Signals Inbox·August 26, 2026·AI Infrastructure

Is Lambda really worth $3B today?

Lambda looks comfortably worth more than $3 billion today, but the proposed $12 billion valuation is harder to justify: the evidence currently points closer to an $8 billion to $10 billion company.

Q1Is Lambda actually worth $3B today?

Lambda is clearly worth more than $3 billion based on its latest financing history; the real question now is whether investors should pay $12 billion or more.

The confusion comes from Lambda's newest fundraising talks. Bloomberg reports that the AI cloud company is discussing a round of up to $3 billion at a valuation of $12 billion or higher. Lambda has already received multiple term sheets, although the deal has not closed and the final terms can still change.

That proposed price would be a huge step up. Lambda raised $320 million at a $1.5 billion valuation in February 2024, $480 million at $2.5 billion in February 2025, then more than $1.5 billion in November 2025. Forge puts that last Series E at a $5.9 billion post-money valuation.

So a $3 billion Lambda would actually look cheap today. At $12 billion, the debate becomes much more interesting because investors would be doubling Lambda's last primary valuation in less than a year.

Lambda funding and valuation history

Q2How did Lambda go from $1.5B to a possible $12B so fast?

Lambda's valuation has increased roughly eightfold in about two and a half years, which is unusually aggressive even by AI infrastructure standards.

Investors did not suddenly discover Lambda. The company was founded in 2012 and had been selling AI infrastructure years before ChatGPT. What changed was the scale of the market around it. Lambda went from a company known for selling GPU servers and relatively small cloud deployments to one signing multibillion-dollar capacity agreements.

Still, the private market gives us a reason to resist the $12 billion number. Forge currently shows Lambda shares at $39.32. Earlier this summer, Forge priced the shares at $49.25 and said that level implied a valuation of about $9 billion. The current share price is roughly 20% lower.

Secondary prices are messy because different share classes, liquidity and transaction sizes can affect them. We would not turn $39.32 into a precise company valuation. But the direction is useful: people actually trading Lambda shares lately have been paying less than they were when the implied value was around $9 billion.

That makes $12 billion look like a genuine financing premium rather than a price the wider private market has already accepted.

Q3How much revenue is Lambda actually making now?

Lambda is expected to generate more than $1.5 billion of revenue this year, but that number is still a forecast rather than audited public-company revenue.

We have a fairly useful trail behind it. Sacra estimated Lambda generated about $425 million in 2024. The Information later reported more than $520 million of revenue during the 12 months from October 2024 through September 2025, alongside a loss of roughly $175 million.

The same reporting said third-quarter sales were up 80% year over year. Bloomberg's latest reporting now says Lambda is expected to exceed $1.5 billion of revenue this year.

Those periods are not perfectly comparable, so we should not claim that revenue has literally tripled year over year. But the order of magnitude has clearly changed. Lambda was a business doing a few hundred million dollars of annual revenue relatively recently and is now trying to become a billion-dollar-plus cloud provider.

That is a much stronger foundation than Lambda had when investors valued it at $1.5 billion or $2.5 billion.

Q4Is $12B expensive for Lambda's current revenue?

At more than $1.5 billion of expected revenue, a $12 billion Lambda would trade at just under 8x sales, which is expensive for infrastructure but far less extreme than the headline valuation jump suggests.

Using Lambda's older $520 million trailing-revenue figure produces a multiple above 23x. That denominator is already too stale for a company growing this quickly.

The more interesting comparison is with Lambda's previous financing. Forge puts the Series E valuation at $5.9 billion. Against the $520 million reported for the 12 months ending shortly before that round, investors were already paying roughly 11x trailing revenue.

If Lambda reaches its latest revenue forecast, the company could double in valuation while its revenue multiple actually falls. The $12 billion price still assumes excellent execution, but investors are not simply doubling what they pay for each dollar of Lambda revenue.

Lambda revenue against a $12B valuation

Q5Is Lambda still growing fast enough to support a $12B valuation?

Yes, Lambda is currently growing fast enough to make $12 billion plausible, but there is very little room for an ordinary cloud slowdown.

The cleanest measured growth figure we have is the 80% year-over-year increase in third-quarter sales reported by The Information. That is exceptional at Lambda's scale.

The change in cloud revenue is even more striking when we zoom out. Forge, citing earlier reporting, says Lambda generated an estimated $20 million of revenue in 2022. A few years later, the business has crossed into hundreds of millions of dollars of reported annual sales and is chasing much larger contracted deployments.

