What Most People Miss About Jim Cramer and Cerebras Stock

Jim Cramer didn't doubt the silicon. He doubted the math.

When Cerebras Systems hit Nasdaq on May 14, 2026, euphoria was practically dripping off the trading floor. The inference-chip specialist priced its upsized IPO at $185 a share, raised $5.55 billion, and popped right out of the gate at $350—briefly pushing its valuation north of $100 billion on a fully diluted basis. Anyone watching Mad Money around that debut heard a familiar refrain from Cramer: the tech is real, the speed is eye-popping, and you're going to have to buy it up there without his blessing.

Six months later, that warning looks uncomfortably prescient. Trading right around $108 in mid-November 2026, Cerebras shares sit roughly 65% below their first-day close and well under the $185 IPO price. Yet the underlying business hasn't stalled out. Third-quarter core revenue jumped 70% year over year to $230.9 million, contracted future revenue sits at $26.1 billion, and management is guiding for fourth-quarter core revenue of $310 million to $330 million.

So where does Cramer actually stand on this newly public AI infrastructure stock now that the IPO fireworks have faded into a bruising autumn selloff? His stance has shifted from valuation skepticism to a much sharper focus on physical bottlenecks, customer timing, and whether AI demand outside Nvidia can survive broader market turbulence.

http://googleusercontent.com/lmdx_content/cEbkCQqKJjUdbSTtzXfzCPIyZSAcAMfpCdRLxogqLCvorctBezsfxWpNKeWFGteQNwJPaQvxuXvXYNKovnZTypaiGXTOBGwAWwoGZHdZImbUNzyovpXasWQDuXoRWpp55531


Why Jim Cramer Pushed Back on the IPO Frenzy

Opening-day hype has a nasty habit of turning great engineering stories into terrible entry prices. That was the core of Cramer's argument when Cerebras went public.

A company that generated roughly $510 million in 2025 revenue was suddenly being priced near $95 billion to over $100 billion depending on how you counted shares. Even if you love wafer-scale computing and fast token generation, paying a triple-digit multiple of sales leaves zero room for supply hiccups, delayed data center energization, or a shift in market mood.

I've watched plenty of hardware IPOs get priced for perfection on day one only to get repriced brutally once quarterly reality sets in. When an offering pops nearly 90% at the open, early backers and institutions have massive liquidity, while late retail buyers inherit all the multiple compression. Cramer's reluctance wasn't a dismissal of Cerebras's inference architecture. It was basic discipline around price versus near-term cash generation.

Speed Alone Doesn't Pay the Electric Bill

Here's the part casual observers keep tripping over. Making an inference chip run blazingly fast is an engineering triumph. Getting a 700-megawatt data center footprint powered, permitted, and online on schedule is a logistics war.

Cerebras has reportedly expanded its contracted power capacity to 700 megawatts for delivery by the end of 2027, up from 600 megawatts, with a return to full schedule targeted for the second half of 2027. Still, data center delays have been hanging over its largest customer deployment. Reported backlog already accounts for estimated delay credits, and management estimates less than 5% of committed capacity is at risk in 2026. Even so, analysts at firms like Morgan Stanley have flagged that slowdowns give customers room to weigh alternatives for fast inference.

Cramer has been hammering adjacent infrastructure realities across his autumn commentary—from permitting and grid constraints on data center expansions to broader market anxiety around how fast AI spending actually converts into returns. If a utility hookup slips by two quarters, your wafer-scale hardware sits waiting on power.

Reading Cramer's Nuanced Stance on AI Infrastructure

If you try to box Cramer into a simple "bullish" or "bearish" label on Cerebras or AI infrastructure right now, you'll miss half the picture. His commentary reads more like a triage nurse checking vitals during a storm.

On one hand, he has explicitly pointed out signs of AI momentum outside Nvidia—noting after mid-November results that both Cerebras and Cisco beat expectations, reinforcing that demand isn't strictly locked inside a single vendor. On the other hand, he has warned investors not to get fooled by brief bounces in a broader tech selloff where AI, software, and semiconductors have all taken hits.

The Hyperscaler Pivot vs. Execution Risk

What makes Cerebras genuinely interesting at depressed levels isn't nostalgia for May's $350 opening print. It's the commercial traction.

✨ Don't miss: The Name on the Tower

Landing cloud and hardware relationships across major players—including Microsoft alongside Amazon Web Services and Oracle—gives Cerebras credibility beyond early AI lab deals. Microsoft-related revenue is slated to begin in the first half of 2027, while Oracle hardware shipments are lined up starting in the fourth quarter. Management is even projecting revenue to triple in 2027, with fourth-quarter gross margin aiming near 48% and operating margin hovering near break-even.

That sounds encouraging on paper. Honestly, though, tripling revenue in a year is a high-wire act with almost no margin for error. Any slip in customer readiness, optical or component supply, or data center buildouts can send a volatile stock swinging 10% or 15% in a single session.

Where Opposing Views Clash

Wall Street isn't singing from one sheet of music on Cerebras or newly public AI infrastructure names. Bulls see a high-growth inference challenger trading at a fraction of its post-IPO peak with a $26.1 billion contracted revenue backlog and big-name cloud wins taking shape for 2027.

Skeptics—including cautious sell-side voices and macro-focused investors—worry about execution timelines, customer concentration risk during delays, and fierce competition across both custom silicon and established GPU ecosystems. Plus, when macro nerves flare around interest rates or major listings like Anthropic's market debut, high-beta hardware stocks often get sold indiscriminately alongside weaker names.

Practical Mistakes Investors Make With Newly Public AI Stocks

Chasing newly public infrastructure stocks is one of the fastest ways to turn conviction into regret if you skip the fundamentals. Most mistakes come down to three habits:

  • Mistaking a hot IPO debut for proven unit economics. First-day pops measure scarcity and momentum, not sustainable margins or free cash flow.
  • Ignoring physical infrastructure bottlenecks. Silicon doesn't run in a vacuum; power availability, cooling, and site readiness dictate when revenue actually gets recognized.
  • Sizing a volatile turnaround like a blue-chip core holding. A stock swinging sharply after earnings or macro headlines demands disciplined position sizing.

If you're looking at Cerebras right now, don't anchor to the $350 debut price—and don't assume a 65% drop from that first close automatically makes it risk-free. Track the concrete milestones instead: fourth-quarter core revenue landing inside the $310 million to $330 million range, gross margins holding near 48%, updates on the 700-megawatt data center schedule, and actual revenue ramp from new hyperscaler contracts in 2027. Scale in gradually only if the execution matches the guidance, and keep enough cash on hand that a choppy market won't force your hand.

Jim Cramer highlights the biggest players in internet infrastructure

This video offers helpful context on how Cramer evaluates cloud and data-center infrastructure beneficiaries amid surging AI-driven network traffic.

LY

Lily Young

With a passion for uncovering the truth, Lily Young has spent years reporting on complex issues across business, technology, and global affairs.