Cerebras just delivered a fascinating earnings report.
Core cloud revenue:
+272% YoY.
Core total revenue:
$230.9 MILLION.
Up:
70% YoY.
Core net profit:
$29.8 MILLION.
And management raised its full-year core revenue outlook to:
$940M–$960M.
The market loved it.
CBRS surged as high as:
$120.44.
But by Friday's close?
$108.72.
Still up 11.22%, but well below its intraday high.
And after hours?
Approximately $106.
Interesting.
Because while investors celebrated Cerebras's accelerating cloud business, Nvidia was dealing with its historic leadership transition.
So is Cerebras finally ready to challenge Nvidia?
I think it has a genuine opportunity.
But there's something hidden inside Cerebras's earnings that investors shouldn't overlook.
ITS TECHNOLOGY IS READY.
SOME OF ITS DATA-CENTRE CAPACITY ISN'T.
And that could determine where CBRS trades next.
🔥 FIRST, THE 272% GROWTH IS REAL. BUT LET'S GET THE NUMBERS RIGHT.
Cerebras reported two sets of results.
GAAP revenue: $210.8M, +55% YoY.
Core revenue: $230.9M, +70% YoY.
The famous 272% cloud-growth figure refers to its adjusted core measure.
GAAP cloud and other services revenue grew 265%.
Still exceptional!
But here's the distinction that matters.
Core results exclude items including stock-based compensation and customer warrant accounting effects.
On a GAAP basis, Cerebras actually recorded a:
$161.2M NET LOSS.
Despite reporting a $29.8M core net profit.
This doesn't make the growth meaningless.
But I wouldn't describe Cerebras as fully profitable just yet.
Its financial economics are improving.
Its path to sustainable GAAP profitability still needs proof.
And management's Q4 target is encouraging.
Core revenue: $310M–$330M.
Core gross margin: 47%–49%.
Core operating margin: approximately breakeven.
That is what I want to see.
NOT JUST FASTER AI CHIPS.
A BUSINESS MOVING TOWARD SUSTAINABLE PROFITS.
🧠 WHY CEREBRAS COULD BECOME A SERIOUS AI INFERENCE PLAYER
Nvidia built an extraordinary advantage in GPUs, networking and CUDA.
Cerebras is taking a different approach.
Its Wafer Scale Engine puts enormous computing capability on a single wafer-scale processor.
This architecture is designed to reduce certain data-movement bottlenecks and deliver extremely fast inference.
Why does that matter?
Because AI is changing.
Earlier, investors focused heavily on training large models.
Now the market is increasingly interested in:
AI AGENTS.
REAL-TIME REASONING.
LOW-LATENCY RESPONSES.
HIGH-VOLUME INFERENCE.
Cerebras claims its CS-4 can achieve up to 30 times faster inference than GPU-based alternatives in selected comparisons.
But I would be careful.
That is a workload-specific performance claim, not proof that Cerebras is 30 times better for every AI task.
Nvidia still has formidable advantages in software, scale, versatility and ecosystem adoption.
So my thesis isn't:
CEREBRAS WILL REPLACE NVIDIA.
It's:
CEREBRAS COULD WIN A VALUABLE SLICE OF THE AI INFERENCE MARKET.
And it doesn't need Nvidia to collapse for that to happen.
🚨 BUT HERE'S THE OPENAI PROBLEM THAT WORRIES ME
Cerebras has a major multiyear agreement with OpenAI involving:
750 MW OF COMMITTED COMPUTE CAPACITY.
That is enormous.
Its September-quarter filing also reported:
$26.1B IN REMAINING PERFORMANCE OBLIGATIONS.
But here's what many investors may miss.
Cerebras disclosed that it delivered less capacity to OpenAI than originally contemplated by September 30.
Why?
Management cited delays involving:
LAND.
POWER.
DATA-CENTRE FACILITIES.
Not a lack of computing demand.
And not, according to management, a failure of its AI chips.
But the result is still significant.
Cerebras expects to return to its planned capacity-delivery schedule in:
THE SECOND HALF OF 2027.
Delayed delivery can trigger service credits.
Under certain circumstances, affected capacity commitments may also be cancelled.
Management estimates less than 5% of committed megawatts could become subject to cancellation this year.