A cloud business growing 20% would struggle to defend an 8x forward-sales multiple. At 80%, the argument is much easier.

The risk starts when Lambda becomes larger. Maintaining 80% growth after reaching billion-dollar scale requires adding enormous amounts of power, GPUs and customer spending every year. We know Lambda can grow unusually fast. We do not yet know whether it can compound at anything close to this rate for several more years.

Q6Does CoreWeave make Lambda look overpriced?

Yes. CoreWeave is currently the strongest reason to call Lambda's proposed $12 billion valuation expensive.

CoreWeave is worth about $49 billion on the public market these days. The company recently raised its full-year revenue guidance to $12.4 billion to $13.2 billion, which puts its market capitalization at roughly 3.8x the midpoint of expected revenue.

Lambda would be closer to 8x on its latest forecast. Investors would therefore be paying around twice CoreWeave's revenue multiple for the smaller private company.

And CoreWeave is hardly a slow incumbent that deserves an obvious discount. Its latest quarter produced $2.58 billion of revenue, up 112% year over year. Revenue backlog reached approximately $104 billion, with more than $25 billion of additional commitments added early in the following quarter.

There are good reasons for some difference in valuation. CoreWeave expects $35 billion to $39 billion of capital expenditure this year and lost $626 million in its latest quarter. Public investors can see exactly how expensive this growth is.

But the gap is hard to ignore. Lambda is asking private investors to pay a premium multiple over a larger competitor that is currently growing even faster and has more than $100 billion of reported backlog.

Q7Does Nebius show that Lambda could be worth much more?

Yes. Nebius shows that public investors will pay dramatically more than 8x revenue when a neocloud is growing at an exceptional rate.

Nebius currently has a market capitalization of roughly $60 billion. Its latest quarterly revenue reached about $575 million, up 514% year over year. Simply annualizing that quarter gives around $2.3 billion of revenue, which would put the company at roughly 26x that revenue level before making adjustments for cash, debt or its other businesses.

The comparison is imperfect, but the gap itself is useful. Public markets are clearly willing to value certain AI infrastructure companies far above the 4x range implied by CoreWeave.

Nebius has given investors unusually strong reasons to do that. Its latest quarter produced about $236 million of adjusted EBITDA, its growth is several times faster than Lambda's latest reported year-over-year rate, and it has been building a much broader full-stack cloud platform.

CoreWeave and Nebius work well as bookends. CoreWeave shows what public markets can pay for a huge, capital-intensive GPU cloud. Nebius shows how far that multiple can expand when growth becomes extraordinary. Lambda currently sits somewhere between those two stories.

Q8Is Lambda more expensive than other private AI clouds?

Lambda would sit near the expensive end of recent private AI cloud valuations at $12 billion, although investors are clearly willing to pay similar prices elsewhere.

Together AI is one of the freshest comparisons. The company recently raised $800 million at an $8.3 billion valuation after reporting more than $1.15 billion of annual bookings. That works out to roughly 7.2x bookings, although bookings should not be treated as revenue.

Crusoe was valued above $10 billion in its last major financing. The Information reported that the company was projecting roughly $2.3 billion of revenue this year, which would put that valuation near 4.3x projected sales.

Fluidstack provides another useful reference. The company recently raised $830 million at a $7.5 billion valuation after much more aggressive valuations had circulated during earlier fundraising talks.

Across those deals, investors are paying heavily for AI infrastructure while still discriminating between companies. A $12 billion Lambda would be expensive. It would not be some completely isolated private-market anomaly.

Recent private AI cloud valuation references

Q9Do Microsoft and Nvidia prove Lambda has real demand?

Yes, Lambda has enough signed business from Microsoft and Nvidia to show that its current growth is backed by very large customers spending real money on compute.

Microsoft signed a multiyear, multibillion-dollar agreement with Lambda covering tens of thousands of Nvidia GPUs, including GB300 NVL72 systems. The exact contract value was not disclosed, but "multibillion-dollar" immediately puts it in a different category from a normal enterprise-cloud contract.

Nvidia separately agreed to rent 10,000 GPUs from Lambda for $1.3 billion over four years, according to The Information. Another agreement worth roughly $200 million covers 8,000 additional Nvidia GPUs.

Taken together, those contracts help explain why Lambda can raise so much money today. Building GPU infrastructure before finding customers is dangerous. Building it against multiyear commitments from Microsoft and Nvidia gives lenders and equity investors something much more tangible to underwrite.

