AMD
AMD launched an AI cost calculator for enterprise clients to compare total cost of ownership for AI workloads across different hardware stacks.
Why it matters for sellers
New product = GTM investment and tooling needs
Signal details
- Reported
- August 26, 2026
- Source
- longbridge.com
From the coverage · longbridge.com
1. S. 64%. S. indices rose, and Nvidia ended its seven-day run. Easing geopolitics boosted global risk appetite, with energy under pressure while tech/semis rebounded. Softer oil alleviates inflation and long-end rate pressures, a near-term tailwind for long-duration growth stocks. 2. S. Jul PCE report is due. PCE directly shapes rate-cut odds and is the core input for global pricing tonight, so the print bears close watching. 1. HK) XPeng closed up nearly 8%. It announced independent financing for its robotics unit that sets a new private-market record (surpassing Unitree), and He Xiaopeng said the first advanced humanoid robot will go into true mass production.
1 bn loss for the half-year, short-term borrowings rose by RMB 5 bn, and the full-year profit target was cut to RMB 3 bn from RMB 5 bn. The bounce was driven by repricing of the long-dated optionality from the robotics narrative and funding milestone, but widening core auto losses and a cautious guide remain fundamental headwinds; sector stabilization in autos also helped. This is still a case of story lifts multiples, profits need to materialize. 2. US) OpenAI and Broadcom co-developed the Jalapeno in-house inference ASIC. Lab tests show better energy efficiency than Nvidia GB300, designed for low-latency inference for LLMs and AI agents, and engineering samples have run model workloads; supply will be prioritized for OpenAI’s own services.
This marks a tangible step by OpenAI to reduce reliance on Nvidia’s supply chain, intensifying competition in in-house inference chips among AI giants. For now, results are only from lab tests, and there is no disclosed timeline for mass production, yields, or external shipments, so it should not directly hit Nvidia revenue near term. 5. US) AMD launched an AI cost calculator for enterprise clients, allowing inputs for model parameters and concurrency to compare total cost of ownership across training and inference under different hardware stacks, covering AMD MI series and competing gear.
It is an enterprise-focused tool that lowers selection barriers and highlights AMD’s cost advantages. Still, it is primarily a marketing enabler; share gains in hardware ultimately require product deliveries and continued build-out of the software stack. 6. US) Per media citing internal docs, Meta will launch a consumer AI agent platform, Hatch, in the coming weeks. It can autonomously complete tasks across apps and integrate with third-party services such as food delivery, email, and e-commerce; a tiered subscription is under consideration, with the top tier approaching $200 per month, distributed via WhatsApp and Instagram.
However, the product has not been officially announced. Whether consumers will accept the high pricing and how privacy permissions are handled remain key to watch. 7.
Get signals like this for every account you sell to
Our engine detects, verifies, and deduplicates thousands of buying signals every day across 50M+ companies — delivered via API, GCS push, or flat file.
More from the last 24 hours
SpaceX
CapEx InvestmentSpaceX has invested over $15 billion in its Starship rocket.
Amazon
PartnershipAmazon and Nvidia announced an expanded partnership that goes beyond chip purchases to include integration of Nvidia's networking hardware, open models, CPUs, software, and robotics platform across Amazon Web Services.
NVIDIA
New CustomerApple is now using NVIDIA GPUs with Confidential Computing for confidential inference in its Private Cloud Compute.
CrowdStrike
EarningsCrowdStrike reported a 26% jump in Q2 revenue from $1.17 billion a year ago.