1. Semiconductor Wafers (5-nanometer)
Source TSMC is the sole manufacturer of Cerebras chips. Cerebras uses TSMC's 5-nanometer node, which results in lower wafer costs and less constrained supply compared to 3-nanometer nodes.
Confidence: high
CBRS
T3Cerebras Systems Inc.
AI '25: Semis & MemoryAI '25: Servers, Storage & Compute SystemsAI '26: Intelligence Infrastructure SupercycleAI '26: Neoclouds
Loading chart…
OverviewCerebras Systems Inc. develops AI compute platforms with its Wafer-Scale Engine for superior, fast AI inference. Offering hardware systems and cloud services, Q
Cerebras Systems Inc. develops AI compute platforms with its Wafer-Scale Engine for superior, fast AI inference. Offering hardware systems and cloud services, Q2 2026 revenue was approximately 61% cloud/services and 39% hardware. The company serves hyperscalers like OpenAI and AWS, and is expanding with new disaggregated solutions via AMD, enhancing throughput for high-demand AI applications and unlocking new markets.
Key Inputs And Sourcing
2. Data Center Space & Power
Source Data center space and power are identified as the primary bottleneck for the industry and Cerebras's growth. Cerebras has secured over 600 megawatts of data center capacity by the end of 2027. Power consumption is a significant factor, with Cerebras systems having a 25kW power draw per node.
Confidence: high
3. Contract Manufacturing Services
Source Cerebras is expanding manufacturing capabilities with partners like Flex and Sanmina, expecting a more than 10x increase in manufacturing capacity in 2026. Flex facilities in Milpitas, California, are scaling production of CS-3 systems.
Confidence: high
4. Skilled Engineering Labor (R&D)
Source Cerebras emphasizes continuous invention and engineering execution for new technology and systems. The company invests in 'world-class people'. Average annual total compensation for employees is around $265k, with machine learning engineers earning an average of $291k.
Confidence: high
5. Rented Cerebras Systems (temporary)
Source Cerebras temporarily rents back some of its own systems from cloud customers to meet overwhelming demand, which reduces gross margin.
Confidence: high
6. Other Semiconductor Components
Source While Cerebras avoids HBM and CoWoS, its systems still require various other semiconductor components for system design, packaging, I/O, and power delivery.
Confidence: medium
7. Logistics & Shipping
Source Global expansion of data centers and manufacturing implies significant logistics and shipping costs.
Confidence: medium
Industry Publications
- The Next Platform (nextplatform.com) — Provides deep insights into high-performance computing, AI infrastructure, data center architectures, and custom silicon, directly relevant to Cerebras's core business and technology.
- SemiAnalysis (semianalysis.com) — Offers expert analysis on semiconductor technology, AI chips, advanced packaging, supply chain dynamics, and AI infrastructure economics, with specific coverage on Cerebras and its competitive landscape.
- HPCwire (hpcwire.com) — A leading publication dedicated to the high-performance computing sphere, covering technologies, improvements, and trends in interconnect technologies, large-scale AI workloads, and hardware/software advancements.
- ServeTheHome (servethehome.com) — Focuses on servers, storage, and networking hardware, providing practical reviews and discussions relevant to data center deployments and infrastructure components used by Cerebras.
- Data Center Dynamics (DCD) (datacenterdynamics.com) — Offers in-depth reporting on sustainability, security, technology trends, and industry shifts in the rapidly evolving field of AI and HPC data centers, directly addressing Cerebras's primary growth constraint.
Economic Data Watch
1. PwC, Morgan Stanley, Goldman Sachs — Global Data Centre Outlook / Hyperscaler Capital Expenditure Forecasts
Metric/field Annual Data Center Capital Expenditure (USD billions)
Cadence Annually/Quarterly
Why it matters Direct indicator of investment in AI infrastructure, which drives demand for Cerebras's products and services.
Signal to watch Accelerating growth is bullish.
Confidence: high
2. 451 Research (S&P Global Energy Horizons), IDCA, JLL — Global Data Center Services & Infrastructure Market Monitor & Forecast / Global Data Center Report
Metric/field Global Data Center Grid Power Demand (Gigawatts)
Cadence Annually/Quarterly
Why it matters Data center capacity is a primary constraint for Cerebras's growth. Increased capacity enables faster deployment and revenue recognition.
Signal to watch Accelerating growth in power capacity is bullish.
Confidence: high
3. World Semiconductor Trade Statistics (WSTS) — Global Semiconductor Market Forecast
Metric/field Global Semiconductor Sales (USD billions)
Cadence Quarterly/Annually
Why it matters Reflects the overall health and demand in the semiconductor industry, which is foundational for AI infrastructure.
Signal to watch Sustained strong growth is bullish.
Confidence: high
4. U.S. Census Bureau / Associated Builders and Contractors (ABC) — Construction Spending Survey (C-30) / Private Nonresidential Construction Spending
Metric/field Value of Construction Put in Place: Private Nonresidential, Data Centers (Millions of Dollars, Seasonally Adjusted Annual Rate)
Cadence Monthly
Why it matters Direct measure of data center build-out in the US, a key market for Cerebras.
Signal to watch Sustained or accelerating growth is bullish.
Confidence: high
5. TrendForce, DRAMeXchange — DRAMeXchange Spot and Contract Price Trends
Metric/field DRAM and HBM Average Selling Price (ASP) Year-over-Year % Change
Cadence Weekly/Monthly
Why it matters Indicates supply/demand balance and cost pressures for competitors, which can influence pricing power and market dynamics in the AI compute space.
Signal to watch Declining prices for HBM could signal easing supply constraints for competitors (bearish for Cerebras's relative advantage), while rising prices could reinforce Cerebras's supply chain advantage (bullish).
Confidence: medium
Free Alt Data Watch
1. Google Trends — Web Search Interest
Metric/field Search volume index for 'AI data center'
Cadence weekly
Why it matters Reflects general public and industry interest in the core demand drivers of the theme.
Signal to watch Rising search interest is bullish.
Confidence: high
2. AltIndex (or direct Reddit monitoring) — Reddit Mentions & Sentiment Tracker
Metric/field Daily mention volume and sentiment score for 'Cerebras Systems' (CBRS)
Cadence daily
Why it matters Gauges retail investor interest and public perception, which can influence short-term stock movements and broader awareness.
Signal to watch Increasing mention volume with positive sentiment is bullish.
