Artificial intelligence giants have entered a “boots-on-the-ground” war, pouring astronomical sums into securing enterprise customers. The starting signal was Microsoft’s (MS) announcement of a new 6,000-person AI deployment organization backed by a $2.5 billion investment (approximately 3.8 trillion won). This move is being followed by Amazon Web Services (AWS) and OpenAI, which are also significantly strengthening their Frontier Deployed Engineering (FDE) strategies—embedding engineers directly into corporate environments. At the same time, OpenAI and Anthropic are accelerating a “customer lock-in” race by distributing millions of dollars in free token credits to dominate the startup ecosystem.

This movement signifies that AI has moved beyond simple pilot programs and entered the “AI Transformation (AX)” phase, where core business processes are being redesigned. Big Tech companies have concluded that merely providing a good model is no longer sufficient. They must now directly transplant AI into a client’s complex internal data, security policies, and workflows to deliver measurable business outcomes.

A $2.5 Billion Bet to Create ‘AI-Native Enterprises’

The “Microsoft Frontier Company,” announced by MS on the 2nd, represents the pinnacle of this strategy. This organization aims to deploy 6,000 industry experts and engineers for long-term assignments inside client companies to co-design and deploy AI systems. Judson Althoff, Microsoft’s Chief Commercial Officer, stated that customers have moved beyond the experimentation phase and are now focused on proving AI return on investment (ROI), emphasizing that this organization will be “the industry’s largest and most capable outcome-focused engineering organization.”

This is more than simple staff augmentation. MS views a company’s data, expertise, and decision-making processes as “unique intelligence” and aims to rewrite workflows by extending this intelligence with AI. The strategy is to facilitate a shift away from simply purchasing software to augment existing tasks, toward becoming an “AI-native enterprise” where AI is the central process driver.

A month earlier, on June 30, AWS announced a $1 billion investment in its own FDE organization. AWS differentiated its approach by explaining that its teams collaborate with a client’s business, engineering, and security teams to build AI systems and, upon project completion, leave behind a “knowledge graph” and operational manuals so the client can run the system independently. Gary Brantley, Chief Information Officer (CIO) of the NFL, stated that collaboration with AWS FDE enabled the launch of a commercial service in just a few weeks.

The originator of this FDE model, Palantir, has been deploying engineers to build systems on-site for U.S. intelligence agencies and corporations since the 2010s. With the proliferation of generative AI, this approach is now spreading across the cloud and AI platform industry. OpenAI has also joined the fray, establishing the “OpenAI Deployment Company” and securing about 150 deployment specialists.

The Startup Scramble: Locking in the Ecosystem with ‘Free AI’

Alongside the FDE competition in the enterprise field, a war of free credits targeting startups to secure future customers is also heating up. According to a Wall Street Journal report on the 7th (local time), OpenAI and Anthropic are competitively offering massive token credits, primarily to startups participating in Y Combinator (YC), Silicon Valley’s premier accelerator.

OpenAI CEO Sam Altman proposed $2 million in credits to YC startups in May in exchange for equity. In response, Anthropic dramatically raised its free credits for YC startups from the previous level of $30,000 to $500,000—with no equity requirement. OpenAI then adjusted its terms, offering a base of $500,000 in credits without equity, plus an additional $1.5 million in exchange for equity. Some startups have received total credit offers exceeding $3 million from both companies.

This is a long-term strategy exploiting the fact that early-stage startups building products on a specific model find it difficult to switch later. Ilya Volkov, co-founder of the AI startup Touchmark, said the benefits allowed them to focus on “token-maxxing,” utilizing as many tokens as possible. Industry estimates suggest the annual value of credits these companies could provide to YC may reach up to $800 million.

This free credit race is a complex phenomenon intertwined with the profitability pressures facing AI companies. OpenAI and Anthropic are under pressure to improve profitability ahead of potential initial public offerings (IPOs). OpenAI, in particular, is preparing for an IPO targeting a $1 trillion valuation but is reportedly reviewing its timeline due to concerns over a cold market reception. CEO Sam Altman faces the challenge of proving the company is a “sustainable machine” through a technological leap with GPT-5, based on annual revenues of roughly $24 billion.

AI Model Companies Shake Up the Cloud Ecosystem

The rise of AI model companies is disrupting the revenue structures and partnership policies of traditional Cloud Service Providers (CSPs). A prime example is Anthropic’s renegotiation of its contract with Amazon, shifting the billing system from a compute-time basis to a token-based one. As Anthropic supplies its “Claude” model to the three major CSPs—AWS, Microsoft Azure, and Google Cloud—this shift demonstrates that the policies of the model company, not the CSP, are beginning to dictate pricing and distribution.

Cloud Managed Service Providers (MSPs) in South Korea are also on edge. One industry insider noted, “In the past, we only had to watch the policies of AWS or MS, but now we must also monitor the pricing and distribution strategies of OpenAI and Anthropic.” They predicted that “these changes will impact the revenue structures and customer proposal methods of domestic MSPs.” In response, AWS is expanding its own “Nova” AI model, and Google is strengthening its “Gemini” model and TPU infrastructure in moves to reduce dependency on specific models.

The Rise of the ‘FDE’: A New Talent Profile for the AI Era

These changes are also causing a seismic shift in the talent market. The FDE is emerging not as a mere outsourced developer or consultant, but as a new profession bridging technology and business. They serve as a critical link, transplanting AI models into real-world tasks at client sites and feeding the results back into model development. Microsoft’s FDE partnerships with global consulting firms like Accenture and KPMG suggest potential future competition with these same consultancies.

Within South Korea, Naver Cloud has announced plans to introduce an FDE-centric system in the defense AI sector, and LG CNS is also emphasizing a dedicated FDE organization through its collaboration with Palantir. As AI enters the phase of redesigning core enterprise processes, the capability to “implement on the ground” is becoming the decisive competitive battleground, surpassing the importance of simply building the models.