Friday, June 19, 2026

OpenAI bulks up for IPO with Transformer inventor and Trump AI policy official

3 stories·49 sources curated中文

Title: AI giants position for dominance ahead of IPOs

01OpenAI bulks up for IPO with Transformer inventor and Trump AI policy official

OpenAI is making strategic hires as it positions itself for a potential public offering, bringing on Transformer co-inventor Noam Shazeer from Google DeepMind. Shazeer, who co-authored the 2017 paper that became the foundational architecture for modern large language models, left Google after the company declined to commercialize his subsequent work on character-level language models. His addition signals OpenAI's intent to maintain technical leadership as it navigates the transition from research lab to publicly traded company.

In the same week, OpenAI also hired Dean Ball, a former Trump administration official who served as a senior policy advisor for AI at the Treasury Department and worked on the National AI Initiative Office. The dual hires suggest OpenAI is building both its technical and regulatory capabilities in preparation for the scrutiny that comes with being a public company, as the startup continues to burn cash while seeking paths to profitability.

02Amazon eyes $50B opportunity by selling AI chips to other data centers

Amazon Web Services is in talks to sell its Trainium and Inferentia chips to external data centers, a move that would directly challenge Nvidia's dominance in the AI accelerator market. CEO Andy Jassy has described the chip sales opportunity as representing a potential $50 billion market for the company. Currently, Amazon uses these custom-designed chips internally to power AI workloads across its cloud platform, but expanding to external customers would mark a significant shift in strategy.

The effort positions Amazon to capture a larger share of the AI infrastructure market, where Nvidia has maintained a near-monopoly on the GPUs that train and run AI models. AWS custom chips are designed to offer competitive performance at lower power consumption and cost, though they currently lack the extensive software ecosystem that has made Nvidia's CUDA platform the industry standard. Industry analysts note that winning external customers will require Amazon to invest heavily in developer tools and support infrastructure.

03AI inference startup Baseten reportedly raising $1.5B at $13B valuation

The inference gold rush continues as Baseten pursues a $1.5 billion funding round at a $13 billion valuation, according to sources familiar with the matter. The San Francisco-based startup, which provides infrastructure for deploying and running AI models at scale, previously raised a large round just months ago, reflecting the unprecedented pace of investment flowing into AI infrastructure companies. Inference—running trained AI models to generate outputs—has emerged as a critical bottleneck as enterprises deploy AI applications at scale.

Baseten's platform helps companies manage the computational demands of serving AI models to end users, handling tasks like autoscaling, latency optimization, and cost management. The company competes with cloud providers and other infrastructure startups in a market that has seen demand surge following the commercial success of generative AI products. The proposed valuation would represent a significant jump from Baseten's previous funding rounds, underscoring investor appetite for picks-and-shovels plays in the AI boom.

Also today

ShareXLinkedIn

Continue reading

Jun 18

Google's AMIE AI Matches Primary Care Physicians in Disease Management, Study Shows

Jun 17

Android 17 launches with new multitasking tools as Google expands Gemini features

Jun 15

Meta Unwinds $2B Manus Acquisition After Beijing's Demand

← BackView all archives →