So this is not evidence that OpenAI has abandoned Cerebras.
But it is evidence that:
A SIGNED AI CONTRACT IS NOT THE SAME AS RECOGNISED REVENUE.
That's a lesson we've also seen across the AI cloud sector.
And it's why I wouldn't blindly buy CBRS just because its backlog is enormous.
🔵 THE BULL CASE IS GETTING STRONGER TOO
Now here's what makes the story interesting.
Cerebras announced new hyperscaler business.
Oracle plans to offer Cerebras inference through its cloud infrastructure in 2027.
Management also disclosed a Microsoft cloud agreement, with more details expected next year.
Meanwhile, Cerebras has secured approximately 700 MW of gross data-centre capacity targeted for delivery by the end of 2027.
These are important developments.
Because broader customer relationships reduce dependence on any single customer over time.
But again, future capacity is not operating capacity.
And future partnerships are not revenue already earned.
EXECUTION WILL DETERMINE HOW MUCH OF THE OPPORTUNITY BECOMES PROFIT.
🔮 MY CBRS PREDICTION: NOVEMBER 16–20
Friday close:
$108.72.
Friday high:
$120.44.
Friday low:
$101.18.
My subjective scenarios:
🟢 BULL CASE: 30%
Buyers defend $105–$108.
Positive momentum returns to AI infrastructure stocks.
CBRS reclaims $115 and challenges Friday's high.
Potential recovery:
$120–$128.
Above $120.44 with sustained volume would strengthen the bullish case.
🟡 BASE CASE: 45%
MY MOST LIKELY SCENARIO.
The earnings beat attracts buyers.
But OpenAI delivery delays, valuation concerns and profit-taking limit momentum.
Expected trading range:
$100–$120.
I expect sharp intraday swings rather than a smooth rally.
🔴 BEAR CASE: 25%
AI sentiment weakens.
Investors focus on infrastructure delays and GAAP losses.
CBRS loses $100 decisively.
Potential downside:
$90–$100.
These are scenario ranges, not guaranteed forecasts.
🎯 MY CBRS TRADING MAP
$105–$108
First stabilisation watch.
$100–$101
Important support test, close to Friday's intraday low.
$112–$115
First meaningful recovery confirmation.
$120.44
Friday's high and an important breakout reference.
$120–$128
Bull-case target.
Below $100
I become more defensive.
And I wouldn't forget:
NVIDIA REPORTS EARNINGS ON NOVEMBER 17.
A reassuring Nvidia report could revive sentiment across AI hardware.
But a disappointing report or further turbulence could pressure Cerebras too.
Cerebras is not automatically insulated from the wider AI cycle.
🔥 MY CALL: CEREBRAS COULD BE AN AI INFERENCE WINNER. BUT I WOULDN'T CHASE THE EARNINGS SPIKE.
I like Cerebras's growth.
I like the technological differentiation.
I like the direction of its core margins.
And I like that more cloud customers are embracing its technology.
But I also see substantial risks.
GAAP losses remain.
OpenAI deployments are behind schedule.
The company still needs to convert contracted capacity into recurring revenue.
And Friday's price action showed that sellers remain active above $110.
For next week?
MY BASE CASE: $100–$120.
Above $115 with strength?
I become more constructive.
Above $120.44?
A stronger breakout could be developing.
Below $100?
I reassess.
My bigger thesis is simple.
NVIDIA DOESN'T NEED TO LOSE FOR CEREBRAS TO WIN.
The AI market could become large enough for specialised inference architectures to gain meaningful share alongside GPUs.
But the stock market won't reward technological promise forever.
Eventually, it will demand:
COMPUTING SPEED → DEPLOYED CAPACITY → REVENUE → CASH RETURNS.
Cerebras has demonstrated impressive speed and demand.
Now it needs to demonstrate the rest.
That's what I will be watching.
I am not a financial advisor. Trade wisely, Comrades!
Would you buy CBRS after its earnings breakout, or wait for confirmation that its OpenAI capacity delays are being resolved?
If you find this post interesting, give it wings! 🪽 Repost and share the insights.
Do consider ‘Follow me’ and get firsthand read of my daily new post. Thank you.
@Tiger_SG @TigerEvents @Daily_Discussion @TigerStars @TigerPM
Comments