ElevenLabs and DXC begin joint voice AI projects as enterprise buyers demand proof of return

DXC Technology has invested in ElevenLabs, uses its voice technology internally and is now selling it to clients. The IT services company announced a strategic partnership with the London-headquartered voice AI firm in late July. It also said it had taken part in ElevenLabs' $500m Series D, which valued the company at about $11bn. DXC has not disclosed the size of its stake.

By mid-September, the two companies were drawing up a joint business plan. They have selected a small group of DXC client accounts for their first projects, which focus on customer experience and growth.

“We are very bullish on ElevenLabs because we are both a customer and an investor, and I would also describe us as a partner,” said Holly Grant, President of AI Innovation and Strategy and LabX at DXC Technology. “So really, we wear three hats in this relationship.”

Mati Staniszewski and Piotr Dąbkowski founded ElevenLabs in 2022. Staniszewski had been a deployment strategist at Palantir, and Dąbkowski a machine learning engineer at Google. The pair have said the idea came from watching badly dubbed American films as children in Poland. The company has offices in London, New York, San Francisco and Warsaw.

ElevenLabs started with text-to-speech and added dubbing, voice cloning, speech-to-text and music generation. Eleven v3, its current model, generates speech in more than 70 languages and offers a library of more than 10,000 voices. A lower-latency model responds in about 75 milliseconds, fast enough for a live phone call. ElevenAgents, its agents platform, lets companies build voice agents that connect to their own systems. Deutsche Telekom, Revolut and Klarna are among its named enterprise customers.

Annualised recurring revenue was about $120m at the end of 2024. It reached $330m by the end of 2025, grew by about $100m in the first quarter of 2026, and passed $500m by May. Staniszewski confirmed the $500m figure in a CNBC interview on 5 May.

A $19m Series A in June 2023 valued ElevenLabs at about $100m. The Series B in January 2024 took it to $1.1bn, and a $180m Series C in January 2025 to $3.3bn. Sequoia led the $500m Series D in February 2026 at $11bn, with Andreessen Horowitz quadrupling its investment and Iconiq tripling its own. By May, Nvidia's venture arm NVentures, BlackRock, Wellington Management, D.E. Shaw, Schroders and Salesforce had joined the round, along with more than 30 entertainment figures including Jamie Foxx and Eva Longoria. DXC came in as a strategic investor after the round first closed. The $11bn valuation is about 22 times ElevenLabs' current annualised revenue.

DXC is repositioning a $12.6bn services business around AI

DXC was formed in April 2017 through the merger of Computer Sciences Corporation and Hewlett Packard Enterprise's Enterprise Services unit. Headquartered in Ashburn, Virginia, it manages infrastructure, modernises applications and supplies industry software for large companies and public sector bodies in more than 70 countries.

Raul Fernandez became President and CEO in February 2024 and has split the company into two tracks. Core Track runs DXC's established services business. FastTrack houses new AI products, platforms and partnerships, including the ElevenLabs deal and OASIS, DXC's AI orchestration platform.

DXC reported revenue of $12.64bn for the fiscal year to March 2026. That was 4.8% lower on an organic basis, as enterprise IT budgets shift from traditional outsourcing towards cloud and AI. Profitability and cash generation held up over the same period. The adjusted EBIT margin was 7.7%, and free cash flow rose 3.8% to $713m. DXC has guided to revenue of $12.11bn to $12.35bn for the current year. DXC shares fell 21.5% on the first trading day after the full-year results in May, as investors weighed revenue against margins and cash flow.

Fernandez said at the time that DXC had “leaned into innovation to reposition DXC for the next phase of enterprise IT and AI-driven transformation, including the recent launch of our AI-based orchestration platform, OASIS, and continued progress across our Core Track and FastTrack initiatives.”

DXC's chief executive tested the technology on his own earnings calls

Fernandez trained an AI model on his own voice, and the cloned voice read his earnings transcript for three consecutive quarters. Grant called it an early “customer zero” example.

