News Grower

Independent coverage of AI, startups, and technology.

Ars Technica Apr 22, 2026 at 17:10 Big Tech Rising Hot

Google unveils two new TPUs designed for the "agentic era"

Google's new generation of Tensor AI chips is actually two chips, one for inference and one for training.

Signal weather

Rising

Momentum is building quickly, so this card is a good early entry point into the topic.

By Ryan Whitwam Original source
Google unveils two new TPUs designed for the "agentic era"

Most of the companies that have fully committed to building AI models are gobbling up every Nvidia AI accelerator they can get, but Google has taken a different approach. Most of its cloud AI infrastructure is based on its line of custom Tensor processing units (TPUs). After announcing the seventh-gen Ironwood TPU in 2025, the company has moved on to the eighth-gen version, but it's not just a faster iteration of the same chip. The new TPUs come in two flavors, providing Google and its customers with an AI platform that is faster and more efficient, the company says. Google is pushing the idea that the "agent era" is fundamentally different from the AI systems that came before, necessitating a new approach to the hardware. So engineers have devised the TPU8t (for training) and the TPU 8i (for inference). Before AI models become something you can use to analyze data or make silly memes, they need to be trained. The TPU 8t was designed specifically for this part of the AI lifecycle to reduce the training time for frontier AI models from months to weeks. Read full article Comments

Stay on the signal

Follow Google unveils two new TPUs designed for the "agentic era"

Follow this story beyond a single article: new follow-ups, adjacent sources, and the evolving storyline.

We send a confirmation link first, then only meaningful digests.

Story map

Understand this topic fast

A quick entry into the story: why it matters now, who is involved, and where to go next for context.

Why it matters now

Fresh coverage with immediate momentum.
There are already 6 connected articles in the same storyline to continue from here.
The story keeps orbiting around Agentic Era, Ars Technica, and chips, so the entity pages are the fastest way to build context.
Ars Technica already has 4 follow-up stories on the same theme.

Topic constellation

Open the live map for this story

See which entities, story threads, sources, and follow-up articles shape this story right now.

Click nodes to continue

Entity Cluster Article Hub Source

Story timeline

Continue with this story

A short sequence of events and follow-up stories to understand the arc quickly.

Apr 22, 2026 at 19:42 Ars Technica

You want your Moon landings in HDTV? So does NASA—here's how it's happening.

"You just push this button, and in three hours, you're counting photons."

Apr 22, 2026 at 19:32 Ars Technica

Microsoft issues emergency update for macOS and Linux ASP.NET threat

When authentication fails, things can go very, very wrong.

Apr 22, 2026 at 18:39 TechCrunch

Google Cloud launches two new AI chips to compete with Nvidia

Google's newest TPUs are faster and cheaper than the previous versions. But the company is still embracing Nvidia in its cloud — for now.

Apr 22, 2026 at 18:34 Ars Technica

Anthropic tested removing Claude Code from the Pro plan

Untenable demand has Anthropic exploring new approaches to rationing its service.

Apr 22, 2026 at 17:55 Ars Technica

Coyote vs. Acme is finally getting released—with a killer trailer

What was Warner Bros. even thinking, shelving this film for so many years?

Apr 22, 2026 at 17:10 Ars Technica

Google unveils two new TPUs designed for the "agentic era"

Google's new generation of Tensor AI chips is actually two chips, one for inference and one for training.

How reliable this looks

Signal and trust for Ars Technica

This source works at a rapid pace: 100% of recent stories land in the hot window, and 0% carry visible search signal.

Trusted

Reliability

92

Freshness

100

Sources in storyline

2

Related articles

More stories that share tags, source, or category context.

More from Ars Technica

Fresh reporting and follow-up coverage from the same newsroom.

Open source page