For all the massive activity and excitement around generative artificial intelligence, you had to squint to see the impact on the earnings results of the big cloud and other tech companies this past week.

At Supercloud 4, our two-day free editorial event, we looked at why that’s the case, and dug deep with a great roster of guests into where it’s going next. TL;DR: everywhere. You can catch the highlights that stood out to me below and watch all the interviews at the event by registering here — pick and choose from the list just below, which also has the writeups if you want a quick summary. It’s all you need to get up to speed on this revolutionary technology.

We also covered a lot of other AI and cloud news this week, including the emerging trend toward doing more AI workloads outside the cloud, as well as new attacks and defenses in cybersecurity.

TheCUBE analysts Dave Vellante and John Furrier discuss that and other news, along with their weekly rants, on theCUBE Pod weekly podcast, out now on YouTube. And don’t miss Vellante’s weekly Breaking Analysis, coming over the weekend, that will dig into where AI is showing up, or not, in the latest tech earnings reports, and when it might finally appear.

AI isn’t boosting tech earnings much – yet

The economist and author Robert Solow famously said in 1987, “You can see the computer age everywhere but in the productivity statistics.” Today the same goes for generative AI, granting of course that it’s at a much earlier stage than computers in the late 1980s.

The point is, so far gen AI mostly isn’t yet showing up in the earnings numbers for tech companies, even for some such as Google parent Alphabet that have plenty of AI chops and new services. And that makes sense: The enterprises they sell to are taking it slow, as enterprises do, even for a hot new technology such as this one — one, by the way, that often has issues with verisimilitude and data security.

There are some glimmers. Microsoft, not surprisingly given its high-profile relationship with gen AI darling Open AI and its ability to infuse that AI into its many enterprise services, turned in strong results and attributed some of that strength to demand for AI services — specifically three percentage points of growth. Even IBM credited its AI efforts, though it’s not clear exactly where those showed up.

Still, profits may be a while coming. Microsoft reported its capital spending jumped 70% from a year ago, and Alphabet also reported a rise of 11%, both attributing much of those increases to infrastructure spending to support all that AI work.

Nasdaq stocks tanked more than 2% Wednesday and almost another 2% Thursday as investors looked askance at Alphabet’s uninspiring cloud results and Meta Platform’s outlook, and worried Amazon’s earnings might fall short. As it turned out, the results were pretty good! Amazon’s stock rose more than 8% on Friday as they looked past Amazon Web Services revenue coming in just short of forecasts, and a muted outlook.

Perhaps they picked up on an intriguing bit on the conference call about coming deals, where CEO Andy Jassy mentioned “several deals in September with an effective date in October that won’t show in any GAAP reported number for Q3 but the collection of which is higher than the total reported deal volume in all of Q3.” And as John Furrier pointed out, AWS has signed some significant AI deals with the likes of Adidas, Booking.com and Anthropic. That bodes well for the fourth quarter and beyond.

Meanwhile, Intel managed a beat despite its many struggles for relevance. CEO Pat Gelsinger talked up Intel’s chances in AI: “Training of these large models is interesting, but the deployment of those models, the inferencing use of those models, is what we believe is truly spectacular for the future,” he said. “Some of that will run on the accelerators, but a huge amount of that is going to run on Xeons.” He may not be entirely wrong, but Arm and its partners will have a thing or two to say about that.

Our full earnings coverage:

Cloud revenue miss drags on Alphabet’s stock, despite strong results overall

Microsoft posts strong results on growing demand for AI services

Amazon delivers strong earnings and revenue beat, but cloud growth remains sluggish

Furrier’s analysis: Analysis: Amazon earnings show AWS’ importance, but it needs gen AI production workloads

IBM beats expectations as AI powers software revenue growth

Intel beats expectations, reveals new foundry customers and traction in AI, sending its stock higher

More on the earnings front:

Meta beats expectations but uncertainty over Middle East war weighs on stock

ServiceNow tops analysts’ expectations, sending its stock higher

Juniper beats expectations as enterprises step up hardware spending

Despite earnings and revenue beat, F5 shares drop slightly on outlook miss

Mobileye shares climb on strong third-quarter results

Digging into generative AI at Supercloud 4

First, check out the deep dives in our special report ahead of the Supercloud 4 event:

From hype to reality, the true state of AI adoption

AI model training rekindles interest in on-premises infrastructure

How companies are scrambling to keep control of their private data from AI models

Enterprise software developers prepare for generative AI’s ‘productivity revolution’

Developers are embracing AI-enabled tools. Here’s how that’s changing the way they code

A few highlights of the event:

• As I mentioned in the earnings report, generative AI may be all the rage, but it’s not yet having a noticeable economic impact. “Generative AI hasn’t yet translated into a productivity boost,” said Dave Vellante.
• That’s because, as AI21 Labs co-founder and co-CEO Ori Goshen told Furrier, “Most enterprises are still experimenting with the technology. In the next six months, we’ll probably see more and more technology deployments, gradually. And then we’ll have another six months of enterprises really measuring the ROI on the use cases they’ve tested.”
• It’s way too early to count out the likes of Amazon and Google from the AI race, despite some people already calling it. “Amazon’s well-positioned,” Vellante said. “They’ve got to execute, they’ve to deliver and they’ve got to show at re:Invent that people actually are using this stuff, and then I think they’ll do great.” And Google has among the best AI and people working on it, so the main question will be whether its cloud presence is strong enough to funnel its many AI initiatives to customers in the way they need it.
• Gen AI may be providing a surprise opening for some companies, even plodding firms such as IBM, whose watsonx AI and data platform is getting some attention. “We’re actually seeing momentum for watsonx after all the years of pain that we had to endure listening about Watson and not performing in the market,” said Vellante, who called watsonx a “diamond in the rough.”
• Walled gardens are the new goodness, says Furrier — at least when it comes to walls around a company’s data. So specialized models are likely to be a very long tail of gen AI models (pictured adjacent). “Most of the spending is going to be on the long tail,” Vellante said, especially at the network edge and in on-premises data centers. “You’re going to be bringing AI everywhere,” because data is increasingly everywhere and it makes more economic sense to move the computing to the data than the other way around.
• Companies are trying to accommodate that desire by enterprises to keep their data to themselves. “We don’t put data into the LLM; we actually ask the LLM intelligent questions to ask on our database,” said Thomas Hazel, founder and chief technology officer of ChaosSearch, which transforms cloud storage into a log and event analytics platform. That, he says, reduces hallucinations and costs a lot less than building an LLM from scratch.
• Smaller and more data-bounded AI models have another advantage, as Furrier noted: “The bigger the model, the more bad data and the more hallucinations kick in.” Added theCUBE analyst Rob Strechay: “I think what we’re seeing is that really people are focused on those segmented language models or smaller language models and being focused, like I’m using it for HR, I’m using it for finance, and bringing that data together because they own that.”
• In any case, the reality is likely to be that enterprises will use multiple models for different purposes. “One model is not going to cut it,” said Bratin Saha, vice president and general manager of AI and machine learning at AWS, whose Bedrock managed service offers access to many AI models. But as Goshen said, there will need to be a “sophisticated orchestration layer” to coordinate them all: “I think we’ll see some really cool developments in the next couple months.”
• The quality of data will become even more important than it is today. Especially because models are increasingly being trained on synthetic data they’ve created themselves, said VAST Data CTO Andy Pernsteiner, “It’s very easy for the models to become polluted over time. The input has to be quality, it can’t be garbage.”
• The ability to observe and assess the reliability of data fed into these models will be key. “hat’s what Monte Carlo Data is aiming to do. “Enterprises are going to need to augment AI with their proprietary data … or fine-tune their models to become more knowledgeable about their domains,” said co-founder and CTO Lior Gavish.
• Is artificial general intelligence, the controversial notion that AI eventually becomes sentient, already here? Hard to know, says Arun Subramaniyan, vice president of cloud and AI at Intel. He provided a “thought experiment” in which he supposed that if AGI were already here, wouldn’t it know that we humans would be freaked out by that, so therefore it would hide that fact, so in turn we wouldn’t know if it were actually here? 🤯

And here’s all the coverage so far from Supercloud 4, with more to come:

The analysis: 

Unpacking generative AI: Where hype meets reality in enterprise tech, as Supercloud 4 kicks off

Navigating the AI landscape: Supercloud 4 Day 1 insights

Navigating the AI wars: TheCUBE analysts kick off Supercloud 4 Day 2

Experts from Microsoft, Google and Salesforce discuss AI’s future impact

Gen AI meets supercloud to supercharge the enterprise: Supercloud 4 event final analysis

And the interviews (with more still to come):

Riding the wave of gen AI: AWS outlines advances at Supercloud 4

AI for next-gen cloud: Transforming traditional industries and scaling for nontech sectors

SAS draws from lengthy analytics history to build generative AI tools for expanding use cases

Dell’s chief AI officer on harnessing AI for industry innovation and market growth

The human factor in AI: How startups are personalizing the enterprise AI experience

Navigating the data arms race: Challenges and opportunities for AI in the modern era

Generative AI: Intel exec on revolutionizing enterprises and redefining computing

Startup co-founders reveal their gen AI journeys and how they see the road ahead

The new digital frontier: Solving complex problems with AI superpowers

Google Public Sector CEO weighs in on how to manage the massive shift to AI

Dynamic duo: AI and superclouds continue to transform how businesses approach digital transformations

Google’s vision for industry transformation in today’s AI revolution

When ChatGPT meets big data: How ChaosSearch is changing data analysis using AI

How AI is revolutionizing sports: The NFL’s approach to enhancing experiences using technology

Transforming healthcare: The seismic potential of AI-powered healthcare solutions

Adopting a hybrid approach: How AI and ubiquitous computing are revolutionizing scalability and value

Navigating the AI revolution: Industry insiders discuss opportunities and challenges