Confidence: medium
3. GitHub (via OSSInsight or direct API) — Trending AI Repositories / GitHub Topics
Metric/field Number of stars/forks/commits for top 'AI inference' or 'AI framework' repositories (e.g., DeepSpeed, OpenVINO, TensorRT)
Cadence daily/weekly
Why it matters Indicates developer interest and innovation in the broader AI inference ecosystem, which could indirectly benefit Cerebras through increased demand for fast inference hardware.
Signal to watch Sustained growth in activity is bullish.
Confidence: medium
4. U.S. Bureau of Labor Statistics (BLS) / U.S. Census Bureau — Quarterly Workforce Indicators (QWI) / Occupational Employment Statistics (OES)
Metric/field Employment in Data Centers (Number of Workers)
Cadence Quarterly/Annually
Why it matters A growing workforce in data centers indicates ongoing expansion and operational scaling of AI infrastructure.
Signal to watch Sustained growth in employment is bullish.
Confidence: high
5. Labrynth's Red Tape Index (or similar news aggregators with sentiment analysis) — Data Center Media Sentiment tracker
Metric/field National media tone score for 'AI infrastructure expansion' or 'data center development' (e.g., on a scale of -100 to +100)
Cadence daily/weekly
Why it matters Provides a real-time view of sentiment around data center development, which can impact permitting, community consent, and investment flows.
Signal to watch Improving or positive sentiment is bullish (less friction for build-out).
Confidence: medium
Paid Alt Data Watch
1. Orbital Insight, Planet Labs — Global Data Center Construction Monitoring
Metric/field New Data Center Construction Starts (Count, Square Footage, Gigawatt Capacity) in key regions (e.g., US, Europe, Canada)
Cadence monthly/quarterly
Why it matters Provides a leading indicator of physical capacity expansion, which directly impacts Cerebras's ability to deploy systems and recognize revenue.
Signal to watch Increasing construction activity is bullish.
Confidence: high
2. Revelio Labs, Thinknum — Company Job Postings Data
Metric/field Total active job postings for 'Cerebras Systems' (CBRS) and specific roles like 'Data Center Engineer', 'AI Engineer', 'Manufacturing Operations'
Cadence weekly/monthly
Why it matters Indicates internal scaling efforts, particularly in critical areas like data center operations and manufacturing, which are key growth enablers.
Signal to watch Sustained increase in relevant job postings is bullish.
Confidence: high
3. ImportGenius, Panjiva — Global Shipment Data
Metric/field Import/export volume and value of '5nm silicon wafers' (for TSMC to Cerebras manufacturing partners) and 'AI server racks' (to Cerebras data center locations)
Cadence monthly
Why it matters Provides insights into the physical supply chain health and potential bottlenecks or accelerations in manufacturing and deployment.
Signal to watch Increasing shipments of relevant components is bullish.
Confidence: medium
4. Similarweb, Sensor Tower — Website Traffic Analytics / App Usage Data
Metric/field Monthly unique visitors / total visits to 'openai.com' (and subdomains like 'platform.openai.com', 'chat.openai.com') and 'aws.amazon.com/machine-learning' (or relevant AWS AI service pages)
Cadence monthly
Why it matters Directly reflects user adoption and engagement with the AI services provided by Cerebras's key partners, indicating underlying demand for Cerebras's inference compute.
Signal to watch Sustained growth in traffic/usage is bullish.
Confidence: high
5. Bloomberg Terminal, Refinitiv Eikon, S&P Capital IQ — Consensus Analyst Estimates / Company Financials
Metric/field Consensus Capital Expenditure (CapEx) forecasts for 'Microsoft Corp (MSFT)', 'Alphabet Inc (GOOG)', 'Amazon.com Inc (AMZN)', 'Meta Platforms Inc (META)' (specifically for AI/Cloud infrastructure)
Cadence quarterly
Why it matters Hyperscalers are the primary drivers of AI infrastructure spending. Tracking their CapEx plans provides a forward-looking view of demand for AI compute.
Signal to watch Upward revisions to CapEx forecasts are bullish.
Confidence: high
Search Keywords Brand Product
- Cerebras CS-3
- Cerebras CS4
- Cerebras CS5
- Wafer-Scale Engine
- Cerebras Cloud
- Cerebras AI Model Studio
- MemoryX
- Cerebras disaggregated inference
- AI inference speed
- wafer-scale architecture
- AI compute platform
- AI data centers
- large language models
- frontier models
- AI productivity
- AI security applications
- agentic AI
- AI cloud services
Search Keywords Event Phrases
- Cerebras Q2 2026 earnings
- Cerebras Supernova Conference
- Cerebras IPO
- What They Do (Plain English & Analogies)
- Cerebras Systems builds specialized supercomputers and chips designed specifically for Artificial Intelligence (AI). Unlike traditional computers that use many small, interconnected chips, Cerebras uses one giant chip, called a Wafer-Scale Engine (WSE), which is essentially an entire silicon wafer. This massive chip acts like a single, enormous brain for AI, allowing it to process information much faster and more efficiently, especially for very large AI models. Think of it like a super-fast, dedicated highway for AI calculations, instead of a regular road with many traffic lights and intersections, enabling AI to get answers much quicker. This speed is critical for 'inference,' which is when AI models use their training to make predictions or generate content, like quickly answering your questions or powering AI agents. They offer these powerful systems for customers to install in their own data centers (on-premise) or provide access to their computing power as a cloud service. They are also pioneering 'disaggregated inference,' where they combine their super-fast chips with other processors (like GPUs) to get both extreme speed and high overall capacity.
- Very Brief History
- Cerebras Systems Inc. was founded in 2015 and is headquartered in Sunnyvale, California. The company has focused on developing its unique wafer-scale technology for AI, releasing its first Wafer-Scale Engine (WSE-1) in 2019, followed by the WSE-3 in 2024, which powers their current CS-3 system. The company recently completed its public offering in Q2 2026.
- "Street Stereotype"
- The 'street stereotype' for Cerebras Systems is that of a disruptive innovator in the AI compute space, known for its 'speed wins' philosophy and its unique wafer-scale architecture that challenges traditional GPU-based AI infrastructure. They are perceived as a high-growth company with significant potential in the rapidly expanding AI market, particularly for fast inference. The recent successful IPO and substantial Remaining Performance Obligations (RPO) further solidify their image as a serious player, though customer concentration and the ability to rapidly scale data center capacity remain key discussion points among investors and analysts.
