260: Amazon Dispatches AWS CEO Adam Selipsky with Prime 2-day delivery

Episode 260 May 24, 2024 01:21:27
260: Amazon Dispatches AWS CEO Adam Selipsky with Prime 2-day delivery
260: Amazon Dispatches AWS CEO Adam Selipsky with Prime 2-day delivery

May 24 2024 | 01:21:27


Show Notes

Welcome to episode 260 of the Cloud Pod podcast – where the forecast is always cloudy! This week your hosts Justin, Matthew, and Jonathan and Ryan are talking about changes in leadership over at Amazon, GPT-4.o and its image generating capabilities, and the new voice of Skynet, Amazon Polly! It’s an action packed episode – and make sure to stay tuned for this week’s after show. 

Titles we almost went with this week:

A big thanks to this week’s sponsor:

Big thanks to Sonrai Security for sponsoring today’s podcast! Check out Sonrai Securities’ new Cloud Permission Firewall. Just for our listeners, enjoy a 14 day trial at https://sonrai.co/cloudpod 

General News 

00:40 Terraform Enterprise adds Podman support and workflow enhancements

01:18 Ryan – “This is for the small amount of customers running the enterprise either on -prem or in their cloud environment. It’s a pretty good option. Makes sense.”

01:42 Justin – “You know, the thing I was most interested in at this actually is that Red Hat Linux 7 is now end of life, which this is my first time in my entire 20 some odd career that I’ve never had to support Red Hat Linux in production because we use Ubuntu for some weird reason, which I actually appreciate because I always like Ubuntu best for my home projects, but I didn’t actually know Red Hat 7 was going away.”

AI Is Going Great (Or, How ML Makes All It’s Money) 

03:58 Hello GPT-4o

08:20 Justin – “there’s so many opportunities for it to be abused. And then also with voicing, if you can start sending the voice to other people, then next thing you know, your mother is calling you desperately in need of money and asking you to wire her money because she ended up in Tahiti. And you’re like, how did you end up there? Don’t answer. I came here with your father and I need money. Weird stories are going to come out of this technology for sure.”

AWS – AWS CEO signs Off

09:30 AWS CEO logs off after three years at the helm

10:44 Ryan – “I wish executive letters were truthful and honest in some way. Like, on one hand  I hope it is just a matter of, of Adam saying, you know, this job’s hard and I only want to do it for a few years and move on. But I doubt it, just because it seems that it’s all performance based and, you know, probably not getting along or not making enough meaningful movements in the right areas.”

16:28 A new generative engine and three voices are now generally available on Amazon Polly

19:46 Jonathan – “Yeah… all I’m thinking is if Kindle doesn’t have this as an option to read any book as an audiobook within six months, I will eat my hat.”

20:55 Build RAG and agent-based generative AI applications with new Amazon Titan Text Premier model, available in Amazon Bedrock

21:44 Jonathan – “I mean, thanks Amazon for playing. I guess I see why Adams is no longer there, because the other guys just kick the crap out of you. that poly thing was pretty good, but everything else, small text LLMs are not pretty exciting right now.”

23:36 Build generative AI applications with Amazon Bedrock Studio (preview)

24:24 Matthew – “They’ve always tried to build these like web console studios for stuff. I’ve never seen them fully take off like cloud nine and let they release another one later on like the studio, these web portals that kind of act as your editors of sorts. And they all seem good, but I feel like most people I know never actually fully get, get all the way into them.”

28:20 AWS Cost Anomaly Detection reduces anomaly detection latency by up to 30%

29:32 AWS CISO tells The Reg: In the AI gold rush, folks are forgetting application security

31:06 Ryan – “Yeah, it’s a whole new world out there. And companies are racing for adoption, right? Because if you’re behind, it feels really behind because of how fast everything’s moving. And so it’s a tricky thing because it’s short -cutting security about it. But there’s also sort of like, we’re having to figure out what the right way to secure development processes for AI are, right? Like, how do you train against, you know, exploiting at the prompt level. How do you segregate data access that’s within the model training? And these are all new questions that we’re sort of in the early days of figuring out. And we’re having to do that while trying to reach into the market at record pace. And something bad is going to happen and reset this.”

33:53 Amazon S3 will no longer charge for several HTTP error codes

35:21 New compute-optimized (C7i-flex) Amazon EC2 Flex instances

36:47 Justin – “Yeah, so that’s the T, they don’t have it on the C7s, but they do on the Ts still. But you know, on the T side, even they got rid of the guaranteed CPU time, but you could basically click a box and you wouldn’t get burst credits anymore. Yeah, on the T3s. So this is kind of like a next evolution of it where, you know, it’s a workload that is more sporadic and you don’t need guaranteed throughput or capacity, but you do want it when you need it for short bursts.”