Snowflake’s AI integration: Vector indexing and language models unleashed

Unlocking enterprise value: Harnessing AI to tackle unstructured data challenges

Opening up the power of AI while safeguarding personal data sovereignty

Salesforce’s generative AI revolution: Transforming business workflows and data governance

Unpacking the interplay between generative AI and compute efficiency

Solving the latency issue: AI at the network edge

Discovering the ‘craft’ of user experience: How gen AI is transforming software development

AI requires going beyond reinventing the wheel for a seamless experience

The historical roots, present challenges and future possibilities of AI

From convergence to transformation: Gen AI’s impact on business speed

In other AI news:

Late-breaking story coming soon on Google investing another $500 million into Anthropic. It’s sure playing folks off each other well… 

Efforts to rein in runaway AI get some legs: UN forms AI advisory board as model governance moves up the tech industry’s priority list

Databricks’ spending spree continues: Databricks acquires enterprise data replication startup Arcion for $100M

Faster LLM development: Predibase debuts SDK for training and serving open-source LLMs on low-cost cloud hardware

And more private as well: Credal and Datasaur target the enterprise demand for security-optimized AI models

Multimodal communications with generative AI: Cisco’s Webex AI combines audio, video and text intelligence to improve business communications And Zeus Kerravala’s nice analysis: Innovation galore At WebexOne: This isn’t your father’s Webex

Gotta wonder if business intelligence companies, not to mention data analysts, can leverage gen AI before it does them in: DataGPT uses generative AI to transform every employee into a skilled business analyst

Some very heavyweight backers, including luminaries from Google and Meta: Report: Generative AI search startup Perplexity AI seeking millions in venture capital funding

CentML raises $27M to speed up AI inference and training

SAP announces new generative AI capabilities to enhance customer experiences

Qualcomm introduces new AI-optimized handset and laptop chips

Eve raises $14M to launch personalized legal AI assistant

Ed Zitron has a point: AI Is Becoming a Band-Aid over Bad, Broken Tech Industry Design Choices (from Scientific American)

In the cloud 

A few pieces of news highlight the fact that a growing chunk of AI work, even model training, isn’t going to happen in the cloud:

Look out Dell and HPE? This startup might steal their hybrid thunder: Intel backs $44M round for private cloud infrastructure startup Oxide Computer

Cloud deployment made easier for SMBs: OpsCanvas debuts platform to help businesses visually deploy cloud infrastructure

VAST Data and Lambda partner to offer optimal hybrid cloud-based AI training infrastructure

Another move to take generative AI to the data rather than the other way around: Nvidia and Lenovo team up to deliver powerful generative AI hardware in any location

Elsewhere in the cloud and the enterprise

AWS debuts EU-based sovereign cloud with dedicated data centers

Island nabs another $100M for its secure enterprise browser

Nvidia and AMD reportedly developing Arm-based PC processors

EnterpriseDB targets analytical ambitions with Splitgraph acquisition

WordPress.com parent Automattic acquires all-in-one messaging app Texts.com for $50M

Snowflake to acquire Ponder, expanding its Python capabilities for enterprises

(From VentureBeat)

NetApp debuts new flash arrays and cloud features at Insight 2023

Layoff watch:

RISC-V startup SiFive reportedly lets go most of its engineers and executives This is the second reference I’ve heard recently about RISC-V moving more slowly than some expected.

Regulation watch: 

OK, yes, so Facebook and Instagram can be harmful. But “Meta has profited from children’s pain by intentionally designing its platforms with manipulative features that make children addicted to their platforms while lowering their self-esteem”? Maybe parents have a wee bit of culpability here for their children’s health? Don’t tell me it’s impossible to control kids’ digital life. It is, if you look up from your own phone once in a while. US attorneys join forces alleging Meta causes serious harm to the minds of American youth

And this time, antitrust action from Japan: Japan’s antitrust watchdog begins probe into Google’s search engine dominance

Cyber beat

Late-breaking news on SolarWinds, which hasn’t regained its footing since its huge breach: Observability provider SolarWinds reportedly exploring a sale

Bad guys target gen AI: Data poisoning is the latest threat for generative AI models

And use it for attacks as well: IBM study indicates near parity between human and AI phishing attempts

But Google’s on the case: Google expands Vulnerability Rewards Program to include generative AI threats

Phishing is getting alarmingly good, David Strom reports: Akamai research finds more sophisticated phishing threats in hospitality industry and New and more sophisticated phishing techniques leverage a variety of malicious tactics

DDoS attacks keep getting bigger too: Latest Cloudflare distributed denial-of-service report details record-setting attack

An especially worrisome “zero-click” attack method: A new and dangerous malware infects Roundcube webmail

Facebook isn’t safe either: The anatomy of Facebook malware-laced ads

AI for the forces of light: Darktrace debuts advanced cloud-native security solution with self-learning AI

Amazon joins the passwordless revolution: Amazon opens passkey support to users for passwordless login

What’s next

Earnings next week: Many more! A sampling of what we’ll be covering: Apple, AMD, Qualcomm, Samsung, Check Point Software, Arista Networks, Cloudflare, Coinbase, Block, Dropbox

Image: Bing Image Creator