OpenAI models hit AWS as AI cloud war intensifies

OpenAI models hit AWS

OpenAI announced that its frontier models and Codex are now available on Amazon Web Services. The move puts the same large‑scale language models that power ChatGPT behind the same infrastructure that runs most of the internet’s workloads. Developers can call the models through standard AWS APIs. OpenAI did not disclose pricing or latency figures in the brief release.

The announcement arrives as cloud providers scramble to lock in AI workloads. AWS already offers its own Bedrock service, but the addition of OpenAI’s flagship models gives customers a single‑vendor path to the most widely used generative AI. The integration also means that enterprises can keep data within the AWS ecosystem while tapping OpenAI’s research‑grade models.

Alphabet pours $80 billion into AI compute

Alphabet disclosed a proposed $80 billion equity capital raise aimed at expanding its AI infrastructure and compute capacity. The fundraising effort signals that Google Cloud will need more silicon and networking to stay competitive in a market where OpenAI, Microsoft, and Amazon are all courting the same customers.

Alphabet’s plan is to use the capital to build new data centers, upgrade GPU clusters, and invest in custom AI chips. The scale of the raise dwarfs most recent cloud‑AI investments and underscores the belief that compute will be the limiting factor for future generative AI services.

Analysts note that the capital raise could pressure Alphabet’s margins in the short term, but the firm expects a “trillion‑market” opportunity in AI‑enabled enterprise software. The exact timeline for deployment was not disclosed.

IBM splits to double‑down on cloud and AI

IBM announced that it will spin off its legacy IT infrastructure business into a separate public company, tentatively called “NewCo.” The spin‑off will contain roughly 90,000 employees and $19 billion in revenue, while the parent company will retain its hybrid‑cloud and AI focus.

CEO Arvind Krishna, who took over in April 2020 after steering the $34 billion Red Hat acquisition, framed the move as the latest in a series of divestments that include networking in the 1990s, PCs in the 2000s, and semiconductors about five years ago. Krishna said shedding low‑margin infrastructure work will let IBM concentrate on higher‑value cloud software and AI solutions.

Industry observers, such as Wedbush analyst Moshe Katri, called the split “smart,” arguing that automation and cloud services have eroded the profitability of traditional managed infrastructure. The new entity will automatically become the world’s leading managed infrastructure services provider, serving about 4,600 customers that cover more than 75 percent of the Fortune 100.

Microsoft’s cloud‑first PC strategy adds pressure

Microsoft has been moving Windows toward a subscription‑based, cloud‑delivered model for several years. The Windows 365 Cloud PC now streams a full Windows operating system from Azure to any device. Pricing for the service sits in the $30‑to‑$60 per‑user‑per‑month range, depending on the chosen configuration.

The bottom‑end offering includes 2 virtual CPUs, 4 GB of RAM, and 64 GB of storage. The service can run on Macs, iPhones, Android tablets, Linux machines, and even VR headsets. Microsoft’s internal slides describe the goal as “moving Windows 11 increasingly to the cloud” and positioning the cloud PC as the default desktop experience.

AI features have been layered on top of the cloud PC, but industry voices remain skeptical. Dell’s head of product, Kevin Terwilliger, warned that “they’re not buying based on AI” and suggested that AI may confuse customers more than it helps them. The criticism echoes a broader sentiment that AI‑infused PCs have yet to prove a compelling value proposition.

What to watch

The convergence of AI models, cloud compute, and subscription‑based desktops creates a competitive triangle. Track three data points over the next six months: (1) AWS pricing and latency metrics for OpenAI’s models, (2) Alphabet’s capital deployment milestones, especially new GPU‑focused data centers, and (3) the adoption rate of Microsoft’s Windows 365 Cloud PC among enterprise customers. Together these signals will reveal which cloud provider secures the most AI‑driven workloads and how the industry balances on‑premise control against cloud‑only convenience.

Updates

• 2026-07-28 — The US is banning foreign robots (source)