The AI race is no longer just about building better models.
It is increasingly about who can secure enough computing power to run them.
That is what makes Anthropic’s latest reported deal particularly interesting. The company has reportedly signed a $35 billion cloud-computing agreement with Lambda, a Nvidia-backed cloud provider, just days after reports of another multibillion-dollar compute agreement with Nscale.
At the same time, Anthropic is reportedly preparing to make its IPO prospectus public after Labor Day.
Put those developments together and a bigger question emerges:
How much does it actually cost to compete at the frontier of AI?
Anthropic reportedly commits another $35 billion to compute
According to reporting cited in the source, Anthropic has signed a $35 billion cloud-computing agreement with Lambda.
The arrangement is tied to a Texas data center being developed by Hut 8 in Nueces County. Nvidia is also deeply involved in the setup. The chipmaker reportedly holds the lease on the facility, supplies the chips and is an investor in Lambda.
That makes the deal notable for more than just its size.
Nvidia is sitting on several sides of the transaction.
The company is involved with the infrastructure, the cloud provider and the hardware powering the facility. That has attracted attention because similar relationships are becoming increasingly common across the AI infrastructure ecosystem.
For Anthropic, however, the immediate issue is straightforward: frontier AI models require enormous amounts of compute.
And securing that compute years in advance is becoming a critical part of the competitive landscape.
Another $45 billion deal came just days earlier
The Lambda agreement would be Anthropic’s second reported compute commitment at this scale in just five days.
On August 26, Anthropic was reportedly negotiating a roughly $45 billion, six-year agreement with Nscale for computing capacity from a West Virginia facility expected to come online toward the end of 2027.
That puts the reported value of the two agreements at around $80 billion.
These are not ordinary cloud contracts.
They reflect the enormous infrastructure requirements behind today’s most advanced AI systems. Training and operating large models requires access to specialized chips, data centers, electricity and networking infrastructure, all at a scale that few companies can secure without long-term commitments.
For investors, this creates an interesting tension.
AI companies need massive infrastructure spending to grow, but that spending also creates massive financial obligations.
The real question is whether future revenue growth can justify the cost.
Nvidia’s role makes the deal even more interesting
Nvidia has become one of the biggest beneficiaries of the AI infrastructure boom because its chips are at the center of much of the industry’s computing capacity.
But the Lambda arrangement illustrates how Nvidia’s role can extend beyond simply selling GPUs.
In this case, Nvidia is reportedly:
- An investor in Lambda
- The leaseholder of the data-center facility
- A supplier of the chips inside the facility
That creates a more interconnected AI infrastructure ecosystem.
For companies like Anthropic, these relationships can help secure access to the compute they need.
For the broader market, however, they raise questions about how AI infrastructure deals are being structured and how much of the capital flowing through the ecosystem ultimately comes from the same group of companies and investors.
That is something investors will likely be watching more closely as more AI companies move toward public markets.
Anthropic’s IPO could change the conversation
The timing of these deals is particularly important.
Anthropic’s IPO prospectus is reportedly expected to become public after Labor Day, potentially giving investors their first detailed look at the company’s financial picture.
Until now, much of the discussion around Anthropic’s valuation, revenue and private-market activity has relied on company announcements, media reports and third-party information.
A public prospectus would change that.
It could provide investors with a clearer picture of:
- Revenue
- Operating costs
- Compute commitments
- Capital structure
- Potential IPO plans
- Lockup and secondary-sale arrangements
That information will be particularly useful when viewed alongside the company’s reported multibillion-dollar infrastructure commitments.
The headline valuation may grab attention, but the economics underneath the valuation will matter much more.
The real test will be revenue versus infrastructure costs
A $35 billion cloud agreement sounds enormous.
A $45 billion agreement sounds even bigger.
But the size of the commitments alone does not tell us whether Anthropic is financially healthy or financially stretched.
Investors will want to understand how these commitments fit into the company’s expected revenue growth.
The important questions will be:
How quickly is Anthropic’s revenue growing?
