weekly active users and paying business users had increased 50% to three million since February, CNBC wrote.

Meanwhile, Anthropic — provider of the Claude AI chatbot — has reportedly reached $4 billion in ARR, according to a report from The Information. This represents “a fourfold increase in revenue since the beginning of the year,” noted OpenTools.

This rapid growth reflects the popularity of Anthropic’s Claude models. These are known for their conversational AI excellence which makes them “powerful tools for automating customer service and other text-based tasks,” reported OpenTools.

Why does Microsoft’s Copilot appear to be stagnating while OpenAI and Claude sprint forward?

My guess is there are two reasons. First, Microsoft and OpenAI are struggling to maintain their partnership from breaking up. OpenAI could lose a $20 billion investment unless Microsoft approves OpenAI’s conversion to a for-profit company, reported the Wall Street Journal.

OpenAI’s deal gives Microsoft first access to startup’s latest technology. However, the two firms are now at odds — with OpenAI seeking to pay $3 billion to buy Windsurf, an AI coding startup that competes with Microsoft’s GitHub Copilot, noted the Journal.

The companies are optimistic they can resolve their differences. “We have a long-term, productive partnership that has delivered amazing AI tools for everyone,” representatives for the two companies said in a joint statement to the Journal. “Talks are ongoing and we are optimistic we will continue to build together for years to come.”

But these partnership problems make me question whether Microsoft may be suffering from the same innovation disease that afflicted IBM for many years. As I wrote in a November 2016 Forbes post, IBM stopped developing products that provided more benefits to end users than competing products did.

For example, IBM imposed numerous internal requirements on its product development teams – such as requiring that products can be usable by vision-impaired individuals regardless of whether there is strong demand for the feature and that the product is available in at least nine languages without regard to market requirements, I wrote.

These executive-imposed mandates could consume as much as half of product development teams’ budgets and time. As a result, too many IBM products became available to customers after rival vendors have introduced their products – and those late products offered customers insufficient benefits to compel them to switch from rivals’ products to IBM’s, noted my post.

I wonder whether a similar mindset is keeping Microsoft from delivering a superior value proposition to AI chatbot end-users. ChatGPT, Claude and Perplexity seem designed to provide a huge increase in value and productivity.

Most notably, I appreciate the deep research capabilities of ChatGPT, Claude and Perplexity to enable me to brainstorm solutions to challenging problems of growth strategy.

By contrast, Copilot seems constrained in what it can do by the need for it to be integrated with Microsoft 365’s many business software applications. Copilot seems designed to win over IT executives and to boost the software giant’s revenue per customer, rather than winning over the people who actually use AI chatbots.

This makes me wonder whether Microsoft will achieve the revenue boost the company envisioned from its heavy generative AI capital bets — including its $80 billion in AI data centers capital expenditures and $14 billion investment in OpenAI.