Stock Markets April 22, 2026 08:22 AM

Alphabet Shares Advance After Google Cloud Unveils Next-Gen AI Chips and Partner Fund

New TPU chips, a $750 million partner fund and expanded cloud agreements underpin gains as token usage and enterprise adoption rise

By Maya Rios GOOGL
Alphabet Shares Advance After Google Cloud Unveils Next-Gen AI Chips and Partner Fund
GOOGL

Alphabet shares rose 1.7% on Wednesday after Google Cloud detailed new AI infrastructure, an ecosystem funding initiative and expanded customer arrangements at its Cloud Next '26 event. The announcements included two eighth-generation Tensor Processing Units tailored for training and inference workloads, a $750 million fund to support partners building agentic AI solutions, and metrics showing sharply increased token processing and customer adoption of AI products.

Key Points

  • Alphabet shares increased 1.7% after Google Cloud announced new AI hardware and partner initiatives.
  • Eighth-generation TPUs announced: TPU 8t for training (up to 9,600 chips, two petabytes shared memory) and TPU 8i for inference (1,152 TPUs per pod, three times more on-chip SRAM).
  • Google Cloud detailed a $750 million fund to support its 120,000-member partner ecosystem and reported substantial increases in token processing and AI product adoption.

Alphabet (NASDAQ:GOOGL) shares climbed 1.7% Wednesday in response to a series of Google Cloud announcements that highlighted new AI hardware, partner investments and expanded enterprise offerings.

At the Cloud Next '26 conference, Google Cloud introduced its eighth-generation Tensor Processing Units. The pair of specialized chips are aimed at distinct AI workloads. TPU 8t is tuned for model training and can scale to configurations of up to 9,600 chips with two petabytes of shared memory in a single superpod, a setup Google said delivers three times the processing power of the prior generation. TPU 8i is intended for inference tasks, able to connect 1,152 TPUs in a single pod and offering three times more on-chip SRAM to support AI agents that need low latency and high throughput.

Alongside the new silicon, Google Cloud announced a $750 million fund meant to bolster its 120,000-member partner ecosystem as those partners develop and deploy agentic AI solutions. The fund will be allocated toward AI value assessments, prototyping, agent construction and upskilling programs targeted at consulting firms, systems integrators and software partners.

Google Cloud also disclosed usage and adoption metrics for its AI services. The company said its models now handle more than 16 billion tokens per minute through direct API usage by customers, up from 10 billion tokens per minute in the prior quarter. Nearly 75% of Google Cloud customers are using the company’s AI products, and 330 customers each processed over a trillion tokens in the past 12 months.

On the customer front, Google Cloud expanded an agreement with Thinking Machines Lab, which will make use of A4X Max virtual machines equipped with NVIDIA GB300 GPUs. In early testing, Thinking Machines reported training and serving speeds doubled when using A4X Max VMs versus the previous generation of GPUs.

The company also unveiled the Gemini Enterprise Agent Platform, a product intended to help organizations construct, scale and govern AI agents. Google Cloud reported that paid monthly active users for Gemini Enterprise grew 40% quarter-over-quarter in the first quarter.

Internally, Google reported a shift in software development practices: 75% of all new code at the company is now AI-generated and subsequently approved by engineers, up from 50% last fall.


Key points

  • Alphabet shares rose 1.7% after Google Cloud revealed eighth-generation TPUs, a $750 million partner fund and new product and customer milestones.
  • TPU 8t targets model training and scales to 9,600 chips with two petabytes of shared memory in a single superpod; TPU 8i targets inference with 1,152 TPUs per pod and three times more on-chip SRAM.
  • Sectors impacted include cloud infrastructure, semiconductor acceleration for AI, enterprise software and consulting through partner deployments.

Risks and uncertainties

  • Performance claims for customer workloads may be preliminary - Thinking Machines’ reported twofold speed increases were described as results from early testing.
  • Token processing volumes are rising, but a portion of usage is concentrated among large customers - 330 customers each processed over a trillion tokens in the past 12 months, which may indicate concentrated demand patterns.
  • The effectiveness of the $750 million partner fund depends on partner uptake and successful deployment of agentic AI solutions, creating execution risk for the intended ecosystem benefits.

Risks

  • Performance improvements cited by partners were reported in early testing, indicating preliminary results.
  • A significant share of token processing appears concentrated among a subset of large customers, which could reflect concentrated demand.
  • Benefits from the $750 million partner fund depend on partner adoption and successful deployment of agentic AI solutions.

More from Stock Markets

USDA Awards $300 Million BPA to Palantir to Support Farm Security Plan Apr 22, 2026 UBS Sticks With Buy Rating, $180 Target for Yum! Brands Ahead of Q1 Results Apr 22, 2026 UBS Raises Biogen to Buy, Cites Multiple Near-Term Pipeline Catalysts Driving Potential Upside Apr 22, 2026 Einride and Legato Move Forward With SPAC Registration Valuing Freight Tech Firm at $1.35 Billion Apr 22, 2026 UBS Says ASML Capacity Concerns Likely Overstated After Management Roadshow Apr 22, 2026