$27B bets, AI wrappers, and a video model pause

Here’s what’s worth your time today.

Meta is reportedly considering layoffs that could impact 20% or more of its workforce, according to Reuters. The move could help offset the company’s massive spending on AI infrastructure, acquisitions, and hiring as it pushes deeper into the AI race. Meta had nearly 79,000 employees at the end of last year.

A company spokesperson said the report was “speculative,” but the discussion comes as tech firms increasingly tie job cuts to AI-driven restructuring. Meta previously carried out large layoffs in 2022 and 2023, eliminating more than 21,000 roles during its earlier cost-cutting push.

Supported by Descope

Building a B2B AI application means making dozens of infrastructure decisions before you can fully realize their impact. Frontend / backend / agentic frameworks, database, AI model providers, auth and identity, observability, eval…the list is long and growing.

This comprehensive founders guide contains insights and learnings from successful AI startups like WisdomAI, Echelon AI, and Different on:

Bookmark the guide and see how your AI app compares

Meta has agreed to spend up to $27 billion over the next five years on AI infrastructure from Nebius Group as it ramps up computing capacity for its AI ambitions. Nebius will provide $12 billion in dedicated data center capacity starting in 2027, with Meta committing to purchase up to $15 billion more as additional infrastructure comes online.

The deal highlights how aggressively Meta is investing to compete with AI leaders like OpenAI and Google. The company is pouring billions into chips, data centers, and infrastructure, including partnerships with Nvidia and Advanced Micro Devices, while also developing its own AI hardware in-house.

Some of the biggest voices in AI, including Sam Altman of OpenAI and Alex Karp of Palantir, have recently warned that AI could massively disrupt jobs and society. Critics say this messaging may help companies raise money and position themselves as the only ones capable of building safe AI — but it risks frightening the public.

Public sentiment is already fragile. Polling cited in the report shows only 26% of U.S. voters view AI positively, and some executives worry growing fear could fuel calls to restrict or ban the technology. While some leaders argue the warnings are honest reflections of AI’s potential impact, the increasingly dramatic messaging could make it harder for the broader public to embrace the technology.

ByteDance has reportedly delayed the global launch of its AI video model Seedance 2.0 after facing backlash from Hollywood studios. The model, which debuted in China earlier this year, went viral online with clips such as a deepfake-style video showing Tom Cruise fighting Brad Pitt.

Studios including The Walt Disney Company quickly issued cease-and-desist letters, accusing the company of using protected film characters and likenesses without permission. ByteDance had planned a global release in March but is now reportedly pausing the rollout while engineers and legal teams work on stronger intellectual property safeguards.

Google and Accel selected five startups for their AI accelerator in India after reviewing more than 4,000 applications — and none of them were simple “AI wrappers.” Around 70% of submissions were rejected for layering chatbots or basic AI features onto existing software without rethinking workflows.

The selected companies will receive up to $2M in funding plus cloud credits through Google’s AI Futures Fund. The cohort includes startups building an AI research assistant, enterprise agents for ERP systems, voice AI for call centers, AI-generated film tools, and automation for industrial manufacturing.