- Subsidiaries On Linked In*
- {"subsidiaries":[]}
- Customer Sectors & Example Clients
- Cerebras serves a diverse clientele including leading hyperscalers, advanced foundation model laboratories, AI-native and digital-first businesses, large enterprises, and key players in Sovereign AI initiatives. Specific top clients mentioned are OpenAI, Amazon Web Services (AWS), Figma (AI coding), Cognition (AI coding), Lovable (AI coding in Europe), Block (agentic flows), AlphaSense (agentic flows), GSK (agentic flows), and CrowdStrike (security).
- New Customers / Segments They'Re Targeting
- Cerebras is actively expanding its customer base by accelerating AI productivity in existing markets like coding and agentic flows. They are also pioneering entirely new areas where speed unlocks opportunities, such as security, citing their recent win with CrowdStrike as an application that only exists if AI is fast. Furthermore, they are targeting other hyperscalers beyond OpenAI and AWS, with discussions ongoing, and are seeing opportunities with emerging 'neocloud' providers who are diversifying their hardware vendors and seeking multi-vendor solutions.
- Sales Geographies And Expansion Plans
- Cerebras currently has operations spanning the United States (with data centers either up or under contract in Alabama, Dallas, Denver, Minneapolis, Santa Clara, Stockton), Canada (Manitoba, Montreal, Saskatchewan, Toronto), Europe (France, Finland, Norway), and the Middle East and Africa. The company is actively expanding its data center capacity globally, having secured over 600 megawatts of data center capacity by the end of 2027, with a pipeline measured in gigawatts. They also have offices in Sunnyvale, San Diego, Toronto, and Bangalore, India, and are in early discussions for data centers in Israel, the UAE, Australia, Singapore, India, and Indonesia.
- How Key Themes May Help/Hurt
- Cerebras is significantly impacted by the 'AI '26: Intelligence Infrastructure Supercycle' theme. The accelerating and insatiable demand for AI compute and networking hardware (Bull1) directly benefits Cerebras, as their fast inference solutions are in high demand, leading to record core revenue and a massive $25.4 billion in Remaining Performance Obligations (RPO). The broadening monetization of AI across diverse industries (Bull2), particularly in new applications like security and agentic flows, expands Cerebras's Total Addressable Market (TAM). The accelerating critical infrastructure build-out, especially in data centers and power (Bull3), is crucial for Cerebras as they are aggressively securing and building out data center capacity globally to meet demand, positioning them to be among the largest non-hyperscale AI clouds. Their unique supply chain, which does not rely on HBM memory, CoWoS packaging, or 3-nanometer fab capacity, makes them more resilient to the persistent and severe supply chain constraints (Bear2 for the theme) that plague the broader industry. However, despite these advantages, Cerebras's growth is primarily constrained by the availability of data center space and power, which is a key challenge in the current market, potentially limiting their ability to fully capitalize on demand and realize revenue from large contracts.
3 Main Long-Term Bull Details
- Unmatched Speed and Wafer-Scale Architecture: Cerebras delivers the fastest AI inference in the world by an order of magnitude due to its unique wafer-scale architecture, which eliminates traditional chip-to-chip communication bottlenecks and utilizes high-speed SRAM. This fundamental architectural advantage is durable and expanding, enabling faster token generation and greater AI productivity across all model sizes and for latency-sensitive workloads.
- Transformative Strategic Partnerships and RPO: Definitive agreements with OpenAI (for over $25 billion in compute RPO) and AWS (for deployment in their data centers, combining Trainium with CS systems for disaggregated inference) provide unprecedented revenue visibility, market validation, and direct insight into frontier model development, positioning Cerebras at the forefront of AI innovation.
- Supply Chain Resilience and Disaggregation Strategy: By not relying on HBM memory, CoWoS packaging, or the most constrained 3-nanometer fab capacity, Cerebras avoids critical supply chain bottlenecks that impact competitors. Their use of SRAM and 5-nanometer technology, coupled with U.S. manufacturing, provides a significant advantage in scaling production. Furthermore, their disaggregation strategy with AMD and AWS expands their market opportunity and improves the economics of inference.
3 Main Long-Term Bear Details
- Data Center Capacity as a Binding Constraint: Despite strong demand and supply chain advantages for their chips, the primary constraint on Cerebras' growth is the availability of data center space and power. The company is aggressively building out capacity, but this remains a significant challenge in the current market, potentially limiting their ability to fully capitalize on demand and realize revenue from large contracts.
- Temporary Gross Margin Pressure: To accelerate capacity deployment and meet significant near-term demand, Cerebras is temporarily renting back its own systems from an existing customer. This additional cost will depress core cloud and other services gross margins in Q3 2026 before expected recovery in Q4 2026 and beyond.
- Intense Competition and Market Adoption Challenges: The AI compute market remains intensely competitive, dominated by established GPU players like NVIDIA. While Cerebras highlights its speed advantage, convincing a broader market to adopt a specialized or disaggregated architecture, especially given potential higher token costs for some workloads, presents an ongoing challenge against entrenched solutions and could limit broader market penetration.
- Competitors And Differentiation
- Cerebras competes with established GPU players like NVIDIA, as well as other chip companies like AMD and hyperscalers developing their own silicon (e.g., AWS with Trainium). Their primary differentiation lies in their unique wafer-scale architecture (Wafer-Scale Engine - WSE) which uses SRAM directly on the logic wafer, enabling significantly faster AI inference (10x faster for GPT-5.6 Sol) compared to HBM-based GPU solutions. This architecture avoids critical supply chain constraints like HBM memory and CoWoS packaging, and operates on the less constrained 5-nanometer node, where wafers are less expensive and supply is less constrained. They are also differentiating through 'disaggregated inference' solutions, partnering with companies like AMD (Helios racks) and AWS (Trainium) to combine their speed with higher throughput from other processors, expanding market opportunities and improving the economics of inference.
