Microsoft is taking a deeper step into the use of artificial intelligence across its legal operations, seeking a senior lawyer with technical expertise to build AI-powered tools, develop automated workflows, and train attorneys to use the technology as the software giant looks to reshape how its legal department handles routine work.

The company is seeking a principal legal engineer for its Customer & Partner Solutions group, which supports Microsoft’s commercial business. According to the job description, the successful candidate will build AI agents, develop and refine prompts, and work directly with Microsoft’s lawyers and paralegals to integrate AI into their day-to-day operations.

“This is a role for a builder,” Microsoft said in the posting. “You will ship real capability rather than advise from the sidelines, partner directly with engineering, and stay close enough to attorneys and paralegals to design with them rather than for them.”

The position offers a clear indication of how the legal engineer role is moving beyond technology companies that sell software to law firms and into the legal departments of major corporations themselves. Rather than treating AI as a tool that lawyers can use independently, Microsoft appears to be seeking a specialist who can redesign legal workflows around the technology.

The shift comes as corporate legal departments face growing pressure to handle more work internally while controlling the cost of outside counsel. AI tools can assist with tasks such as contract review, drafting, and policy development, potentially allowing in-house lawyers to complete more preliminary work before sending complex matters to external law firms.

For Microsoft, that could have a significant financial and operational impact given the scale of its commercial operations and the volume of contracts and legal documents handled by its lawyers.

The company has already begun integrating AI into its legal work. Microsoft said in a blog post that its lawyers use Copilot, its flagship AI assistant, to accelerate routine tasks. In contract reviews, for example, Copilot can compare a draft agreement with similar contracts, identify potential issues, and recommend changes consistent with Microsoft’s internal policies.

Microsoft has also started deploying Harvey across its Corporate, External, and Legal Affairs organization. Harvey is an AI platform designed specifically for legal work, and its adoption by Microsoft places the company among major businesses seeking to integrate specialized AI systems into professional workflows.

The new legal engineer would sit at the intersection of law, software engineering, and AI. Microsoft is seeking an individual with a law degree and at least seven years of experience as a practicing attorney, while hands-on experience with AI legal tools such as Harvey or Copilot is preferred.

The position carries a base salary ranging from $147,000 to $278,900, according to the job posting.

The emergence of the role is notable because legal departments have historically been cautious about adopting technologies that could produce inaccurate or incomplete results. Legal work is particularly sensitive to errors because a fabricated case, incorrect contractual provision, or overlooked obligation can create financial and regulatory exposure.

AI’s growing presence in legal work is beginning to change that calculation. Clients are increasingly pressing law firms to deploy tools such as Harvey and Anthropic’s Claude for Legal to accelerate research and drafting, with the expectation that greater efficiency will eventually translate into lower legal costs.

That pressure is now moving inside corporations.

The emerging model is less about replacing lawyers and more about changing where lawyers spend their time. AI can handle some of the repetitive first-pass work, while attorneys focus on judgment-intensive tasks, negotiations, strategy, and matters where legal risk is higher. The legal engineer could become an important part of that transition. Unlike a conventional technology specialist, the role requires an understanding of how lawyers actually work. Unlike a traditional attorney, the position requires the ability to build and customize technical systems.

Microsoft’s emphasis on a “builder” rather than an adviser is particularly revealing. The company is not simply looking for someone to recommend AI products. It wants someone capable of turning those recommendations into working systems and embedding them directly into legal workflows.

The development also mirrors a trend among companies that build legal technology. Startups have increasingly hired legal engineers, often people with legal backgrounds and technical skills, to work closely with lawyers and translate their needs into software products.

Palantir, for example, recently advertised an embedded legal engineer position. The role involves working directly with the company’s lawyers to “transform high-touch, manual workflows into scalable, automated solutions.”

Microsoft’s move suggests that the same concept is now gaining traction among large corporate legal departments that are building internal AI capabilities rather than relying solely on external legal-tech vendors.

There is also a broader implication. Microsoft is simultaneously an AI developer, cloud provider, and enterprise software company, giving its legal department access to much of the infrastructure required to experiment with AI internally. Its lawyers can use Microsoft’s own Copilot ecosystem while also deploying external specialized tools such as Harvey. That creates an opportunity to develop a legal AI stack tailored to the company’s own requirements, potentially connecting contract databases, internal policies, legal knowledge, and workflow automation.

The challenge will be ensuring that greater automation does not introduce new forms of legal risk. AI-generated work still requires human oversight, particularly when the consequences of an error can involve contracts worth millions or billions of dollars, regulatory obligations, or litigation.

OpenAI’s preparation for a public-market debut in 2027 could be accelerated if the ChatGPT maker’s business continues to grow rapidly, Chief Financial Officer Sarah Friar told employees on Wednesday.

“The IPO is not a finish line, it is a milestone, another fundraise,” Friar said during an all-hands meeting, according to two people familiar with her comments quoted by CNBC. “We raised $122 billion in March, and that gives us flexibility.”

The comments provide OpenAI’s clearest indication yet of its expected path to the public markets. The company confidentially filed its IPO prospectus with the U.S. Securities and Exchange Commission in June, but has not committed publicly to a listing date.

Friar’s 2027 target also signals that OpenAI is not prepared to rush into an IPO simply because rival Anthropic may reach the market first.

Anthropic has also confidentially filed IPO documents and has begun preliminary discussions with prospective investors. Friar told OpenAI employees that Anthropic could make its confidential filing public as soon as September, but said that would not alter OpenAI’s plans.

