Why Big Tech is Spending Billions on Forward Deployed Engineers

For the past two years, Big Tech has been busy in the race to outcompete one another on frontier models. But a different trend is now taking shape. As frontier AI companies look to accelerate enterprise adoption, they’re moving closer to customers by building separate companies of forward-deployed engineers.

Within days of one another, both Amazon Web Services and Microsoft announced major investments in dedicated engineering organisations that will work directly with enterprise customers to deploy AI.

AWS is committing an initial $1 billion to a new Forward Deployed Engineering unit, while Microsoft has unveiled the Microsoft Frontier Company, backed by $2.5 billion and around 6,000 engineers to help enterprises build and run AI models. 

“Incredible goal to hire 6000 FDEs, Satya Nadella! There are about 7000 FDEs globally. Microsoft by itself will double this! There simply has not been a better time to be an FDE,” wrote Rajan Anandan, Managing Director at Peak XV Partners, in a post on X.

However, Microsoft and Amazon are not the first to deploy engineers. 

In April, Google Cloud unveiled a $750 million fund to expand its forward-deployed engineering capacity and support the deployment of agentic AI at customer sites.

Anthropic followed in May with a $1.5 billion joint venture with Blackstone and Goldman Sachs to embed Forward Deployed Engineers within enterprise organisations. Days later, OpenAI launched the OpenAI Deployment Company with an initial $4 billion investment backed by TPG, Advent, Bain, and Brookfield. 

The AI industry is now moving into the next phase of the race, deploying its products inside the world’s largest enterprises.

“The competitive advantage has shifted from building AI models to delivering measurable business outcomes. Most enterprises have moved beyond pilots and now demand production-grade AI integrated with business processes, enterprise data, and governance,” Arun Chandrasekaran, Distinguished VP Analyst at Gartner, tells AIM.

He explains that this requires engineers who can bridge product development with customer implementation. Once an AI model becomes deeply integrated with an organisation’s workflows, data infrastructure and internal applications, switching providers becomes significantly more difficult.

Successful deployments also increase cloud consumption, expand AI workloads, and create opportunities to sell additional enterprise software and services.

The Deployment Problem

Forward-deployed engineers, or FDEs, are not traditional consultants or sales engineers. FDEs work alongside customers, write production-grade code, integrate AI models with enterprise systems, work alongside security and engineering teams, and solve deployment challenges that generic software cannot.

Most large organisations already use LLMs to summarise documents, generate code, or for research. Their biggest challenge is connecting those models to decades-old enterprise systems, proprietary data, governance frameworks, and regulatory requirements.

Chandrasekaran says models are no longer the bottleneck. “The biggest hurdles include poor data quality, fragmented enterprise architectures, integration with legacy systems, AI governance and security requirements, and redesigning workflows,” he says.

Organisations also struggle to demonstrate measurable business value, establish clear operating models for AI, and manage organisational change. “Successful deployment requires combining AI engineering, domain expertise, and business transformation, all of which are far beyond just model development,” he suggests.

Holger Mueller, VP and Principal Analyst at Constellation Research, further notes that FDEs can help enterprises accelerate AI deployment. “Contrary to consultants—who can only customise and extend the product—FDEs can not only finish a product for an enterprise during an engagement but also have feedback capabilities for general development,” he says.

He adds that early adopters of new technologies often want the confidence that they’re deploying a mature, enterprise-grade product rather than one that’s still being built or customised.

That has also shot up the demand for forward-deployed engineers and similar customer-embedded AI engineering roles, which grew 42-fold between 2023 and 2025, according to a LinkedIn report.

New Challenge for IT Services

The rise of forward-deployed engineering also directly threatens the traditional IT services industry. For years, companies like TCS, Infosys, Wipro, Accenture, and Deloitte positioned themselves as the go-to partners for enterprises implementing large-scale technology projects. That is now changing.

Even former Infosys CEO Vishal Sikka recently launched Hang Ten Systems, backed by Mayfield and Saudi Aramco, with a similar focus on helping enterprises adopt and scale AI through FDEs.

Chandrasekaran says the relationship will evolve into both collaboration and competition. “AI vendors will increasingly own the high-value activities closest to the model, including solution design, reference architectures, and deployment of strategic use cases,” he says, adding that system integrators will continue to lead enterprise-scale transformation, process redesign, integration, governance, and managed operations.

“Over time, services firms will need to differentiate through industry expertise, multi-vendor AI orchestration, and long-term AI operations rather than implementation alone.”

Mueller says providing deployment services can reduce margins for software and LLM vendors, but enterprises prefer to rely on consulting and implementation partners post the initial deployment. He adds that while vendors may step in to accelerate early enterprise adoption or handle strategic use cases, they are generally happy to hand over the work to consulting firms as the ecosystem matures.

The trend could play to India’s strengths. The country’s deep engineering talent and long experience in enterprise software services make it a natural candidate to supply FDEs to global AI companies.

Harveen Singh Chadha, an engineer at Sarvam AI, believes the domain knowledge and relationships that Indian IT companies have in the retail, BFSI, and healthcare sectors can’t easily be cloned by any new-age company. 

“What they urgently need is to reskill a major chunk of their workforce as forward-deployed engineers and execute relentlessly,” he wrote on X.

“India could become the FDE factory for the world,” said Sanjeev Aggarwal, Co-founder of the VC firm The Fundamentum Partnership, in a recent interview with CNBC.

He added that India has a very strong training infrastructure. “We have a lot of domain expertise. These 20–30 years that IT services companies have been in business, they have learned all the different industries, whether it is telecom, financial services, or healthcare,” he said.

Similarly, Cisco President Jeetu Patel recently said that “every engineer will become a Forward Deployed Engineer,” describing it as a significant opportunity for India’s engineering talent.

In India, entry-level FDE positions are drawing packages in the range of ₹9–12 lakh per annum, according to data from staffing firm TeamLease Digital, as demand for FDEs rose nearly 800% over the last year alone.

Mid-level practitioners—those with demonstrated experience in production AI deployments and a track record with enterprise clients—are commanding ₹25–50 lakh, while senior-level FDEs boast packages of ₹50–80+ lakh per annum, according to data from ADaSci.

Humans + Tokens 

Behind FDEs lies the opportunity for businesses to create systems that allow human and AI capabilities to learn from one another over time.

“The real opportunity is not in picking the best model but instead in building a learning loop on top of models where human capital and token capital compound,” said Microsoft CEO Satya Nadella in his recent post on X.

“Our ambition is to help every enterprise build its own AI capability, and to help create a frontier ecosystem where every organisation can turn its knowledge, workflows, and judgement into its own AI systems that continuously improve,” he added.

Nadella argued that companies should be able to replace underlying AI models without losing institutional knowledge accumulated through their own systems. He said businesses need architectures that allow agentic systems to improve while maintaining control over intellectual property.

Chandrasekaran says forward-deployed engineering will become a standard capability for enterprise AI vendors, particularly those targeting large enterprise customers. “As AI solutions become more agentic and deeply embedded into enterprise operations, successful adoption will increasingly depend on co-innovation with customers.”