Key Takeaways
- Cerebras (CBRS) shares have retreated significantly from their post-IPO peak of $300, hovering near $211 in Monday’s early premarket session
- Six Wall Street firms launched coverage Monday with Buy or Outperform recommendations and price objectives spanning $250 to $300
- The AI chipmaker boasts a $24.6 billion order backlog, primarily driven by a substantial agreement with OpenAI and an additional contract with AWS
- Cerebras’ flagship Wafer-Scale Engine represents the world’s largest commercial chip, specifically designed for rapid AI inference operations
- Trailing twelve-month revenue reached $510 million with 76% year-over-year expansion, though shares command a premium 225x earnings valuation
The market debut of Cerebras Systems was nothing short of spectacular. Following last month’s initial public offering, the artificial intelligence semiconductor company watched its shares surge beyond $300 almost instantly. That euphoria proved short-lived.
Shares of CBRS changed hands around $211.80 during Monday’s premarket hours, reflecting a 5.4% intraday gain while remaining substantially below those initial peaks. The stock absorbed a 6.7% decline the prior week amid a broader selloff across semiconductor equities.
Enter Wall Street’s research analysts — and their timing couldn’t be more deliberate. A coordinated series of coverage launches suggests the Street believes current pricing presents opportunity.
Mizuho analyst Vijay Rakesh broke cover first, assigning an Outperform recommendation alongside a $300 price objective. Wedbush’s Matt Bryson matched that enthusiasm with a Buy rating and $270 target. Barclays established an overweight stance at $280. Both UBS and Rosenblatt arrived at $300 targets. Morgan Stanley took the most measured approach, initiating overweight coverage with a $250 price goal.
That’s a unanimous chorus of optimism from major investment banks in a single trading session.
Cerebras’ Technological Edge
The investment thesis centers on the company’s unique silicon approach. Cerebras manufactures the Wafer-Scale Engine — recognized as the largest commercially available chip on the planet. This semiconductor was purpose-built for AI inference, the computational stage where trained models generate real-world outputs.
Where Nvidia’s strategy relies on connecting thousands of discrete GPUs through intricate networking infrastructure, Cerebras consolidates everything onto a single wafer-sized chip. This design eliminates substantial computational overhead and accelerates token generation speeds.
“With the industry focused on inference to deliver Agentic AI solutions, we see Cerebras well-positioned as the industry leader in fast inference,” Mizuho’s Rakesh wrote.
Wedbush characterized the architecture as truly differentiated and observed that the marketplace is “now learning to pay for speed” in AI inference applications.
Massive Backlog Meets Concentration Questions
Cerebras disclosed a revenue pipeline totaling $24.6 billion at the close of 2025. That figure commands attention. The complication: a substantial portion originates from a single customer relationship with OpenAI.
This customer concentration has generated investor skepticism. Yet the company secured an additional agreement with Amazon Web Services, introducing a second blue-chip customer that provides some counterbalance to concentration concerns.
Wedbush’s Bryson positioned it this way: “With a differentiated architecture, a step-change in contracted revenue from OpenAI and AWS, and a market only now learning to pay for speed, we see an asymmetric, upside-skewed setup.”
His $270 valuation derives from applying a 40x multiple to his 2028 earnings projection, adjusted for net cash holdings.
Trailing twelve-month sales totaled $510 million, representing 76% growth compared to the prior year. The equity trades at a 225x earnings multiple, a level InvestingPro identified as exceeding fair value estimates.
By mid-morning Monday, CBRS shares traded approximately 14% higher at the premarket price near $211, based on refreshed market data.