How much does it cost to serve its customers?
How much compute does each dollar of revenue require?
And how much additional capital will the company need to keep expanding?
Those answers could tell investors much more about Anthropic’s long-term business model than a headline valuation ever could.
OpenAI is taking a different route
Anthropic is not the only AI company trying to build a large and diversified business.
OpenAI said its advertising business has reached a $1 billion annualized revenue run rate, around 200 days after it began testing ads in ChatGPT in the US.
Ads are now available in more than 40 countries, with self-service access expanding across regions including India, Europe, the Middle East and North Africa.
The significance is not just the $1 billion figure.
It shows that OpenAI is trying to build revenue streams beyond subscriptions and enterprise AI.
The company says roughly 1 billion people use ChatGPT every week, giving it an enormous potential audience for monetization.
OpenAI also confidentially filed IPO paperwork in June and completed an employee tender offer of roughly $7 billion at an $852 billion valuation in August.
So while Anthropic is preparing for potentially greater financial disclosure, OpenAI is already giving investors selected pieces of its growth story.
The two companies are following different paths, even as they compete for customers, talent and compute.
Shein offers a very different lesson for private markets
There is another interesting data point from the source that has little to do with AI directly.
Shein priced its Hong Kong IPO at a valuation of about $26.5 billion.
That is roughly one-quarter of the fast-fashion company’s reported $100 billion private valuation from 2022.
Shein raised around $1.74 billion in the IPO, but its shares fell about 7% on the first trading day.
The lesson here is important for private-market investors.
A private valuation is not necessarily the valuation the public market will accept.
Private companies can raise capital during periods of strong investor enthusiasm and receive valuations based on expectations about future growth.
Once a company enters public markets, however, investors have access to more information and the stock is repriced continuously.
Shein’s experience is therefore worth watching as more highly valued private companies approach IPOs.
AI valuations are entering a new phase
Several other developments mentioned in the source reinforce the same theme.
Polymarket is reportedly raising $1 billion at a $21 billion valuation, up from a reported $15 billion valuation in April.
Nvidia is reportedly planning a $3.5 billion investment in MediaTek, while the Taiwanese chipmaker expects around $2 billion in AI chip revenue this year.
Chinese AI chipmaker Enflame is also moving toward a Shanghai IPO, reportedly targeting around $908 million in proceeds.
Meanwhile, SB Energy could reportedly become another public-market example of the AI infrastructure boom, with a potential IPO that could raise $5 billion to $7 billion.
The common thread is clear.
Capital is continuing to move aggressively into AI, but the market is also moving toward greater disclosure and public-market scrutiny.
What investors should watch next
The next major development could be Anthropic’s prospectus.
If the filing becomes public as expected, investors will finally have more information to put the company’s reported infrastructure commitments into context.
Three things could be particularly important.
1. Anthropic’s actual revenue growth
The company’s ability to turn AI demand into recurring revenue will be central to the investment story.
2. The cost of compute
The reported cloud commitments show how much infrastructure Anthropic expects to need. Investors will want to understand how those costs affect margins and cash flow.
3. The IPO valuation
A reported private valuation can create headlines, but the public market will ultimately determine what investors are willing to pay.
That makes Anthropic’s potential IPO an important test for the broader AI market.
The bigger picture
The AI boom is moving beyond the question of which company has the best model.
The next phase is increasingly about who can afford the infrastructure needed to scale that model.
Anthropic’s reported $35 billion Lambda deal, combined with its reported Nscale commitment, shows just how large that infrastructure race has become.
At the same time, OpenAI is demonstrating the importance of finding new ways to monetize its enormous user base, while companies such as Shein are showing how dramatically private valuations can change once a company enters public markets.
For investors, the next chapter may therefore be less about AI hype and more about AI economics.
How much revenue can these companies generate?
How much will it cost to serve that revenue?
And can their growth justify the enormous capital being committed to the technology?
Those are the questions that could matter most as the AI giants move from private-market stories toward public-market scrutiny.