TRO | Anthropic Revenue Hits US$30 Billion

Plus, Uber Backs Amazon's Chips and NASA Deploys AI to Protect Artemis II Crew

Author

Meliora Company
April 13, 2026 • Estimated Reading Time: 4 minutes

Subscribe | 13th April, 2026

In this fast-moving Vertical AI economy, headlines are everywhere but optimism is rare.

Here’s our take on 3 stories that will help you be relentlessly optimistic about the future.

1. Anthropic Revenue Hits US$30 Billion

Anthropic has announced a major compute expansion, securing close to 5 gigawatts of next-generation TPU capacity through a partnership with Google and Broadcom, with the bulk of capacity coming online from 2027. Around 3.5GW will be delivered through Google's hardware infrastructure built on Broadcom-designed chips, with the remainder coming through separate arrangements. The vast majority of the capacity will be sited in the United States, extending Anthropic's November 2025 commitment to invest $50 billion in American computing infrastructure. The announcement comes as Anthropic revealed its annualised revenue run rate has surpassed US$30 billion, with demand from Claude customers described as having accelerated sharply through 2026. As part of the agreement, Anthropic will also develop and supply custom TPUs for Google through to 2031.

Here’s why this matters: Anthropic's revenue tripling in under a year is the clearest signal yet that enterprise AI spending has moved from pilot to infrastructure commitment at scale. The Google-Broadcom deal is significant beyond its size: by securing its own custom TPU supply chain and locking in capacity through 2031, Anthropic is making a deliberate move to reduce its dependence on Nvidia at precisely the moment every major AI company is competing for the same constrained pool of compute. For telco, media and technology executives, the strategic question this raises is a structural one: the companies defining the AI frontier are now vertically integrating. That changes the competitive dynamics not just for chip vendors, but for any enterprise that assumed the compute layer would remain a commodity they could simply buy their way into.

2. Uber Backs Amazon’s AI Chips

Uber has expanded its AWS contract to run more of its ride-sharing infrastructure on Amazon’s in-house chips over Oracle and Google. The move is a notable shift for a company that in 2023 signed high-profile multi-year cloud deals with Oracle and Google as part of a wholesale migration off its own data centres. Uber now joins Anthropic, Open AI and Apple as major organisations that have expanded their AWS commitments off the back of Amazon’s homegrown silicon. Amazon CEO Andy Jassy confirmed Trainium was already a multibillion-dollar business by the end of last year.

Here’s why this matters: Cloud infrastructure loyalty now has a price. Uber signed seven-year deals with Oracle and Google just three years ago, and is already expanding onto a third provider. This is a significant signal for any large organisation currently mid-way through a long-term cloud commitment. The hyperscaler with the most compelling homegrown chip story now has a meaningful lever to pull at renewal time. Nvidia remains dominant, but the fast that companies of Uber’s scale are actively trialling alternatives tells us that the window is open. The chip layer of the AI stack is fast becoming the sharpest competitive battleground in enterprise technology.

3. NASA Deploys AI To Shield Artemis II Crew

Ahead of the Artemis II crewed lunar mission, NASA tested two new solar radiation forecasting tools developed by researchers at the University of Michigan. The first was a machine learning model trained on decades of satellite imagery from the Sun that estimates the probability of a dangerous solar particle event up to 24 hours in advance. The second was a physics-based model designed to predict the timing and duration of radiation storms at Earth and the Moon, running on 3,000 reserved processing units on NASA’s supercomputer to ensure it can respond fast enough. Artemis II is flying during the peak of the Sun’s 11-year activity cycle, making the forecasting capability critical for a mission operating largely beyond Earth’s protective magnetic field.

Here’s why this matters: Space is the most unforgiving environment AI could be asked to operate in, and this is a meaningful test of whether machine learning can move from pattern recognition into genuine life-safety infrastructure. The two-model approach is smart engineering - probabilistic forecasting tells you whether to worry, and physics modelling tells you when and for how long. The same combination of satellite data ingestion, real-time AI inference, and edge compute under strict latency constraints that NASA is deploying here is structurally identical to what telecoms and media companies are being asked to build for autonomous systems, live broadcast, and network resilience. The hardest problems in AI aren’t the ones happening inside a data centre, they’re the ones where the model has to be right, fast, and without a human in the loop.

Still Curious?

Our Pursuit Of Better

Giggles, a meme-based prediction market built by 19-year-old Minecraft YouTuber Justin Jin, has raised >US$1.23m in seed funding. The short-form video app lets users invest “aura points” they think will go viral. With 450,000 sign ups already, Jin’s ambition is to build the first crypto app where users spend more than 30 minutes a day.

Defense start up Hermeus has raised US$350m to keep developing what it calls the “fastest unmanned aircraft”, in a round that has pushed its valuation to US$1bn. The LA-based company flew a demonstrator the size of an F-16 last month and is targeting supersonic flight on its next iteration. Its rapid prototyping approach takes inspiration from SpaceX, in a sector where VC investment crossed US$9bn globally last year.

Xoople, a Spanish Earth observation start up, has raised US$130m in a Series B led by Nazca Capital, pushing its total funding to US$225m and its valuation into unicorn territory. The company is building a satellite constellation to collect high-precision optical data for AI applications, and has spent seven years pre-embedding its platform into Microsoft and Esri.

The Story of the First Multi-Screen Australian Open

It’s a race against time to deliver on what has been promised - every point, of every match, on every court, available to watch live. This is a story that ends with nearly half a million app installs, millions of video streams, and a team behind it all that became more like family.

Meliora Team Spotlight

Ricky Sutton - Australia

Ricky’s career spans three decades at the intersection of journalism and technology. He has reported from conflict zones, led global newsrooms, and advised global companies from News Corp and CNN to Microsoft.

Murray Barnett - UK

Murray provides deep expertise in sports media and commercial rights across global markets. He specialises in transforming how sports properties package and monetise their media, sponsorship and commercial rights.

The Pursuit of Better, Together

If this week’s newsletter sparked something, pass it on to a colleague. The best conversations often start with “You’ve got to read this”.