The Microsoft contract is particularly convincing because the wider market shows the same behavior. Microsoft has also signed huge AI-capacity deals with companies including CoreWeave, IREN, Nebius and Nscale. Lambda is participating in a real procurement wave rather than relying on one strange customer decision.

Q10Is Lambda too dependent on Nvidia, Microsoft and Amazon?

Yes, Lambda's customer concentration is high enough that we would discount some of the excitement around its growth.

The Information reported that Amazon and Microsoft together accounted for just under half of Lambda's $114 million of cloud revenue in one reported quarter. Nvidia subsequently became Lambda's largest customer through its own GPU-rental agreements.

That makes the Nvidia relationship particularly complicated. Nvidia supplies Lambda with the GPUs, owns equity in Lambda and now pays Lambda to rent infrastructure built around Nvidia GPUs. Nvidia's researchers genuinely need compute, so the revenue has an underlying service attached to it. Still, we would value $1 of revenue from a diversified group of independent customers more highly than $1 produced inside that tightly connected ecosystem.

The pattern is wider than Lambda. Nvidia has been investing in, financing and buying capacity from several companies across the AI infrastructure market. That strategy helps create alternatives to AWS, Azure and Google Cloud while increasing demand for Nvidia hardware.

For Lambda investors, the important question now is customer diversification. We already know that a few huge buyers can fill Lambda's clusters. We need to see whether the next wave of capacity attracts a much broader group of AI labs and enterprises.

Q11Can Lambda actually make good money renting GPUs?

We still do not have enough public financial data to know whether Lambda can earn consistently attractive returns on its GPU infrastructure. This is probably the biggest hole in the $12 billion case.

The Information reported roughly $175 million of losses during a 12-month period with more than $520 million of revenue. Losing an amount equal to about one-third of revenue is substantial, even for a company growing this quickly.

Lambda's financing tells us something else about the economics. The company recently expanded a senior secured credit facility to $1 billion. It then priced another $926 million secured term loan at SOFR plus three percentage points. That second transaction was oversubscribed and received a Baa2 investment-grade rating from Moody's.

Lenders clearly believe contracted GPU deployments can support large amounts of debt. Those deals also show how much outside capital Lambda needs to turn demand into revenue.

CoreWeave offers a useful warning about what happens at scale. Its latest quarter generated $1.51 billion of adjusted EBITDA on $2.58 billion of revenue, yet $640 million of net interest expense helped push the company to a $626 million net loss.

Renting GPUs can generate excellent operating cash flows when utilization is high, but financing costs, depreciation and the next hardware upgrade cycle still have to be paid. Lambda should not be valued like software until the economics actually look like software.

Q12Is the AI cloud market big enough for Lambda to keep growing?

Yes, the AI cloud market is growing fast enough that Lambda can reach several billion dollars of revenue without needing an unrealistic share of the category.

Gartner's latest forecast puts spending on AI-optimized infrastructure as a service at $42.3 billion this year, up 96.4% from the previous year. Gartner expects another jump to $66.1 billion next year.

The composition of that spending is getting more interesting too. Gartner expects inference to account for $23.3 billion this year versus $19 billion for training. Inference should reach 59% of the category next year as more AI systems move from experiments into applications that run continuously.

That is good news for Lambda because demand no longer rests entirely on a handful of frontier labs training giant models. Production inference can create a larger and more repetitive compute market.

The recent company numbers point in the same direction. CoreWeave has just reported triple-digit growth at multi-billion-dollar scale, while Nebius has been growing several times faster. Several providers are expanding simultaneously rather than fighting over a stagnant pool of revenue.

Market size gives us little reason to reject Lambda's valuation today. The harder problem is whether Lambda can capture that growth at good enough returns.

Q13What can Lambda do that CoreWeave or AWS cannot easily copy?

Lambda's strongest advantage today is execution: getting new Nvidia systems, financing them and turning them into working AI clusters quickly.

That skill is more valuable than it sounds. Lambda recently announced a 100-plus-megawatt AI factory in Kansas City that will initially include more than 10,000 Nvidia GB300 GPUs. It has also been among the early operators deploying new Nvidia architectures in production.

Lambda has spent more than a decade working specifically around machine learning infrastructure, giving it credibility with developers and AI teams that a generic cloud provider had to build later.

Yet we do not see a deep proprietary moat around the underlying compute. CoreWeave can buy Nvidia systems. Nebius can build highly optimized AI infrastructure. AWS, Microsoft and Google can spend tens of billions of dollars on their own clouds while also designing custom accelerators.