39:31 What’s new with Active Assist: New Hub UI and four new recommendations 

41:06 Ryan – “I like that one because I like this optimization overall. I laugh at anything that’s optimization because it’s a great way to build a report that no one’s going to read or act on. But you still have to do it because otherwise you’ll get no traction. But some of the other optimizations and insights that they’re building into this experience, I think, will be really powerful for cloud providers and practitioners.”

42:36 Kubernetes 1.30 is now available in GKE in record time

44:29 Announcing Trillium, the sixth generation of Google Cloud TPU

45:17 Justin – “I mean, I’m sure someone cares who’s building really large models like Coheer or OpenAI or, you know, people who build models probably care about these because like every second, you know, you save, the scale they’re trying to build these models in, you know, can result in days of savings and building a new model.”

47:01 Vertex AI at I/O: Bringing new Gemini and Gemma models to Google Cloud customers

50:01 Justin – “Imagen 3 reminded me of something about ChatGPT 4 .0. So one of the things that annoys me the most about taking text and generating images, is that if you give it text, it never produces the text correctly. So you’d be like, yeah, like it’ll, you’re like, Hey, I want you to draw me a Sherpa climbing Mount Everest and below it, I want you to write the word cloud sherpa, right? Cause you’re trying to make a sticker or something. And it will be like, ‘clud serpa’ Even though you spelled it exactly right, what you wanted, it is never correct. So in the new chat tpt 4 .0, one of the demos they showed you is actually where you gave it text input to have it generate into the image and it actually uses the proper text. It doesn’t modify it or change anything.”


52:23 Bringing generative AI to Azure network security with new Microsoft Copilot integrations

55:38 Matthew – “That’s one of the big things they’re pushing is copilot helps all your tier one and tier two people really get, you know, solve all the problems that your tier three and four, you know, and that’s one of the things that they’re pushing is, Hey, this is something that will actually like help your tier ones get done more of the work. And that’s where they’re stating that copilot will actually help everything. Also, have you ever met a junior developer or really a senior developer that actually looks at logs? So like, yeah, sometimes you just have to tell people to look at logs.”

58:00 Announcing the General Availability of GPT-4 Turbo with Vision on Azure OpenAI Service

58:44 Introducing GPT-4o: OpenAI’s new flagship multimodal model now in preview on Azure

58:59 Matthew – “Well, it was in preview for a while. And then if you really look at this, turbo is really available, I think in like Sweden and East two and like maybe one other region with very limited quantities.”

59:43 Microsoft and LinkedIn release the 2024 Work Trend Index on the state of AI at work

1:01:11 Ryan- “No, I mean, there’s no doubt that it’s useful and there’s a whole bunch of really mundane things that this is going to get rid of. And I’m pretty stoked about that, right? Like I’m not too jazzed about, you know, everyone who’s an entry level position being sort of wiped out and, you know, they still don’t know what to do about like, you know, if you don’t have any entry level positions, how do you get to the next level? But, you know, at a certain point, like these things are, you know, like every automation enhancement.they replace a very fundamental part that’s easy to recreate and saves a lot of time. And I think it’s great. But yeah, shadow AI is going to be a big problem for a long time to come just because there’s a whole lot of legal ground that has to be developed to even know how to handle this.”

1:02:06 Public preview: Azure Application Gateway v2 Basic SKU


1:06:50  It’s here—Red Hat OpenShift on OCI!

1:07:31 Ryan – “What a strange announcement. Because it’s not quite a managed OpenShift service. It’s just we’ve proven we can run it on there. So it’s like, cool.”

1:08:40 Elon Musk’s xAI nears $10 bln deal to rent Oracle’s AI servers, The Information reports

1:09:17 Matthew – “Wasn’t there a news article where Elon like picked up a bunch of servers from AWS and like, just like V to C P to V them all back to on -prem and then like drove them from one data center to the other or something like that. Yeah. So it feels like why would he want to run it on the cloud if he thinks data centers make sense for everything.”


1:11:07 Bringing Project Starline out of the lab


And that is the week in the cloud! Go check out our sponsor, Sonrai and get your 14 day free trial. Also visit  our website, the home of the Cloud Pod where you can join our newsletter, slack team, send feedback or ask questions at theCloud Pod.net or tweet at us with hashtag #theCloud Pod

Other Episodes


September 14, 2019 57m47
Episode Cover

US East 1’s on Fire,  The Cloud Pod says let the !#*&@ burn – Ep 38

US-East-1 has a hiccup in a single AZ, Lambda fixes cold start launches inside a VPC, Google gets an AD service and Microsoft goes...


Episode 149

January 31, 2022 01:11:49
Episode Cover

149: The Cloud Pod BreaksFormation

On The Cloud Pod this week, the team decides 2022 is already a long, cursed year — bring on 2023. Plus nuggets of wisdom...


Episode 90

October 29, 2020 00:43:58
Episode Cover

Episode 90: The Cloud Pod gets a NanoDegree on podcasting

On The Cloud Pod this week, Peter turns into an old man in his yard, yelling at cloud providers.  A big thanks to this...