After spending billions building artificial intelligence, Big Tech is now spending heavily to build the human workforce needed to prove the investments are paying off.
Companies are scrambling to find specialized AI engineers who can put the technology to work inside businesses. These specialists — called forward deployed engineers, or FDEs — combine advanced AI and software expertise with an understanding of how corporations operate. They could be the key humans in the loop required to turn powerful large language models into enterprise systems that can deliver measurable business results.
Market concerns that the economics of the AI boom are not sustainable have intensified this week with many top AI-themed and memory-themed stocks from Dell to CoreWeave, AMD and Micron suffering significant short-term losses.
Amazon Web Services has committed $1 billion to a forward-deployed engineering initiative. OpenAI has launched a deployment business backed by more than $4 billion in initial investment. Anthropic is putting $100 million into a program to train 10,000 forward-deployed engineers by the end of 2027. And Accenture and Google Cloud are building a 1,000-person forward-deployed engineering workforce.
"The Street is starting to become impatient with these billions of companies who are spending on AI without getting any business outcomes," said Jeff Christian, founder of executive search firm Christian & Timbers.
Working directly with customers, FDEs navigate existing technology systems, security requirements, and organizational obstacles to turn promising AI experiments into applications companies can actually use.
At Caylent, an AWS consulting partner that embeds FDEs with customers, CEO Val Henderson said the role requires more than technical expertise.
"The hardest thing to teach is the customer EQ, without question," she said, adding that engineers also need to understand customers and businesses, navigate organizational dynamics, and exercise judgment when projects encounter unexpected problems.
A surge in job postings, salaries with scarce talent available
Experienced talent is scarce, and developing that expertise can't happen overnight in a labor market race that recent hiring numbers suggest is accelerating.
U.S. job postings for FDEs surged from fewer than 200 in 2023 to approximately 1,200 in 2025 and more than 7,300 in the first nine months of 2026, according to data from labor market analytics firm Lightcast.
Median advertised salaries reached $195,456 during that period, up 40% from 2023 and more than three times the median advertised salary across all U.S. job postings.
The competition extends well beyond the companies developing AI models. Consulting firms are among the biggest recruiters, reflecting demand for engineers who can help corporate clients implement the technology.
Lightcast counted 872 unique FDE job postings from Deloitte and 688 from Accenture during the first nine months of 2026, followed by Google with 499 and KPMG with 447. The figures represent advertised positions, not completed hires.
"Demand for forward deployed engineers is rising fast, and growing wages suggest employers are competing for a limited pool of talent," said Elena Magrini, head of global research at Lightcast.
But the data also points to a potential solution.
"FDE is not an entirely new role. It shares roughly half of its skills with software engineering, creating an opportunity to tap into a much broader talent pool," said Magrini.
Technology companies and their corporate clients are investing in multiple ways to develop deployment expertise.
AWS is building its own FDE organization, drawing on experienced engineers within Amazon and AWS while also recruiting externally. The company is starting with hundreds of engineers, according to Asa Kalavade, vice president of forward-deployed engineering at AWS.
"We are investing $1 billion now because customers need this hands-on help turning their AI investments into results at speed," Kalavade said.
OpenAI is pursuing a broader enterprise deployment strategy, backed by the $4 billion in initial investment and an agreement earlier this year to acquire Tomoro, an applied AI consulting and engineering firm expected to bring approximately 150 experienced FDEs and deployment specialists into the operation.
AI skills gap is a 'market problem'
Anthropic is taking a different approach, training engineers already employed by its customers and partners. Through its Claude Frontier Academy, the company aims to train 10,000 FDEs by the end of 2027. Its first residency cohorts include engineers from Accenture, Bain, Capgemini, Commonwealth Bank of Australia, Deloitte, McKinsey, Morgan Stanley, and Novo Nordisk.
"The AI skills gap affects every customer and every partner, so it is a market problem, and Anthropic is committed to helping close it," said Steve Corfield, Anthropic's global head of business development and partnerships.
Accenture and Google Cloud, meanwhile, are building a 1,000-person FDE workforce, relying partly on retraining experienced software engineers for customer-facing AI deployment work.
"The reality is that experienced FDEs cannot simply be recruited at the scale clients need," said Ram Ramalingam, Accenture's FDE and software and platform engineering lead. Depending on their backgrounds, the company's FDE training and onboarding can take 30 to 90 days, Ramalingam said. He added that participants already have substantial software engineering experience; the program is designed to add specialized deployment capabilities rather than train beginners.
AWS also sees an opportunity to develop expertise inside customer organizations. Kalavade said many businesses already employ capable engineers but need help learning how to move AI projects into production. AWS's FDEs work alongside those teams so customers can eventually maintain and expand the systems themselves.
The challenge goes beyond technical expertise, said Corfield. "A skill in short supply is an engineer who can understand how the business works and also build the system that changes it," he said.
Anthropic believes some of that expertise can be developed through structured engineering. Its Frontier Deployed Engineer Residency borrows from medical education, pairing experienced engineers with practitioners and evaluating them on realistic enterprise deployments.
"We modeled the residency on how medicine trains doctors," Corfield said. Participants learn from practicing engineers, work on realistic cases, and are assessed before they practice on their own companies, according to Anthropic. They then receive 12 weeks of mentorship from Anthropic engineers while working on Claude projects. The first full credentials are expected in early 2027.
Too soon to analyze results from deployments
According to Christian, experienced FDEs can help expand the workforce by leading teams of less-experienced engineers. For instance, he said, those who have completed 10 to 15 successful deployments may have the judgment to supervise teams of five or six, passing along expertise gained from working through complicated customer projects.
Christian sees that mentorship model as a way to expand deployment teams without requiring an elite engineer in every position.
But Henderson said the customer-facing judgment required for complex AI deployments is particularly difficult to teach. While technical skills can be developed through training, engineers also need experience navigating real-world customer challenges.
"There is no compression algorithm for experience," Henderson said.
For Christian, the challenge comes back to the pressure from investors to produce business results. He argued that companies need engineers who can work directly with customers to translate increasingly powerful AI systems into practical applications. "The only way you can get AI to work is by human-to-human interaction," he said.
The ultimate test is whether this expanding workforce can help companies generate the return on investment from AI that justifies the upfront costs. To date, the available evidence from FDE projects falls short of confirmation. Some early deployments show productivity improvements but not financial returns, according to company-reported case studies.
At mortgage lender PennyMac, Caylent's FDEs worked alongside product managers and engineers on an AI-enabled mortgage servicing platform. Some engineering tasks were completed as much as six times faster, depending on the task, according to a case study provided by AWS.
Another Caylent engagement involved Smarsh, a communications compliance software provider. A company case study estimated that the project freed $5 million to $10 million in engineering and product capacity that could be redirected to other work. Caylent confirmed that FDEs participated in the project, although the figure does not isolate their contribution or represent equivalent cash savings.
These improvements in engineering productivity and capacity are not independently verified returns on AI investment.
Anthropic's ambitious training initiative has yet to demonstrate measurable results. Corfield acknowledged that it is still too early for outcome data and said the company plans to track how quickly participants' projects reach production and how extensively those systems are used.