- Recent Performance & What The Market'S Focused On
- Cerebras had a strong Q2 2026, delivering record core revenue of $209.9 million, up 103% year-over-year, and beating guidance on all metrics including core gross margins and core operating margin. Core cloud and other services revenue nearly quadrupled, increasing by 287% year-over-year. The company successfully completed its public offering and ended the quarter with over $8.6 billion in cash. Management raised full-year 2026 guidance for core revenue to a range of $880 million to $890 million, core gross margin to 41% to 43%, and core operating margin to negative 19% to negative 17%. The market is focused on Cerebras's ability to more than triple core revenues in 2027 and continue significant growth in subsequent years, driven by its $25.4 billion in Remaining Performance Obligations (RPO). Key areas of market focus include the continued build-out of data center capacity (with over 600 megawatts secured by the end of 2027 and a gigawatt pipeline), the ramp-up of the OpenAI deployment, the commencement of AWS solutions via the Bedrock platform in Q1 2027, the commercialization of disaggregated inference solutions (expected to be deployed and available in Q4), and the recovery of gross margins as lower-cost owned systems replace rented capacity. The upcoming unveiling of the CS4 and the planned launch of the CS5 in the second half of 2027 are also key points of interest.
- Revenue Segments And Estimated Mix
- Core Cloud and Other Services Revenue — Mix: ~60.8%; Source: Q2 2026 transcript; Trend: Up 287% year-over-year; expected to see increasing year-over-year growth rates for each quarter in 2026, with more revenue later in the year as cloud capacity deployments accelerate. Expected to improve significantly in Q4 2026 as more Cerebras-owned systems come online.
- Core Hardware Revenue — Mix: ~39.2%; Source: Q2 2026 transcript; Trend: Up 17% year-over-year; expected to decrease sequentially for the next few quarters as production shifts to Cerebras Cloud.
- Product Brands
- CS-3
- CS4
- CS5
- Wafer-Scale Engine
- WSE-3
- Cerebras Cloud
- Cerebras AI Model Studio
- MemoryX
Bull / Bear DetailsCerebras Systems, an AI infrastructure innovator, is poised for massive growth, driven by its wafer-scale architecture delivering unparalleled inference speed a
Thesis
Cerebras Systems, an AI infrastructure innovator, is poised for massive growth, driven by its wafer-scale architecture delivering unparalleled inference speed and efficiency. Strategic partnerships with OpenAI (>$25B RPO) and AWS, alongside new disaggregated solutions with AMD, validate its technology and expand market reach. Despite near-term data center capacity bottlenecks and temporary margin pressures, aggressive capacity build-out and a robust product roadmap position Cerebras to capture a substantial share of the expanding, speed-driven AI compute market. (Updated: 2026-09-09)
Bull case
Cerebras maintains its industry-leading AI inference speed, demonstrated by 10x faster GPT-5.6 Sol serving. New disaggregated solutions with AMD and AWS will increase throughput by 5x, while the internal roadmap aims for 20x throughput and annual speed doubling by the end of 2027, significantly enhancing economics and expanding the total addressable market.
The $25 billion RPO, primarily from OpenAI, provides exceptional revenue visibility and market validation. New partnerships with AWS for Bedrock (Q1 2027 GA) and AMD for disaggregated solutions, along with emerging hyperscaler and enterprise deals, are diversifying Cerebras' customer base and positioning it for sustained, multi-year revenue growth.
Cerebras is aggressively addressing data center constraints, having secured over 600MW by the end of 2027 with a gigawatt pipeline, and increasing manufacturing capacity 10x in 2026. Its 5nm TSMC node and avoidance of HBM/CoWoS bottlenecks ensure supply chain resilience, enabling rapid scaling to meet surging demand and capitalize on its RPO.
Bear case
Despite aggressive efforts, data center space and power remain the industry's primary bottleneck, directly impacting Cerebras' ability to deploy solutions and realize revenue from its substantial RPO. Delays in data center build-out could slow growth and market penetration, despite strong demand and an otherwise robust supply chain for its core technology.
Cerebras anticipates Q3 2026 to be the low point for core gross margins due to the higher cost of renting third-party capacity to meet immediate cloud demand. While expected to recover significantly in Q4 2026 and beyond, this temporary pressure and current negative operating margins could impact near-term profitability and investor sentiment.
The AI compute market remains intensely competitive, dominated by established GPU players. While Cerebras offers a significant speed advantage, convincing a broader market to adopt its specialized or disaggregated architecture, especially against entrenched solutions and potential concerns about total cost of ownership, presents an ongoing challenge to market penetration.
Bull / Bear Case
- Bear Case
- Despite aggressive efforts, data center space and power remain the industry's primary bottleneck, directly impacting Cerebras' ability to deploy solutions and realize revenue from its substantial $25 billion RPO. Delays in data center build-out could slow growth and market penetration, despite strong demand. Cerebras anticipates Q3 2026 to be the low point for core gross margins due to the higher cost of renting third-party capacity, impacting near-term profitability and investor sentiment, especially with current negative operating margins. The AI compute market is intensely competitive, dominated by established GPU players, and convincing a broader market to adopt its specialized or disaggregated architecture against entrenched solutions presents an ongoing challenge to market penetration.
- Bull Case
- Cerebras Systems is poised for massive growth, driven by its industry-leading AI inference speed, demonstrated by 10x faster GPT-5.6 Sol serving, and its unique wafer-scale architecture. Strategic partnerships, including a $25 billion RPO with OpenAI and new disaggregated solutions with AWS (Q1 2027 GA) and AMD (Q4 2026 deployment), provide exceptional revenue visibility and diversify its customer base across hyperscalers, enterprises, and new markets like security. The company is aggressively addressing data center constraints, having secured over 600MW by the end of 2027 with a gigawatt pipeline, and increasing manufacturing capacity 10x in 2026. Its 5nm TSMC node and avoidance of HBM/CoWoS bottlenecks ensure supply chain resilience, enabling rapid scaling to capitalize on surging demand and its robust RPO, with expectations to more than triple core revenues in 2027.
- More Compelling & Why
- Bear. The stock's significant underperformance post-earnings suggests the market is heavily discounting the near-term operational challenges. The primary bottleneck of data center capacity, acknowledged by management, directly impedes revenue realization from the substantial RPO, while temporary gross margin pressure and negative operating margins further weigh on profitability. This makes the current valuation, likely a high Price/Sales ratio given the growth expectations, appear stretched. My view would flip if Cerebras consistently demonstrates faster-than-expected data center deployments, leading to accelerated revenue growth above guidance, and a clear, sustained recovery in gross margins towards its 60%+ target.
