AI buyers need to plan beyond forward-deployed engineers

FDE growth makes delivery ownership a buying decision

Enterprise buyers can end up paying an ongoing engineering premium for AI deployment work that eventually becomes suitable for conventional services. Forward-deployed engineers, or FDEs, work directly with customers to put technology into use, helping address data safety concerns, rising costs and talent shortages that obstruct AI adoption. Their value depends on the assignment’s difficulty and the expertise they bring. So buyers need to decide which work requires specialized support and who should own delivery as the assignment becomes clearer.

The stakes could be substantial. Gartner forecasts that roughly 70% of vendor-led AI engineering engagements will be abandoned by 2028, drawing on IT services company data, client consultations and proprietary modeling. This is expected abandonment; it neither measures an observed failure rate nor establishes that abandoned engagements deliver no value. Mukul Saha, Gartner senior director analyst, also sees good outcomes from well-run FDE engagements.

Saha’s assessment comes from a company with a commercial interest in demand for technology advice. Still, his argument gives buyers two specific questions to examine: whether the assignment warrants an FDE team and whether the provider can supply the required expertise. The second matters because vendors are expanding their deployment commitments while experienced engineers are already scarce.

Deployment commitments face a limited supply of engineers

Saha estimated that there were 2,000 active FDEs at the time of reporting. “Demand is probably quadruple that, so there aren’t enough engineering resources available,” he told Channel Dive. The shortage makes access to a suitably skilled team a practical buying constraint. A provider’s investment announcement describes its commitment, while the people available for an engagement determine what it can deliver.

Google Cloud’s recruitment effort shows how large vendors are competing for those people. In May, it advertised 59 distinct FDE roles across the U.S. and international locations. Google Cloud CEO Thomas Kurian promoted the opportunity on LinkedIn, inviting builders to work at global scale in what he called the “agentic era,” referring to AI systems that can carry out tasks. Hiring was still needed to turn those advertised positions into delivery capacity.

Other cloud providers announced substantial spending to help customers deploy their products, giving them a commercial reason to expand their engineering teams. AWS announced in June that it would invest $1 billion to establish a dedicated FDE unit. Microsoft spent $2.5 billion to create an AI deployment division in July and promised to send 6,000 engineers into the field to help customers put AI into operation. The spending and staffing promise describe different parts of that expansion.

Services firms are also seeking deployment business through their existing practices. In early September, Accenture expanded its Google Cloud-based Gemini Business Group with the aim of adding 1,000 AI deployment specialists. Lemongrass promoted FDE support within its SAP ERP migration practice in August; ERP systems manage core business processes such as finance and operations. Both firms stand to gain service revenue as customers seek help introducing AI and changing enterprise systems.

Training plans offer another measure of expansion alongside investment and recruitment. IT services and consulting company DXC Technology announced in June that it would work with Anthropic to train tens of thousands of Claude-certified FDEs. On a July 31 earnings call, CEO Raul Fernandez told investors that DXC had 86 FDEs working across 57 customer environments. The training ambition and operating team describe different stages of building capability, and the short interval does not establish a missed target.

Premium fees require delivery capability

Limited access to experienced engineers can justify higher fees when an assignment requires exceptional expertise. But buyers still need to establish what the FDE label means in each proposal. In a Gartner brief published Tuesday, Saha said providers were using the term for implementation, professional services, solution engineering and AI consulting, sometimes thoughtfully and sometimes for its strategic appeal. The title alone therefore tells a buyer little about the work or the organization responsible for delivering it.

The organization matters because technical expertise is only one part of deployment. “Some charge premium fees without the delivery depth, program management, or change management maturity to justify them,” Saha said. Program management coordinates work and decisions across an engagement, while change management helps people adopt new processes. An engineer can solve a difficult technical problem while the wider team still lacks the ability to turn that solution into a usable business result.

Paying for that wider team can create substantial ongoing costs. Gartner estimates that consulting fees alone could exceed $200,000 quarterly per use case. It warns that customers who fail to develop internal engineering skills can remain dependent on specialized assistance even as the work becomes better understood. Early technical progress therefore leaves a separate operating decision: who will maintain the deployment and extend it over time?

As the work becomes clearer, buyers have a stronger basis for comparing delivery options. Gartner warns that customers could keep paying “premium rates for work that, once the scope is clear, a traditional services or partner model could potentially deliver more cost-effectively and with more predictable results.” Continuing uncertainty may warrant specialist involvement, while a well-defined assignment makes other delivery models easier to assess. Scope, capability and cost should therefore be reviewed as the engagement develops.

That review still leaves a substantial role for FDE teams. “If you truly have the right FDE model, we see it bringing good outcomes,” Saha said, adding that the resources required are unusually specialized and highly capable. Gartner sees potential when expectations are realistic and governance, the assignment of decision rights and oversight, is sound. Saha’s favorable assessment concerns those conditions; Gartner’s abandonment forecast concerns the broader market.

Palantir provides a precedent for matching exceptional people to demanding work. The software provider, which benefits commercially from successful deployments of its products, is widely credited with creating the FDE framework. Saha points to its work for high-profile government clients that needed exceptional people to complete difficult assignments quickly. His concern is the attempt to reproduce that approach for every customer, regardless of the assignment’s demands or how long specialist support remains necessary.