The Nvidia relationship helps Lambda move faster, but Nvidia wants several healthy clouds buying its chips. It has strong incentives to support CoreWeave, Nebius, Together AI and other customers too.

Lambda's edge currently looks operational rather than exclusive. That can produce a very large company during a capacity shortage. It is less reassuring when thinking about long-term pricing power once GPU supply becomes easier to obtain.

Q14How much revenue would make Lambda's $12B valuation look reasonable?

Lambda needs around $2 billion to $3 billion of annual revenue for a $12 billion valuation to start looking normal beside public AI infrastructure companies.

At 6x revenue, Lambda needs $2 billion. At 5x, it needs $2.4 billion. At 4x, which is close to where CoreWeave trades on current guidance, Lambda needs $3 billion.

Those numbers are reachable from Lambda's current trajectory. Moving from $1.5 billion to $2.4 billion would require about 60% growth. Reaching $3 billion would require roughly another doubling.

This is probably the simplest way to frame the valuation. Investors paying $12 billion today are effectively betting that Lambda can add somewhere around $500 million to $1.5 billion of annual revenue before its growth rate falls sharply.

Revenue required to support a $12B valuation

Q15What has to go right for Lambda to actually deserve $12B?

Lambda can make $12 billion look cheap if its next wave of growth brings more customers, better economics and less dependence on a few giant technology companies.

The demand side already looks strong. Lambda has shown that it can win projects large enough to fill enormous GPU deployments, and the overall AI infrastructure market is still expanding rapidly.

The next proof point has to come from the quality of that growth. A broader mix of large customers would reduce the risk around Microsoft, Nvidia and Amazon. Positive operating economics after depreciation and financing would tell us that Lambda is creating equity value as it expands instead of simply converting billions of dollars of new capital into billions of dollars of revenue.

Software matters more from here. CoreWeave, Nebius and Together AI are all trying to capture higher-value workloads above basic GPU rental, particularly inference and managed AI services. Lambda will have a stronger long-term story if customers eventually choose its platform for more than access to scarce Nvidia hardware.

If those pieces come together while revenue keeps growing rapidly, $12 billion could end up looking conservative.

Q16What could knock Lambda's valuation back down?

Lambda's $12 billion case gets fragile very quickly if growth slows before the company improves customer diversification and returns on capital.

GPU scarcity has helped neoclouds grow because customers could not get enough compute from traditional providers. That shortage will not necessarily last forever. Nvidia keeps shipping more hardware, hyperscalers continue adding enormous amounts of capacity, and customers now have a growing list of specialized clouds to choose from.

Competition can therefore hit Lambda twice. More available capacity can push rental prices down, while AWS, Azure and Google can use their scale to compete aggressively for the same enterprise workloads.

Hardware ages quickly too. Every new Nvidia generation raises the question of how much older clusters can still earn and how quickly providers have to finance another round of equipment.

Investors have lately become more conscious of that problem. Recent research highlighted by MarketWatch argued that major neocloud companies are producing much weaker returns on invested capital than the hyperscalers funding their own AI infrastructure.

The danger for Lambda is straightforward: revenue can keep rising while the economics underneath each dollar get worse. At some point, investors will care much more about cash returns than headline growth.

Q17So is Lambda really worth $3B today?

Yes, Lambda looks worth comfortably more than $3 billion today; our harder conclusion is that $12 billion still feels slightly ahead of the evidence.

We would put the better-supported range around $8 billion to $10 billion for now. That judgment lines up reasonably well with the secondary-market evidence, where Lambda shares have recently traded below the level that had implied roughly $9 billion earlier in the summer.

CoreWeave also keeps us from getting too enthusiastic. Its current market valuation works out to around 3.8x guided revenue despite 112% quarterly growth and enormous contracted backlog. Paying roughly twice that multiple for Lambda requires us to believe the private company has considerably more upside left in its growth curve.

There is a credible case for believing exactly that. As seen above, Bloomberg expects Lambda to generate more than $1.5 billion of revenue this year, while major customers have already committed to very large future deployments. If Lambda keeps compounding at anything close to its recent pace, today's proposed price can become reasonable surprisingly quickly.

We just would not say Lambda has fully earned $12 billion yet. The company still needs to show that its explosive growth can survive billion-dollar scale, that revenue can spread beyond a few huge customers, and that all this GPU infrastructure produces strong returns after debt, depreciation and replacement costs.

So the answer to the original question is unusually clear: $3 billion looks too low. $12 billion looks aggressive. Somewhere around $8 billion to $10 billion fits the evidence better today.