Key Factors
| Key Factor | Why It Matters | What To Watch | What It Signals | Where/How To Track | Free Alt Data | Paid Alt Data |
|---|---|---|---|---|---|---|
| Full Year 2026 Core Revenue Guidance Achievement & 2027 Revenue Growth Outlook | This factor provides an overarching view of Cerebras' financial performance and its ability to execute on its ambitious growth plans, including the expectation to more than triple core revenues in 2027. | Any updates or revisions to the full year 2026 core revenue guidance ($880 million to $890 million) in Q3 2026 earnings. Commentary on the 2027 revenue growth outlook (more than 3x core revenues) in subsequent calls. | Bullish: An upward revision of the full year 2026 core revenue guidance, or reporting full-year revenue at the high end or above the current range. Strong reaffirmation or further increase in the 2027 revenue growth outlook. Bearish: A downward revision of the 2026 guidance, or reporting full-year revenue below the low end of the current range. Any softening of the 2027 growth expectations. | Cerebras Q3 2026 earnings call (expected November 2026) and Q4 2026 earnings call (expected February 2027). | Financial news aggregators (e.g., Bloomberg, Reuters) for analyst consensus updates. | FactSet/Refinitiv Eikon: Analyst consensus estimates for revenue and growth rates. |
| Core Cloud and Services Revenue Year-over-Year Growth Rate | This segment is central to Cerebras' long-term strategy, reflecting the success of strategic partnerships (OpenAI, AWS) and the rapid ramp-up of Cerebras Cloud. It is a direct indicator of RPO conversion and the demand for fast inference. | The reported Core Cloud and Services revenue year-over-year growth rate in Q3 and Q4 2026 earnings reports. | Bullish: Growth rate continues to accelerate or remains significantly above Q1 2026's 167% and Q2 2026's 287%, indicating strong demand and effective capacity deployment. Bearish: Growth rate decelerates significantly or fails to meet management's expectations for continued acceleration. | Cerebras Q3 2026 earnings report (expected November 2026) and Q4 2026 earnings report (expected February 2027). | Company investor relations website for historical financial data. | Bloomberg Terminal/Refinitiv Eikon: Consensus estimates for cloud revenue growth. |
| AWS Disaggregated Inference Solution General Availability via Bedrock | This partnership expands Cerebras' market opportunity, provides global enterprise reach through a trusted industry leader, and is a significant driver for 2027 revenue growth, not yet reflected in the current RPO. | Public announcements from Cerebras or AWS detailing the general availability of Cerebras' disaggregated inference solutions on AWS Bedrock in Q1 2027. Initial customer adoption and any early revenue contributions mentioned in Q1 2027 earnings. | Bullish: General availability announced in Q1 2027 as planned, with positive commentary on customer traction or specific revenue contribution. Bearish: Delays in general availability beyond Q1 2027 or lack of significant customer adoption. | Company press releases, AWS announcements, Cerebras Q1 2027 earnings call (expected around May 2027). | AWS Blog/News, Google Trends: 'AWS Bedrock Cerebras', 'Cerebras AWS partnership'. | Apptopia/Sensor Tower: AWS Bedrock app/service usage trends (if applicable). Thinknum: AWS job postings mentioning 'Cerebras' or 'wafer-scale engine'. |
| Data Center Capacity Additions & Core Cloud Gross Margin Recovery | Data center capacity is the primary bottleneck for Cerebras' growth, directly impacting its ability to convert its $25.4 billion RPO into revenue. Gross margin recovery is crucial for profitability as the company scales its cloud services. | Management updates on new data center capacity brought online (e.g., progress on the 600+ megawatts secured by end of 2027, or gigawatt pipeline conversion). Reported Core Cloud and Services gross margin in Q3 and Q4 2026 earnings reports. | Bullish: Core Cloud and Services gross margin reported above 40% in Q3 2026 (after Q3 is the low point), with clear trajectory towards 60%+ in Q4 2026 and beyond, or faster-than-expected data center deployments. Bearish: Margins remain depressed below 38% in Q3 or decline further, or reported delays in data center build-out schedules. | Company earnings calls and press releases (Q3 2026 earnings expected November 2026, Q4 2026 earnings expected February 2027). | Google Trends: 'AI data center construction', 'data center power capacity'. Industry news sites (e.g., The Next Platform, Data Center Dynamics) for general data center build-out news. | Orbital Insight/Planet Labs: New Data Center Construction Starts (Count, Square Footage, Gigawatt Capacity) |
| Deployment of Disaggregated Inference Solutions with AMD and other GPUs | Disaggregation expands the market for Cerebras by enabling GPUs to participate in fast inference and improving throughput by up to 5x, leading to better economics (more tokens per system, lower cost/power per token). | Public announcements from Cerebras regarding the deployment and availability of disaggregated inference solutions with AMD (Helios racks) or other GPUs in Q4 2026. Any specific customer wins or early performance metrics. | Bullish: Deployment announced in Q4 2026 as planned, with positive commentary on performance, customer interest, or early revenue. Bearish: Delays in deployment beyond Q4 2026 or lack of clear market traction. | Cerebras press releases, Supernova Conference announcements (next week), Q3 2026 earnings call (expected November 2026). | Tech news outlets (e.g., AnandTech, ServeTheHome, The Register) covering Cerebras/AMD announcements. Google Trends: 'Cerebras AMD disaggregation'. | Gartner/IDC reports on AI infrastructure trends, specifically disaggregated architectures. |
Key Reported Metrics, Reratings Triggers & ResultsManagement explicitly stated Q3 will be the low point for core gross margin due to temporary costs. Its performance will be closely watched for signs of recover
Upcoming print · 2026-11-12
| Key reported metrics | ||
|---|---|---|
| Metric | Last period | Why it matters |
| Core Gross Margin | 30.13% | Management explicitly stated Q3 will be the low point for core gross margin due to temporary costs. Its performance will be closely watched for signs of recovery towards the long-term 60%+ target, indicating improving profitability. |
| Core Cloud and Other Services Revenue YoY Growth | $127.7 million (287% y/y growth) | This segment is central to Cerebras' long-term strategy, reflecting the success of strategic partnerships like OpenAI and AWS, and the rapid ramp-up of Cerebras Cloud. Continued acceleration indicates effective capacity deployment. |
| Core Revenue | $82.1 million (17% y/y growth) | Core Revenue is the primary top-line indicator of Cerebras' overall business performance and market demand. Meeting or exceeding guidance signals strong execution and validates the company's growth trajectory amidst aggressive capacity expansion. |
Last reported · 2026-08-12
| Key reported metrics | Rerating thresholds | Earnings results | ||||||
|---|---|---|---|---|---|---|---|---|
| Metric | Last period | Why it matters | What's needed for rerating | Rerating context | Earnings date | Actual reported | Hit target? | Notes |