“As you know we are confidentially under file, and Anthropic is also under file,” Friar said. “There is a chance they pull the cover off that confidential file in the coming weeks and become public in September. That’s OK, we are running our own race.”

The timing matters because OpenAI is approaching the public markets with an extraordinary valuation to justify. The company is valued at about $852 billion, meaning investors will expect evidence that its explosive revenue growth can eventually translate into sustainable profits and cash flow.

OpenAI is now giving investors several indicators designed to support that case.

Friar told employees that OpenAI’s overall revenue run rate is up 35% quarter to date, while its enterprise revenue run rate has increased 50%. The company’s AI coding and work products have reached 20 million weekly active users, according to slides presented during the meeting.

OpenAI generated $6.7 billion in revenue in the second quarter, according to a Wall Street Journal report cited in the information provided, an 18% increase from the first quarter. Its annualized revenue run rate has since surpassed $40 billion.

The scale of that growth is significant, but it also highlights the challenge facing the company. OpenAI must sustain extraordinary expansion while spending heavily on computing infrastructure, model development and talent.

The comparison with Anthropic makes that challenge more pronounced.

Anthropic’s annualized revenue run rate reached $65 billion at the end of July, according to investor figures reported over the weekend, while its preliminary second-quarter revenue was $11.5 billion. The company has therefore emerged as a formidable commercial competitor even as OpenAI remains the larger and more established consumer AI platform.

The race is increasingly moving beyond chatbot popularity toward enterprise contracts, coding, AI agents and other commercial applications that can generate recurring revenue.

Enterprise adoption has become necessary for OpenAI. Consumer subscriptions helped establish ChatGPT as a mass-market product, but corporate customers offer the potential for larger and more predictable spending. The 50% increase in enterprise revenue run rate cited by Friar suggests the company is attempting to deepen that part of its business ahead of a potential listing.

The 20 million weekly active users for coding and workplace products point in the same direction. OpenAI is trying to turn its models from general-purpose chatbots into infrastructure embedded in the daily operations of companies and professionals.

That expansion could help support the valuation, but it does not eliminate the costs of serving those users.

The economics of AI remain heavily dependent on computing capacity. OpenAI has been committing enormous sums to data centers, chips and cloud infrastructure as it seeks to support growing demand. Its March fundraising, which Friar said reached $122 billion, gives the company substantial financial flexibility, reducing the immediate need to tap public markets for capital.

That is one reason the IPO can be treated as a strategic milestone rather than an urgent financing event.

Still, a public listing would provide OpenAI with access to a much broader pool of capital at a time when the AI industry is entering a capital-intensive phase. It would also give existing investors and employees a potential mechanism for realizing returns on their holdings.

The public markets, however, will impose a level of financial scrutiny that OpenAI has not previously faced.

Investors will want to know not only how quickly revenue is growing, but how much the company spends to generate that revenue, how quickly inference costs are declining, what proportion of revenue comes from a small number of large customers, and whether AI models can eventually generate margins comparable with conventional software businesses.

Competition is another major issue.

OpenAI and Anthropic are facing pressure from lower-cost open-weight models developed by companies including Alibaba, DeepSeek, and other Chinese AI firms, as well as Meta and other U.S. technology companies. If increasingly capable models become available at lower prices, OpenAI could face pressure to reduce prices even as its own computing costs remain substantial.

The competitive environment also extends beyond model quality. Google’s Gemini, Meta’s open-weight models and a growing ecosystem of specialized AI systems are competing for developers, enterprise customers and computing workloads.

That makes OpenAI’s revenue growth particularly important ahead of an IPO. A $40 billion annualized revenue base would represent a substantial commercial business, but sustaining rapid growth while defending market share will be central to whether investors accept the company’s enormous valuation.

There is also an internal stability issue.

OpenAI has experienced a series of high-profile executive departures in recent months. Revenue chief Denise Dresser left after about eight months at the company, while Brad Lightcap, a longtime executive, announced that he was ending an eight-year tenure. Fidji Simo stepped down from her product business role in July to focus on her health.

The departures have raised concerns among some investors about management continuity just as OpenAI prepares for greater scrutiny.

President Greg Brockman has pushed back against that interpretation, explaining that OpenAI’s unusual visibility makes executive departures appear more dramatic.

“I actually think that the difference between OpenAI and other organizations is that we are so much in the spotlight, so every departure gets scrutinized in a way that it doesn’t otherwise,” Brockman said in an interview.

However, investors’ concern has been largely about whether OpenAI can demonstrate institutional stability alongside rapid growth.

The IPO race with Anthropic is therefore only one part of a much larger contest. Both companies are trying to convince public-market investors that the extraordinary spending required to build frontier AI systems can ultimately produce durable and highly profitable businesses.

OpenAI has the advantage of scale, brand recognition, and a massive user base. Anthropic is growing rapidly and has demonstrated particularly strong traction with businesses and developers. The decision to target 2027 rather than rush to market could give OpenAI more time to strengthen its financial profile, expand enterprise adoption and demonstrate that its revenue growth is translating into improving economics.

But the delay also raises the stakes. By the time OpenAI becomes public, investors will expect considerably more than evidence that people want to use ChatGPT. They will want proof that the company can convert AI’s extraordinary demand into a business capable of supporting an $852 billion valuation.

The IPO, as Friar put it, may not be the finish line. It could become the moment when the market begins measuring OpenAI by a different standard: not simply how quickly it can build the world’s most widely used AI products, but how efficiently and profitably it can turn that technological lead into a lasting enterprise.