Google Cloud to accept cryptocurrencies as payment

Plus: AI services, security tools and more from Google Cloud Next this week

GCN Google Cloud today said it will start accepting select cryptocurrencies as payment for its online services.

This method will be offered first to a chosen few customers in the Web 3.0 space, according to the internet giant; it's expected that this availability will expand over time. Crypto-payment collection will be powered by Coinbase Commerce, which accepts the most popular digital coins: Bitcoin, Ethereum, Dogecoin, Shiba Inu, Litecoin, and a few others, including stablecoin USDC.

This announcement was part of Google Cloud Next (GCN), a three-day event in which the web ads titan boasts about its latest as-a-service offerings.

"Today, we’re announcing a new partnership with Coinbase, who has selected Google Cloud to build advanced exchange and data services," Google Cloud supremo Thomas Kurian briefly mentioned in among a raft of other news from GCN. "We will also enable select customers to pay for cloud services via select cryptocurrencies by using Coinbase Commerce."

We note that cryptocurrencies have generally sunk in price since a peak around November last year. Bitcoin and Ethereum were flat around time of writing on the news, and Doge was up about a couple of percent. That volatility might well explain the exclusive availability of payment-by-select-cryptocoin for now. USDC, for one, is pegged to the US dollar.

Coinbase also said it will tap up Google Cloud to use the internet goliath's networking infrastructure and analytics, and to host a bunch of its backend systems. Meanwhile, Google will dip into Coinbase's secure custody and reporting tools.

It's basically a Google-Coinbase love in, and an interesting edge for the cloud industry's third-place player. According to CNBC, Coinbase will move some of its software from Amazon Web Services to Google Cloud.

Also for GCN this week, Mandiant and Google shed some more light on why the latter bought the former for $5.4 billion and where it fits in. You can catch that and security-related updates here. Let's take a look at what else is going down at GCN.

Machine learning

Google Cloud announced OpenXLA, an open source project that may help developers build and run AI models on all sorts of hardware.

When creating machine-learning applications, you may end up tying yourself to a particular framework or optimizing the code for a particular accelerator, such as an Nvidia GPU or Google TPU, making it difficult to port to alternatives.

OpenXLA hopes to prevent that kind of lock in, by promising to be an open, universal compiler that is compatible with various frameworks and hardware backends. Sachin Gupta, veep of infrastructure at Google Cloud, said during GCN that the web giant will make OpenXLA available by open sourcing its XLA compiler and decoupling it from TensorFlow.

Speaking of hardware...

Also seen at GCN: the availability of TPUv4 AI acceleration, network offload with IPUs from Intel, and a sneak preview of the x86 giant's delayed Sapphire Rapids processors.

For a deeper dive into the TPUv4s, check out The Next Platform.

"ML development is often stymied by incompatibilities between frameworks and hardware, forcing developers to compromise on technologies when building ML solutions," he explained in a blog post. "[OpenXLA] will address this challenge by letting ML developers build their models on leading frameworks (TensorFlow, PyTorch, JAX) and execute them with high performance across hardware backends (GPU, CPU, and ML accelerators)."

The project is backed by AMD, Arm, AWS, Intel, Meta, and Nvidia. Its developers will start by building an open source community, extracting XLA from TensorFlow, and building StableHLO, which acts as a portability layer between machine-learning frameworks and compilers. 

Google also launched Vertex AI Vision. Vertex AI is a cloud-based service to help customers deploy pre-trained machine learning models or build their own more easily. The latest Vertex AI Vision feature is aimed at helping companies roll out computer vision-aided products using their own data. Developers can start adding data by dragging and dropping files to be processed by Google's AI models to complete tasks such as product recognition, object detection, or occupancy counting. 

"We are now releasing Vertex AI Vision to extend the capabilities of Vertex AI to be more accessible to data practitioners and developers," Gerrit Kazmaier VP and GM of Database, Data Analytics, and Looker, said in a blog post. "This new end-to-end application development environment helps you ingest, analyze, and store visual data: streaming video in manufacturing plants, for example, to help ensure safety, or streams from store shelves to improve inventory analysis, or following traffic lights for management of busy intersections."

Google claimed Vertex AI Vision will help developers build computer vision products in minutes at one tenth the cost of other platforms. ®