| Core Hardware Revenue YoY Growth | 60% | This metric indicates demand for on-premise hardware deployments and the ongoing shift in revenue mix as production is increasingly deployed into Cerebras Cloud. | Core Hardware Revenue YoY Growth needs to be reported above 45%. This would indicate a less severe sequential decrease than anticipated, or stronger-than-expected resilience in on-premise hardware demand, defying the company's guidance for decreasing hardware revenue over the next few quarters. | A Core Hardware Revenue YoY Growth above 45% would signal sustained demand for Cerebras's on-premise hardware, even as the company prioritizes its cloud services. This would alleviate concerns about the pace of hardware revenue decline and validate the continued market penetration of its core technology, supporting a higher valuation. | $82.1 million (17% y/y growth) | No | The actual reported growth of 17% year-over-year for Core Hardware Revenue was below the rerating trigger of 45%. This aligns with management's expectation for decreasing hardware revenue sequentially as production shifts towards Cerebras Cloud deployments. | |
| Core Cloud and Other Services Revenue YoY Growth | 167% | This segment's growth is crucial as it reflects the success of strategic partnerships like OpenAI and AWS, and the rapid ramp-up of Cerebras Cloud, which is central to long-term strategy. | Above 175% YoY growth. | This metric directly validates Cerebras' core investment thesis: its ability to scale high-margin cloud services through strategic partnerships and overcome data center capacity constraints. Exceeding 175% YoY growth would signal strong execution in a critical segment, confirming demand for its unique wafer-scale architecture and providing a clear path to long-term profitability and market share gains in the rapidly expanding AI compute market. | $127.7 million (287% y/y growth) | Yes | Core Cloud and Other Services Revenue grew by 287% year-over-year, significantly exceeding the rerating trigger of 175%. This strong performance reflects tremendous demand for Cerebras' fast inference service and the rapid ramp-up of its private cloud business. | |
| Core Total Revenue YoY Growth | 92% | This metric reflects the overall top-line performance and the company's ability to meet or exceed its Q2 guidance, indicating strong market demand and execution. | Core Total Revenue YoY Growth of 95% or more. This would significantly exceed the company's implied Q2 2026 guidance of approximately 88% and surpass the Q1 2026 growth of 94%, signaling re-acceleration and strong execution. | Achieving this growth validates Cerebras's ability to convert strategic partnerships and unique technology into accelerating top-line growth, despite data center constraints. It signals strong execution, reinforcing the bullish thesis of capturing a substantial share of the expanding AI compute market, driving higher valuation multiples. | $209.9 million (103% y/y growth) | Yes | Cerebras reported a Core Total Revenue growth of 103% year-over-year, surpassing the rerating trigger of 95%. This indicates strong overall top-line performance and execution, with the company delivering record core revenue and beating guidance on all metrics. | |
Key QuestionsWill Cerebras meet or exceed its Q3 2026 core revenue guidance of $214 million to $216 million, driven by continued data center build-out and the ramp of its cl
Will Cerebras meet or exceed its Q3 2026 core revenue guidance of $214 million to $216 million, driven by continued data center build-out and the ramp of its cloud and services segment?
- Question 2
How will Cerebras' Q3 2026 core gross margins be impacted by the temporary costs of renting third-party capacity, and will management demonstrate a clear path for margin recovery towards its long-term 60%+ target in Q4 2026 and beyond?
- Question 3
Will Cerebras successfully deploy its disaggregated inference solutions with AMD (Helios racks) and other GPUs in Q4 2026, and will this demonstrate tangible progress towards the Q1 2027 general availability on AWS Bedrock?
Earnings Transcript Summary
· 2026Q2 Earnings Call
| 3 Things Management Is Most Focused On | Call Takeaway & Tone | Prior Quarter'S Y/Y Growth By Segment | 3 Things Analysts Most Pressed On (And Mgmt Responses) | Revenue Segments |
|---|---|---|---|---|
| 3 Things Management Is Most Focused On1. **Expanding Capacity:** Management is intensely focused on securing and building out data center capacity globally, increasing manufacturing capacity (10x in 2026), and ensuring supply chain resilience (e.g., TSMC wafers, avoiding HBM/CoWoS). This is crucial to meet the $25 billion RPO and enable massive revenue growth in 2027 and beyond. 2. **Advancing Capabilities:** Cerebras is focused on inventing new technology to extend performance, throughput, and power efficiency, including delivering new systems that double speed annually, increasing throughput by more than 20x by the end of 2027, and pioneering disaggregated inference solutions with partners like AMD and AWS. 3. **Expanding Customer Base and Market Opportunities:** Management is focused on leveraging existing partnerships (OpenAI, AWS) and expanding into new high-growth areas like security, AI coding, and agentic flows, which are unlocked by fast inference. They are also pursuing opportunities with other hyperscalers and neoclouds. | Call Takeaway & ToneThe overall takeaway of the call was highly positive and confident. Cerebras delivered a strong Q2 2026, beating guidance on all metrics (core revenue, gross margins, operating margins), and successfully completed its IPO. Management emphasized that 2026 is a 'foundation building year' with significant progress in expanding capacity, advancing capabilities (including new product roadmap and disaggregated inference), and growing its customer base. The company expects to more than triple core revenues in 2027 and continue exceptional growth in subsequent years, with significant margin expansion. The tone was optimistic, highlighting the 'unbound demand for fast inference' and Cerebras' unique position to capitalize on it, despite acknowledging data center capacity as a key bottleneck they are actively addressing. | Prior Quarter'S Y/Y Growth By SegmentCore hardware revenue was up 60% year-over-year in Q1 2026. Core cloud and other services revenue grew 167% year-over-year in Q1 2026. | 3 Things Analysts Most Pressed On (And Mgmt Responses)1. **Customer Concentration:** Analysts questioned the concentration of revenue from OpenAI and AWS for 2027. Management responded that while OpenAI will remain a meaningful portion, AWS and other rapidly growing customers in coding and security will increase their percentage of revenue over time, leading to diversification. 2. **Economics and Go-to-Market for AWS and AMD Disaggregated Solutions:** Analysts asked about the economics and timeline for the AWS and AMD partnerships. Management stated that the AWS service will be live in Q1 2027 via Bedrock, and for AMD, the joint solution (Helios for prefill, Cerebras for decode) is extremely strong, delivers vastly faster speed and more throughput, and they already have buyers. 3. **Impact of 20x Throughput Increase on Disaggregated Compute Need:** Analysts inquired if Cerebras' internal 20x throughput increase by the end of 2027 would reduce the need for disaggregated solutions. Management explained they are exploring all ways to drive throughput up, and that increasing throughput (whether internally or through disaggregation) makes each system more profitable, generates more tokens, and reduces cost and power per token. Disaggregated inference with GPUs will be deployed and available in Q4. | Revenue SegmentsCore total revenue was $209.9 million, up 103% year-over-year. Core cloud and other services revenue was $127.7 million, up 287% year-over-year. Core hardware revenue was $82.1 million, up 17% year-over-year. |
