The Apex Times
Above the Noise
Business Wire
BusinessNvidia posts another surge in revenue, as analysts point to outlines of continued demand strengthThe Apex TimesBusinessUber tells Axios it stabilized AI spending after earlier cost burn, as usage roseThe Apex TimesBusinessEli Lilly’s oncology sales are rising, prompting a question of whether growth can broaden beyond its GLP-1-era lineupThe Apex TimesBusinessNasdaq Edges Higher as Earnings News Lifts Nvidia and Other Mega-Cap NamesThe Apex TimesBusinessHoneywell shares draw attention after leadership shake-ups and a sharp pullbackThe Apex TimesBusinessU.S. stock futures lean higher after Nvidia results lift tech sentimentThe Apex TimesBusinessCitizens analyst reiterates Market Outperform on Amazon, sets $315 target as shares price 2027 earnings at about 24 timesThe Apex TimesBusinessNvidia’s “hidden” data-center business beyond hyperscalers is growing faster, raising questions about demand and concentrationThe Apex TimesBusinessModerna to raise up to $2.3 billion via convertible notes, indicating a push toward oncology growth while using proceeds to repay debtThe Apex TimesBusinessWall Street’s Bullish Take on FedEx Hinges on Consensus Views, Not Clear New CatalystsThe Apex TimesBusinessTarget’s turnaround announcement gets louder, but investors are still waiting for proof in the numbersThe Apex TimesBusinessNVIDIA shares rise about 6% after guidance points to 70% growth, even as management flags a margin hit from memoryThe Apex TimesBusinessNvidia posts another surge in revenue, as analysts point to outlines of continued demand strengthThe Apex TimesBusinessUber tells Axios it stabilized AI spending after earlier cost burn, as usage roseThe Apex TimesBusinessEli Lilly’s oncology sales are rising, prompting a question of whether growth can broaden beyond its GLP-1-era lineupThe Apex TimesBusinessNasdaq Edges Higher as Earnings News Lifts Nvidia and Other Mega-Cap NamesThe Apex TimesBusinessHoneywell shares draw attention after leadership shake-ups and a sharp pullbackThe Apex TimesBusinessU.S. stock futures lean higher after Nvidia results lift tech sentimentThe Apex TimesBusinessCitizens analyst reiterates Market Outperform on Amazon, sets $315 target as shares price 2027 earnings at about 24 timesThe Apex TimesBusinessNvidia’s “hidden” data-center business beyond hyperscalers is growing faster, raising questions about demand and concentrationThe Apex TimesBusinessModerna to raise up to $2.3 billion via convertible notes, indicating a push toward oncology growth while using proceeds to repay debtThe Apex TimesBusinessWall Street’s Bullish Take on FedEx Hinges on Consensus Views, Not Clear New CatalystsThe Apex TimesBusinessTarget’s turnaround announcement gets louder, but investors are still waiting for proof in the numbersThe Apex TimesBusinessNVIDIA shares rise about 6% after guidance points to 70% growth, even as management flags a margin hit from memoryThe Apex TimesBusinessNvidia posts another surge in revenue, as analysts point to outlines of continued demand strengthThe Apex TimesBusinessUber tells Axios it stabilized AI spending after earlier cost burn, as usage roseThe Apex TimesBusinessEli Lilly’s oncology sales are rising, prompting a question of whether growth can broaden beyond its GLP-1-era lineupThe Apex TimesBusinessNasdaq Edges Higher as Earnings News Lifts Nvidia and Other Mega-Cap NamesThe Apex TimesBusinessHoneywell shares draw attention after leadership shake-ups and a sharp pullbackThe Apex TimesBusinessU.S. stock futures lean higher after Nvidia results lift tech sentimentThe Apex TimesBusinessCitizens analyst reiterates Market Outperform on Amazon, sets $315 target as shares price 2027 earnings at about 24 timesThe Apex TimesBusinessNvidia’s “hidden” data-center business beyond hyperscalers is growing faster, raising questions about demand and concentrationThe Apex TimesBusinessModerna to raise up to $2.3 billion via convertible notes, indicating a push toward oncology growth while using proceeds to repay debtThe Apex TimesBusinessWall Street’s Bullish Take on FedEx Hinges on Consensus Views, Not Clear New CatalystsThe Apex TimesBusinessTarget’s turnaround announcement gets louder, but investors are still waiting for proof in the numbersThe Apex TimesBusinessNVIDIA shares rise about 6% after guidance points to 70% growth, even as management flags a margin hit from memoryThe Apex TimesBusinessNvidia posts another surge in revenue, as analysts point to outlines of continued demand strengthThe Apex TimesBusinessUber tells Axios it stabilized AI spending after earlier cost burn, as usage roseThe Apex TimesBusinessEli Lilly’s oncology sales are rising, prompting a question of whether growth can broaden beyond its GLP-1-era lineupThe Apex TimesBusinessNasdaq Edges Higher as Earnings News Lifts Nvidia and Other Mega-Cap NamesThe Apex TimesBusinessHoneywell shares draw attention after leadership shake-ups and a sharp pullbackThe Apex TimesBusinessU.S. stock futures lean higher after Nvidia results lift tech sentimentThe Apex TimesBusinessCitizens analyst reiterates Market Outperform on Amazon, sets $315 target as shares price 2027 earnings at about 24 timesThe Apex TimesBusinessNvidia’s “hidden” data-center business beyond hyperscalers is growing faster, raising questions about demand and concentrationThe Apex TimesBusinessModerna to raise up to $2.3 billion via convertible notes, indicating a push toward oncology growth while using proceeds to repay debtThe Apex TimesBusinessWall Street’s Bullish Take on FedEx Hinges on Consensus Views, Not Clear New CatalystsThe Apex TimesBusinessTarget’s turnaround announcement gets louder, but investors are still waiting for proof in the numbersThe Apex TimesBusinessNVIDIA shares rise about 6% after guidance points to 70% growth, even as management flags a margin hit from memoryThe Apex Times
Back to front
NVIDIA begins shipping Vera, its first agentic AI-focused custom CPU, with early deliveries to cloud and AI partners
The Apex Times

