Cerebras Systems Inc. (NASDAQ: CBRS) and Cisco Systems Inc. (NASDAQ: CSCO) shares declined significantly on Thursday, despite both companies presenting fresh evidence of sustained demand in the AI infrastructure sector. Cerebras fell 13% and Cisco dropped approximately 8% at the time of writing, signaling a shift in investor priorities from raw demand metrics to growth acceleration and margin sustainability.

Strong Demand Metrics

Cerebras reported record core revenue of $209.9 million, representing a 103% increase year over year. Core cloud and services revenue surged 287%. The company’s remaining performance obligations reached $25.4 billion, prompting an upgrade to its full-year guidance.

Cisco also demonstrated robust order inflow, booking $4 billion in hyperscaler AI infrastructure orders during the fourth quarter. This brought fiscal 2026 total orders to $9.3 billion. Cisco expects $7.5 billion in hyperscaler AI revenue for fiscal 2027. Total revenue rose 18% to $17.25 billion, beating analyst estimates.

Growth Deceleration Concerns

The market reaction suggests investors are increasingly focused on the rate of acceleration rather than absolute growth figures. Jake Behan, Direxion’s head of capital markets, noted that Cisco entered earnings with considerable optimism already priced in, meaning strong results were viewed as confirmation rather than a new catalyst.

Cerebras guided third-quarter core revenue to between $214 million and $216 million. This implies just 2.4% sequential growth at the midpoint, a sharp deceleration from the triple-digit annual growth seen previously. Additionally, Cerebras expects core gross margin to contract to 38%-40% before recovering in the fourth quarter.

What the Numbers Show

The divergence between Cerebras’ annual performance and its forward guidance highlights a critical tension in the current AI trade. While the company achieved a 103% year-over-year revenue increase, the implied sequential growth of only 2.4% for the next quarter suggests that the base effect is fading faster than anticipated. This slowdown, combined with expected margin compression, indicates that execution risks are rising as the company scales capacity to meet its $25.4 billion backlog.

Market Sentiment and Outlook

Prediction-market data indicates skepticism about an imminent breakdown in the broader AI boom. Polymarket odds of a severe AI-industry downturn by year-end stood at 14%, down from 27% in late July, based on roughly $2.9 million in trading volume.

Morgan Stanley analysts stated that "execution remains the key debate" for Cerebras given the scale and speed of the capacity build required. Investor Dan Ives described Cisco and CoreWeave as "pieces of the puzzle" for the AI trade, arguing the narrative is shifting beyond capital expenditure toward monetization. The reactions to both companies underscore that investors now require evidence that strong demand translates into accelerating revenue growth and expanding margins.

Cerebras Systems (NASDAQ: CBRS) announced it is powering Ultrafast mode, a new service tier in the OpenAI API for GPT-5.6 Sol. The service is available initially in limited preview to OpenAI customers. Andrew Feldman, CEO and co-founder of Cerebras, stated that the partnership demonstrates that speed and intelligence are no longer mutually exclusive.

Ultrafast runs GPT-5.6 Sol at up to 750 output tokens per second. This represents processing speeds of up to 14 times faster than Standard processing. Sachin Katti, VP Compute Strategy & GPT-Infra at OpenAI, noted that the company is starting with a small group of customers to learn where that speed creates meaningful value before expanding the service.

Performance Benchmarks

The performance differential highlights the infrastructure advantage provided by Cerebras. While both modes deliver the same intelligence as GPT-5.6 Sol Standard, the Ultrafast tier achieves this frontier intelligence at significantly higher throughput. Based on output speeds for Anthropic models reported by Artificial Analysis, Ultrafast is 5x faster than Claude Opus 4.8 in Fast mode and 11x faster than Claude Fable 5.

Benchmarks focused on economically valuable work show how faster token generation translates into higher productivity:

• On Humanity’s Last Exam, a 2,500-question benchmark spanning graduate-level chemistry, economics and literature, GPT-5.6 Sol Ultrafast answered the full question set in just over 11 hours. This compares to more than three days of continuous compute for Claude Fable 5, with GPT-5.6 Sol Ultrafast reaching comparable accuracy nearly 7x faster.
• On GDP-Val, a benchmark of economically valuable knowledge-work tasks such as legal briefs, financial models, and engineering reports, Ultrafast delivered a 5.6x end-to-end speedup with no loss in quality.

What the Numbers Show

The data reveals a divergence between raw model capability and inference efficiency. By maintaining the same intelligence level as the Standard tier while achieving a 14x speed increase, Cerebras’ architecture effectively decouples latency from model size. The comparison with Claude models further contextualizes this advantage: being 5x faster than Claude Opus 4.8 suggests that Cerebras’ Wafer-Scale Engine architecture provides a distinct competitive edge in high-throughput inference scenarios where memory bandwidth is typically the bottleneck.

Technical Architecture

Ultrafast’s speed comes from Cerebras’ Wafer-Scale Engine architecture, which keeps model weights on-chip — 44 GB of SRAM on each wafer-sized chip — rather than shuttling them between on-chip memory and off-chip storage as GPU-based inference must. This eliminates the memory-bandwidth bottleneck that constrains frontier-model inference speed on conventional hardware.

To get notified when capacity expands to more customers, please visit the Cerebras website.