· 2026Q1 Earnings Call
| 3 Things Management Is Most Focused On | Call Takeaway & Tone | Prior Quarter'S Y/Y Growth By Segment | 3 Things Analysts Most Pressed On (And Mgmt Responses) | Revenue Segments |
|---|---|---|---|---|
| 3 Things Management Is Most Focused On1. **Maintaining AI inference speed leadership and wafer-scale architecture advantages:** Management emphasized that Cerebras delivers the fastest AI in the world, by an order of magnitude, across all model sizes and for latency-sensitive workloads, attributing this to their unique wafer-scale architecture which avoids HBM and CoWoS constraints and uses 5-nanometer node technology. 2. **Expanding strategic partnerships and customer adoption:** The company highlighted definitive agreements with OpenAI for over $20 billion of compute and AWS for deploying Cerebras solutions in their data centers, seeing these as transformative for revenue and market positioning. 3. **Aggressively building out data center capacity:** Management is focused on rapidly adding data center space globally to meet significant demand, stating that data center capacity is the primary constraint on growth. | Call Takeaway & ToneThe overall takeaway of the call is highly positive and confident, emphasizing Cerebras Systems' strong Q1 2026 performance, significant technological advantage in AI inference speed, and transformative strategic partnerships with OpenAI and AWS. Management conveyed an optimistic tone regarding the long-term market opportunity driven by AI's expanding footprint, despite acknowledging that data center capacity is currently the primary constraint on their rapid growth. The company is aggressively investing in infrastructure to meet robust demand and expects continued strong revenue growth, though with some temporary gross margin compression due to initial costs of scaling cloud capacity. | Prior Quarter'S Y/Y Growth By SegmentInformation on the prior quarter's (Q4 2025) year-over-year growth for specific segments (Core hardware revenue and Core cloud and services revenue) is not explicitly available in the provided source material or search results. The available information indicates Cerebras Systems' total revenue for FY2025 was $510 million, up 76% year-over-year. | 3 Things Analysts Most Pressed On (And Mgmt Responses)1. **Timing and supply for AWS partnership:** Analysts inquired about the timing of the AWS agreement's impact and Cerebras' ability to supply. Management responded that they have sufficient supply for their 2026 plan and beyond, and expect AWS's impact to be seen in 2027. 2. **Total Addressable Market (TAM) for fast inference and token pricing:** Analysts questioned the TAM for fast inference and whether customers are willing to pay a premium for faster tokens. Management asserted that the entire inference market is available to Cerebras for fast inference, drawing parallels to the market's preference for fast search and internet access, and noted that fast tokens are currently sold at a premium because they are more valuable. 3. **Constraints on growth (data centers vs. wafer capacity):** Analysts asked if the constraint on growth was 5-nanometer wafer capacity, space and power for cloud build-out, or other factors. Management explicitly stated that demand is not the constraint, supply is not the constraint, but rather, the constraint is data centers. | Revenue SegmentsCore total revenue was $191.3 million, representing 92% year-over-year growth. Core hardware revenue was $111.6 million, up 60% year-over-year. Core cloud and other services revenue reached $79.8 million and grew 167% year-over-year. |
Transcript Tidbits
| About Expanding Eligible Market | About Competition | About The Broader Industry | Where Things Are Headed | Updates On Theme | Broader Themes Emerging | Bullish-Leaning Quotes (Short) | Bearish-Leaning Quotes (Short) | Hiring |
|---|---|---|---|---|---|---|---|---|
| About Expanding Eligible MarketFast inference unlocks new applications and new markets, with security being a key example where fast AI enables new applications like invisible, AI-based security devices. AI coding and agentic flows are also rapidly growing markets. Disaggregation with AMD and AWS expands the market for both GPU providers and Cerebras, enabling GPUs to participate in fast inference and Cerebras to reach more price-sensitive customers. The AWS partnership provides global reach through an industry leader. | About CompetitionCerebras serves OpenAI's GPT-5.6 Sol 10x faster than competitors, laying to rest concerns about supporting large frontier models. This collaboration provides a strategic advantage previously only available to NVIDIA. Cerebras' SRAM-based architecture supports blisteringly fast tokens, unlike HBM-based architectures (used by GPUs and ASICs) which force a trade-off between throughput and speed. Cerebras avoids critical supply chain constraints like HBM memory, CoWoS packaging, or 3-nanometer fab capacity, which impact competitors. | About The Broader IndustryData center space continues to be the bottleneck for the entire industry. The AI '26: Intelligence Infrastructure Supercycle is characterized by unprecedented and escalating capital expenditures by hyperscalers, persistent and severe supply chain constraints for critical AI components (like HBM and advanced packaging), and insatiable demand for AI compute and networking hardware. | Where Things Are HeadedCerebras expects to more than triple core revenues in 2027 and continue multi-year growth. The company is securing gigawatts of data center capacity globally and plans to increase manufacturing capacity by over 10x in 2026 and further in 2027. New systems are expected to double speed annually for several years, starting with a 15x advantage, and throughput will increase by more than 20x over the next 18 months. The CS4 will be unveiled at the Supernova Conference, and the CS5 is on track for launch in the second half of 2027. Disaggregated inference solutions with AMD are ready for deployment in Q4, and AWS solutions will be generally available in Q1 2027 via Bedrock. First revenue from other hyperscalers is expected mid-2027, ramping through 2028 and beyond. | Updates On ThemeIntelligence | Broader Themes EmergingSpeed as a Competitive Differentiator in AI; Disaggregated AI Compute Architecture; Data Centers in Space. | Bullish-Leaning Quotes (Short)Q2 was a strong quarter. We delivered record core revenue and beat guidance on all metrics. We see unbound demand for fast inference. We expect to more than triple our core revenues in '27 and continue to grow at multiples in the years following. Our data center pipeline of new opportunities for expansion continues to grow and is now measured in gigawatts. We expect to deliver new systems that double our speed each year for the next several years. Our $25 billion in RPO does not reflect any backlog of business from AWS or any other hyperscaler at this time. Core revenue was $209.9 million, up 103% year-over-year. We ended Q2 with more than $8.6 billion in cash. | Bearish-Leaning Quotes (Short)Data center space continues to be the bottleneck for the entire industry, and we are no exception. In the short term, it reduces gross margin as we have a higher cost for this rented capacity. We expect Q3 to be the low point for core gross margin. Core operating loss was $33.6 million. Core operating margin was negative 16%. | HiringToday, we are investing in world-class people, manufacturing and data center capacity and company infrastructure to support the significant increase in scale we expect to deliver over the next several years. |