DXC is putting more weight on voice agents: conversational systems that talk to people, respond to what they need and hand the call to a human when required.

“Part of that conviction is really around the importance of voice as AI becomes a more natural interface in our day-to-day work and life,” Grant said. “We all know and work with AI in a text-based format. That's really where it began, when you were going back and forth with ChatGPT or Claude and having great conversations. I think the next frontier is this more fluid, natural way of conversing, like we are now.”

She said voice agents can “converse with individuals and really attune to their state, attune to their set of needs and respond in a very natural way, and of course escalate to the human beings who need to come into the conversation to take things further as required.”

“I think it dramatically changes customer experience,” she added. “I think it dramatically changes how AI can be both a thought partner in a live coaching session and a delegate to go and perform work on our behalf.”

“Voice is becoming a primary interface for how enterprises engage with customers and employees,” Fernandez said when the deal was announced. “Our partnership with ElevenLabs allows DXC to move faster in embedding AI into everything we do, from internal operations to the solutions we deliver for clients. This is a key step in accelerating our AI-native products and solutions.”

DXC's core business earns roughly $12bn a year and plans in annual cycles, like most large corporates. FastTrack, which includes LabX, DXC's AI-native product incubator, has its own governance and investment structure.

“If we're really going to innovate at the edge, use some of the frontier AI capabilities and discover new business models that we believe are required for the next era, we need a different structure around governance, investment and relationship building to govern that set of activity within DXC,” Grant explained.

“Our business planning cycles in the rest of DXC operate in what you would expect for a large corporate, right, years,” she said. “In FastTrack, we're talking about 90-day cycles, where we can actually adapt to wildly changing market conditions and wildly changing technology swings. That's a very different motion.”

Grant said the separate structure keeps small projects visible inside a $12bn business. “It's easy in a very large company for early signs of traction to be overlooked if we're using the typical planning and governance structures,” she noted.

Most enterprise AI pilots have failed to show a return, and buyers have noticed

Research published by MIT's NANDA initiative in August 2025 found that 95% of organisations had seen no measurable financial return from generative AI, despite $30bn to $40bn in enterprise spending. The study analysed 300 public deployments. Of the enterprise systems it tracked, 60% were evaluated, 20% reached pilot, and 5% reached production.

“There was just this frenzy of money and activity flowing into AI, and it almost didn't matter initially what the outcome was,” Grant said. “Now I think our buyers are becoming far more sophisticated, far more sceptical in a great sense. They're asking the right questions, which is: if I spend this money, if I spend this time, what is the return on the other side of that?”

DXC will not start a project until the client outcome is defined. “We always have to start with what the customer outcome is we're trying to solve for,” Grant said. “If we can't answer that question, it's actually not time to begin with that customer on an AI deployment of any kind.”

“Once we can answer it, then we can bring the best of governance, auditability, trust and the operational know-how that we have, because we've been operators of mission-critical systems for decades at this point,” she added. “That's where we can get very realistic about the constraints, about where things break down, and we need to show our customers proof that we're actually fine-tuning these systems in practice.”

“So the idea is, while we want to start fast, we want to start small and focused on an outcome,” she said. “We need to demonstrate an initial return, and then we build on that.”

DXC tests voice AI on its own workforce before clients see it

Voice AI raises questions about jobs, copyright, privacy and consent. ElevenLabs' tools have been used to imitate celebrities and public figures, and researchers linked a robocall impersonating President Joe Biden before the January 2024 New Hampshire primary to its technology. ElevenLabs has since tightened its voice cloning safeguards and released a classifier that detects audio made with its tools.

Grant did not address jobs or copyright directly. On governance, she pointed to DXC's practice of deploying each technology internally first, across a workforce of about 125,000 at its last public count. Russell Jukes, DXC's Chief Information and Digital Officer, acts as the internal buyer.

“Our Chief Information and Digital Officer, Russell Jukes, is essentially our de facto customer zero,” Grant said. “He works with us very closely to deploy a lot of technologies in very specific ways and make sure that we have a command of how to do that well, how to do it safely, how to do it in a trusted manner.”