THE APEX TIMES

Business/The Apex Times/Aug 27, 9:17 AM EDT

NVIDIA begins shipping Vera, its first agentic AI-focused custom CPU, with early deliveries to cloud and AI partners

The new chip, built for the CPU-heavy work behind AI agents, is moving from concept to production as NVIDIA hands systems to major customers including AWS, Oracle Cloud Infrastructure, and leading AI labs.

4 min readEditor-approved Apex article

NVIDIA says it has started shipping Vera, its first custom CPU designed specifically for agentic artificial intelligence, the computing style where AI systems not only answer questions but also carry out multi-step tasks. In an update to its original announcement, the company describes the first Vera CPU server deliveries moving across the AI ecosystem at scale, starting with a handoff to Amazon Web Services in Seattle.

The most recent delivery highlighted by NVIDIA went to AWS. The company says its vice president of Hyperscale and HPC, Ian Buck, traveled to AWS headquarters to hand over AWS’s first Vera CPU server along with a Vera Rubin GPU, placing the hardware in front of the Amazon EC2 team. AWS is the latest stop in Vera’s “rapid journey,” following earlier hands-offs to Oracle Cloud Infrastructure and to multiple AI labs.

NVIDIA also ties Vera’s rollout to the broader CPU moment it says is emerging as model usage shifts from static responses to tool use, orchestration, and long-context operations. Buck is quoted describing agentic AI as creating a new kind of CPU demand in the AI factory, arguing that conventional “core-density” CPU designs were not built for the concurrent, real-time workload patterns now expected from AI assistants and autonomous systems.

Technically, NVIDIA positions Vera as a purpose-built processor for the CPU work it says agents require even when GPUs remain central. The company says AI agents involve orchestration, tool-calling, reinforcement-learning workloads, data analytics, agent sandboxing, and long-context state management. Vera is presented as designed to handle “real-time” and “concurrent” tasks that, in NVIDIA’s view, can bottleneck traditional server CPUs.

NVIDIA provides several specifications and performance claims. The Vera CPU is described as packing 88 custom NVIDIA-designed Olympus cores, delivering 1.2TB/s of memory bandwidth, and targeting up to 1.8x faster per-core performance on agentic AI workloads under sustained load. NVIDIA adds that Vera and Rubin share a unified memory architecture in paired systems, and that this combination is meant to keep accelerated compute highly utilized.

