AI Capacity Crunch: 90% of Code Now Agent‑Generated, Latency and Cost Rise
When I opened my cloud dashboard this week, the latency numbers were already nudging higher and the bill was creeping past the usual ceiling. In a handful of pipelines, about 90 % of the code is now being churned out by autonomous agents - the stats are hard to ignore. The tech itself looks slick, but the price tag isn’t getting the same applause.
Companies that used to lean on human programmers are starting to feel longer response times and a noticeable jump in compute spend; some analysts think the agents are simply bumping into their own limits. At the same time, the chatter in the big labs is shifting. Reinforcement learning, which used to sit on the fringe, seems to be the hot-button topic for data scientists at OpenAI, Anthropic and Gemini.
They’re eyeing it as a possible lever to stretch capacity, even as pricing models tip toward surge-pricing thresholds. So we’re left with a tug-of-war between the boost in productivity and the rising costs - and that tension is what the next observation is all about.
It's now estimated that in some cases, 90% of software is generated by coding agents. Now that agents have essentially come of age, Bercovici noted, reinforcement learning is the new conversation among data scientists at some of the leading labs, like OpenAI, Anthropic, and Gemini, who view it as a critical path forward in AI innovation.. It blends many of the elements of training and inference into one unified workflow," Bercovici said.
"It's the latest and greatest scaling law to this mythical milestone we're all trying to reach called AGI -- artificial general intelligence," he added. "What's fascinating to me is that you have to apply all the best practices of how you train models, plus all the best practices of how you infer models, to be able to iterate these thousands of reinforcement learning loops and advance the whole field." The path to AI profitability There's no one answer when it comes to building an infrastructure foundation to make AI profitable, Bercovici said, since it's still an emerging field.
Is the AI world ready for a surge-pricing model? At VentureBeat’s AI Impact session, a few speakers warned that capacity is already tight, latency spikes and cloud lock-in are starting to bite. Bercovici said roughly ninety percent of new software now comes from coding agents, so the agents seem to have finally grown up.
Meanwhile, labs like OpenAI, Anthropic and Gemini are betting on reinforcement learning, which adds another wrinkle to scaling. Prices are creeping up, and the Uber-style surge analogy makes the pressure feel immediate. It’s still unclear whether such pricing will smooth things out or just add more strain to the infrastructure.
Companies could also get stuck with a vendor, making a move to another cloud harder than hoped. Developers are already seeing higher latency, and that might hurt user experience if it isn’t tamed. The shift toward agent-generated code is real, yet we can’t say for sure how it will shape productivity or budgets in the long run.
We’ll have to juggle ambition with the hard limits of the current stack as this picture evolves.
More in Business & Startups
AI shifts hiring: companies favor generalist engineers over specialists
Why are hiring managers rethinking their talent playbooks? The rise of AI tools is nudging firms...
Rivian showcases new AI chip and silicon inspection labs in office tour
Rivian is opening its doors to a behind‑the‑scenes look at the hardware that could define the next...
Anthropic's AI store profits as CEO acknowledges eternal transcendence
Anthropic’s new AI storefront is finally turning a profit, a milestone that feels almost ceremonial...
Gnani.ai launches Vachana STT model as core IndiaAI infrastructure
Gnani.ai, the Bengaluru‑based speech‑technology firm, has just put its newest speech‑to‑text engine...
PC Company Invests in AI Engines, Context‑Aware Computing, Intelligence
A PC maker long known for assembling laptops and desktops is now positioning itself as a backbone...
Common Questions Answered
Why is the AI capacity crunch causing latency to increase for developers?
The capacity crunch stems from coding agents operating near their limits, which slows response times. As agents generate up to 90% of code, the heightened demand strains compute resources, leading to noticeable latency spikes.
How does the rise of coding agents affect cloud bills for companies?
With autonomous agents producing the majority of software, compute usage has surged, inflating cloud expenses. Companies that previously relied on human coders now see higher spend due to the intensive processing required by these agents.
What role does reinforcement learning play in the current AI scaling challenges?
Reinforcement learning is being explored by labs like OpenAI, Anthropic, and Gemini as a way to unify training and inference workflows. However, its integration adds complexity to scaling efforts, contributing to the overall capacity crunch.
Is the claim that 90% of software is now generated by coding agents accurate?
According to Bercovici, in certain workflows up to 90% of code is churned out by autonomous agents. This figure highlights the rapid adoption of agents, though it may vary across different development environments.