Vendor urgency shapes the delivery model

The need for specialist support can change during an engagement, while the vendor has an immediate incentive to demonstrate value. “Product companies want to make their product sticky,” Saha said, explaining that they also want results they can show to Wall Street investors. Competition gives them a limited time to produce those results, so intensive engineering support can help turn a sale into visible use. Customers benefit when that urgency moves a stalled deployment forward.

Faster deployment, however, does not automatically give the customer the skills to operate independently. A vendor can demonstrate product value while the customer continues to rely on external engineers for maintenance and further development. This can happen without a deliberate effort to create dependency because proving value quickly and building the customer’s capability require different work. Buyers need to make responsibility for that capability explicit.

Vendors also bear costs when they retain a large consulting workforce between assignments. Saha argues that bypassing partners creates another problem because those partners already have customer relationships and know the relevant contacts. Training partners can address both concerns, and Google Cloud, AWS and Microsoft have continued to support their networks of service and sales partners. Direct FDE investment and partner training can therefore serve complementary commercial purposes.

Partner development changes who can own delivery

Partner training gives vendors another way to help customers deploy their products while services firms build revenue-generating expertise. Anthropic and OpenAI sought help from partners with AI enablement, the skills and support needed to put AI into use, and Saha said most large AI vendors were creating standardized FDE credentials. These efforts broaden the range of organizations that may be able to take responsibility for an engagement. Buyers still need to examine what each provider can deliver.

Microsoft’s Frontier Transformation Engineer certification shows how vendors are developing that partner capacity. Added to its technical certifications in July, it is intended to train partners in FDE methodologies. Alongside Microsoft’s own deployment division, the certification builds expertise in organizations that already sell and deliver services to customers. Microsoft benefits when those organizations help customers adopt its technology, while partners gain skills they can use in their service businesses.

Credentials address training, but building a customer-facing practice also requires delivery experience. Technology marketplace Pax8 supplies its own FDEs to help its service-provider clients establish those practices, giving Pax8 a commercial interest in their growth. Its engineers temporarily work under the partner’s identity while the partner develops a service it can offer to customers. Support continues until the partner is ready to take responsibility for the work.

Craig Donovan, COO of Pax8, sees this arrangement as especially relevant to smaller customers. “The MSP channel will be the forward deployed engineers for small and midsize businesses,” he told Channel Dive. A managed service provider, or MSP, manages technology services for customers and can give organizations without large engineering teams access to deployment expertise. Midmarket organizations seeking to use AI are also potential customers for these partner-delivered services.

Reaching those customers is difficult when deployment depends on a small pool of highly capable engineers. Some buyers will struggle to secure those people, while others will find sustained premium engagements hard to afford. Pax8’s arrangement puts temporary expertise inside a service organization that can develop its own ongoing delivery capability. Donovan described the purpose as helping partners become active more quickly.

The MSP establishing its practice receives that initial help, so developing the customer’s skills remains a separate responsibility. Pax8’s engineers can help a partner prepare to deliver a service without establishing how the eventual customer’s team will learn to operate or extend its deployment. A buyer considering this arrangement needs to identify both transfers of responsibility: from temporary engineers to the partner, and from external specialists to any customer staff expected to own the system.

Existing service businesses give partners a foundation for taking on that work. FDE assignments include activities familiar to consulting, integration and management-services firms, which is why Saha expects AI implementation to remain substantially within the channel. Those firms already organize delivery around customer requirements and ongoing service responsibilities. Specialized AI expertise can extend those practices, provided the organization can also manage delivery and support changes in how customers work.

For a consulting or managed-service leader, the practical question is what the temporary engineers will help the practice learn to do. Training and credentials can prepare staff, while work alongside experienced deployment engineers can help them apply those skills. The engagement should identify who will manage delivery, support organizational change and continue the customer relationship after the temporary team leaves. These responsibilities give the practice a concrete basis for deciding when it is ready to take over.

For the customer, that division of work belongs in procurement and contract discussions. The buyer needs to know which organization supplies the expertise, which manages the engagement, and which will own ongoing delivery. The agreement should also establish what the customer or partner is expected to learn and when scope, capability and cost will be reviewed. It can accommodate continuing specialist support without a fixed handoff date, provided an owner is named for subsequent delivery and any extension receives an explicit decision.

Key takeaways for leaders

• Match specialist support to the assignment: Buyers need to establish which deployment problems require forward-deployed engineers and reassess that need as scope becomes clearer. Well-defined work may suit conventional services with lower costs and more predictable delivery.
• Verify available engineering capacity: Procurement teams need evidence that a provider can assign suitably experienced engineers. Investment announcements, recruitment plans and training targets describe different stages of building delivery capacity.
• Test what premium fees cover: Gartner estimates consulting fees could exceed $200,000 quarterly per use case. Buyers need to assess program management, change management and skills transfer alongside technical expertise.
• Assign ownership beyond initial deployment: Vendors have commercial incentives to demonstrate product value quickly. Customer contracts need to name who will maintain and extend the deployment, what internal staff will learn and when continued specialist support requires approval.
• Assess partners’ readiness to take over: Service partners offer another route to ongoing AI delivery, but credentials alone provide limited evidence of capability. Buyers need to examine delivery experience and define handoffs from temporary engineers to partners and customer staff.