The rollout narrative in the blog spans multiple buyers with different use cases. For Oracle Cloud Infrastructure, NVIDIA quotes a partner executive describing an internal expectation to deploy “hundreds of thousands” of Vera CPUs beginning in 2026, citing sustained performance needs at massive scale for agentic AI. For AI labs, NVIDIA describes deliveries to Anthropic and OpenAI earlier in the process, with both teams reviewing Vera’s features and discussing how the CPU fits into their workloads.

The company also describes the hardware arriving at SpaceXAI’s offices in Palo Alto for evaluation. NVIDIA says SpaceXAI is assessing Vera for reinforcement learning workloads and for agent-based simulation pipelines that support its training stack. It further recounts that the process included direct technical discussion with Elon Musk, including questions about cores, memory layout, and cooling.

Vera is positioned as part of NVIDIA’s wider co-design platform for data centers, described in the blog as running alongside Rubin GPUs and other building blocks such as BlueField-4 data processing units (DPUs), Spectrum-X networking, and MGX rack architecture. In addition to standalone CPU systems, NVIDIA says Vera is the host processor for Vera Rubin NVL72, where it pairs using second-generation NVLink-C2C with a pair of Rubin GPUs, implying a system-level approach rather than a CPU sold in isolation.

What NVIDIA does not disclose in the blog are commercially specific terms, detailed customer acceptance metrics, or any timetable for when Vera server builds will reach specific production volumes beyond the statement that it is “shipping now” and moving “across” the ecosystem. The company also does not provide independent benchmarks, third-party performance validation, or cost details that would allow customers to compare Vera directly to competing CPU offerings on the same agentic AI workloads. As with early platform launches, readers should treat NVIDIA’s performance framing as marketing language until customers publish results of their own.

In the near term, the most important indicates to watch are whether cloud providers expand deployments of Vera beyond the initial machines delivered, and whether customers document real-world improvements for orchestration, tool-calling, and other CPU-heavy agent functions. NVIDIA’s inclusion of AWS, OCI, and multiple AI labs suggests it is seeking broad feedback from different workload patterns, and the next updates are likely to focus on deployment scale, integration experience, and system-level performance with Rubin GPUs in production settings.

Why It Matters

  • Agentic AI workloads can shift more compute pressure onto CPUs due to orchestration, tool calls, and long-context retrieval tasks, which could change how data center buyers think about server design.
  • If Vera’s claimed per-core and sustained-load performance translates into customer results, NVIDIA’s custom silicon approach could further tighten the coupling between CPUs and GPUs in AI infrastructure.
  • Cloud providers’ decisions to deploy Vera at scale would indicate whether the market views agentic AI as a durable driver for custom CPU platforms rather than a short-term experiment.
  • NVIDIA’s broader co-design ecosystem message suggests it wants customers to adopt integrated systems, which could influence competitive dynamics in networking, DPUs, and rack architecture alongside GPUs.

Sources

Key Facts

  • NVIDIA says it has begun shipping Vera, its first custom CPU built for agentic AI workloads.
  • AWS received its first Vera CPU server and Vera Rubin GPU, delivered by NVIDIA vice president Ian Buck at AWS headquarters in Seattle.
  • NVIDIA describes Vera as handling CPU-intensive agent tasks such as orchestration, tool-calling, long-context state management, and agent sandboxing.
  • Vera is specified by NVIDIA as having 88 custom Olympus cores and 1.2TB/s of memory bandwidth, with up to 1.8x faster per-core performance on agentic AI workloads in sustained load.
  • NVIDIA says Oracle Cloud Infrastructure plans to deploy hundreds of thousands of Vera CPUs beginning in 2026, according to comments from an OCI executive quoted in the blog.
  • NVIDIA describes Vera evaluation at AI labs and companies including Anthropic, OpenAI, and SpaceXAI, including reinforcement learning and agent-based simulation pipelines at SpaceXAI.

Technology Related

NVIDIA begins shipping Vera, its first agentic AI-focused custom CPU, with early deliveries to cloud and AI partners | The Apex Times