“Then we can go to customers and say, look, we know we can deliver this. Look what it did in our own environment. So that's powerful,” she added.

Gartner expects AI agents to resolve 80% of common service issues by 2029

In March 2025, Gartner predicted that by 2029 agentic AI will resolve 80% of common customer service issues without human involvement, cutting operational costs by 30%. A February 2026 Gartner survey of 321 customer service leaders found 91% under pressure to implement AI this year.

In an April 2026 release, Gartner reported that 85% of service leaders were expanding the responsibilities of human agents. Some 31% had carried out or planned AI-driven layoffs through the first quarter of 2027, and 63% were reducing frontline headcount gradually through attrition. In a separate Gartner survey of 5,801 US consumers, 54% said they trust human agents more than AI for recommendations, against 32% who trust AI more.

“As human beings, we spend time trying to get certain things done on the phone with systems that don't understand us,” Grant said. “Those systems can be fully human, or some combination of technology-enabled and human, and often it's a really bad experience for everybody. You can spend a lot of time and not actually get anything done.”

“What I've seen through these voice agents is a dramatic shift in speed to resolution and in overall customer satisfaction,” she added. “I think we can turn people into believers that this combination of humans and voice agents is the way to go.”

Grant also expects multilingual support to widen access to services. “I think of the potential for all of the different languages that can be supported, so that more people can get the help they need in the languages they need it,” she said. “I love to travel, so I'm very excited as I extrapolate out what it means for having more meaningful interactions more broadly.”

Under the partnership, DXC will deploy voice agents and copilots for service desks, training and knowledge management, and build multilingual voice interfaces into customer service platforms. It will also add voice AI to its industry offerings and combine ElevenLabs' models with its application modernisation and orchestration platforms.

ElevenLabs gains access to regulated enterprises, and DXC gains a stake in fast-growing voice AI

ElevenLabs has the model, the capital and revenue that grew by about $170m in the first four months of 2026. Banks, insurers, airlines and government agencies are harder to reach. They run complex legacy systems and have strict procurement rules, and voice agents deployed there must work with call centre software, customer records and compliance requirements. DXC has run those systems for decades and holds the client relationships. The MIT research found that AI tools bought from specialist vendors and deployed with partners reached production more often than tools companies built themselves.

For DXC, the partnership adds a fast-growing product line alongside a core business that is adjusting to changes in enterprise IT spending. The stake gives it a share in one of the fastest-growing AI companies. Internal deployment gives it evidence to show clients before they buy. For ElevenLabs, DXC's accounts across more than 70 countries provide a sales channel that would take years to build alone.

“DXC operates at global enterprise scale across some of the world's most complex environments,” said Ben Budde, Revenue Leader at ElevenLabs. “Partnering with DXC allows us to bring our voice AI technology into mission-critical workflows, unlocking new possibilities for automation, accessibility, and how people work with technology.”

IT services companies worldwide are moving from billing for people-hours towards selling AI products with measurable outcomes. DXC has chosen to invest in, deploy and resell the same company's technology, a combination few of its peers have tried. Its results will show whether AI partnerships can build a second source of growth next to an established services business.

DXC expects to publish before-and-after case studies from the first client projects

LabX and DXC's AI Platforms Engine will co-develop solutions with ElevenLabs for clients worldwide, starting with the accounts selected for the first engagements.

“We are working very closely with our ElevenLabs counterparts to establish a joint business plan and conduct our first few engagements with specific accounts that we've selected,” Grant said. “We believe the outcomes where ElevenLabs' technology and DXC's know-how really move the needle are in the realm of customer experience and growth.”

“In the coming months, I hope we can come back and report real case studies with these customers, where we've brought a solution to bear that's very outcome-focused, to give the before and after comparison,” she said. “Here's what this process was like before we brought this human-agentic solution. Here's what it looks like after.”

“The whole ethos is start small, start focused, demonstrate success, tell the stories, and then expand from there,” she added. “That's really what to expect.”