| About Expanding Eligible Market | About Competition | About The Broader Industry | Where Things Are Headed | Updates On Theme | Broader Themes Emerging | Bullish-Leaning Quotes (Short) | Bearish-Leaning Quotes (Short) |
|---|---|---|---|---|---|---|---|
| About Expanding Eligible MarketAI provides new capabilities to computers, giving them 'purchase on whole swaths of the world that had previously been foreclosed,' which Cerebras believes increases the market addressable to compute by 'many thousands of times'. AI opens up the world of human experience to computers, exponentially increasing the market size. Text, images, video, agents, and robotics are all identified as opportunities for Cerebras as AI expands computers' ability to understand, participate, and take actions in the world. The company views the entire inference market as available for fast inference, stating, 'who doesn't want answers in less time? And who doesn't want more productive agents?' | About CompetitionCerebras claims to deliver the fastest AI in the world 'by an order of magnitude' compared to GPUs, solving problems that competitors couldn't. Their wafer-scale architecture, with chips 58x larger than the largest competitor, allows them to use SRAM for blistering speed, while competitive offerings use HBM, which is described as slow, expensive, and in short supply. In a demo comparing Cerebras to a leading GPU-based inference cloud (understood to be using B300s), Cerebras completed a task 13x faster. Cerebras is one of only two hardware vendors currently serving OpenAI models. The company notes that the GPU architecture struggles with the sequential nature of decode, where Cerebras excels. Cerebras also observed that competition has increased pricing due to higher costs for HBM and other components, which has raised the market floor. | About The Broader IndustryThe AI revolution is driving a 'tectonic shift in compute demand' because AI makes the world tractable for computers, leading to a need for 'vastly more compute' for decades. AI has become 'profoundly useful' since 2025, spurring an explosion in compute demand. The industry faces significant supply chain challenges, particularly with HBM memory, which is in short supply and expensive, and the CoWoS process at TSMC, as well as 3-nanometer capacity. Data center capacity is currently at a premium, described as 'a dog fight out there.' Despite this, AI power users represent less than 1% of the world's population, yet compute and memory are already in tight supply. | Where Things Are HeadedCerebras expects to expand its performance lead with next-generation solutions, as wafer-scale technology fundamentally advantages future technologies like memory stacking and optical integration, and even data centers in space. The company is actively working with OpenAI to bring GPT 5.5 onto Cerebras as the next phase of their rollout. Cerebras' partnership with AWS will combine Trainium 3 chips with Cerebras' CS-3 in a disaggregated solution, expected to deliver an order of magnitude faster tokens at massive throughput, with AWS's impact anticipated in 2027. Cerebras anticipates decreasing hardware revenue for the next few quarters as production shifts towards deploying hardware in Cerebras Cloud to fulfill significant contracts. The company has an 'exciting product road map' over the next several years, including disaggregated inference solutions with multiple hardware partners expected in the second half of this year. Cerebras aims to achieve a long-term profitability profile of approximately 60% gross margin and 40% operating margin. The company is aggressively adding data center space globally, with new data centers coming online in Q3, Q4 of 2026, and throughout 2027. | Updates On ThemeSemis | Broader Themes EmergingSpeed as a Competitive Differentiator in AI; Disaggregated AI Compute Architecture; Data Centers in Space. | Bullish-Leaning Quotes (Short)We delivered core revenue of $191.3 million, up 92% year-over-year. Cerebras delivers the fastest AI in the world, bar none, not by a little bit, but by an order of magnitude. We signed a definitive agreement with OpenAI on December 24, 2025, for the purchase of more than $20 billion of Cerebras compute over the next several years. GPT 5.4 on Cerebras is currently available to OpenAI engineers and to select OpenAI customers. We ended the quarter with $3.3 billion in cash, cash equivalents, restricted cash and marketable securities. We completed the largest semiconductor IPO in history, raising another $6.4 billion. Demand is not the constraint. | Bearish-Leaning Quotes (Short)We plan to see decreasing hardware revenue for the next few quarters. The additional cost of renting third-party capacity will depress core cloud and other services margin temporarily. Core operating margin in the range of minus 30% to minus 32%. Data center capacity is at a premium. It's a dog fight out there. The constraint is data centers. |
Upcoming Events
| Catalyst ID | Estimated Timing | Estimated Date Start | Estimated Date End | Catalyst | Why It Matters | Ticker Or Theme Specific | Transcript Date | Source Type |
|---|---|---|---|---|---|---|---|---|
| CBRS_fa4e3403 | Next week at our annual Supernova Conference | 2026-09-14 | 2026-09-18 | Cerebras will unveil the CS4, its fourth-generation system, at the annual Supernova Conference, with expectations for significant product announcements. | This event will showcase Cerebras' latest hardware advancements, potentially demonstrating increased speed and throughput, which could enhance its competitive position and drive future sales and market interest. | Ticker | 2026-08-12 | earnings_transcript |
| CBRS_67724dec | Q1 2027 | 2027-01-01 | 2027-03-31 | Cerebras expects its solutions to be generally available through AWS' Bedrock platform. | This partnership expands Cerebras' market opportunity and provides global reach through a leading hyperscaler, potentially driving significant revenue growth from a broad base of enterprise customers. | Ticker | 2026-08-12 | earnings_transcript |
| CBRS_8fb2061f | Q4 '26 | 2026-10-01 | 2026-12-31 | Core gross margin is expected to improve significantly in Q4 2026 as Cerebras brings on more data centers filled with lower-cost Cerebras-owned systems. | This improvement indicates enhanced profitability and operational efficiency, validating the strategy of replacing higher-cost rented capacity with owned systems, which is crucial for achieving long-term margin expansion targets. | Ticker | 2026-08-12 | earnings_transcript |