It wasn’t exactly on the scale of the introduction of the iPhone or even the iPod, but Apple this week managed to make a credible splash in artificial intelligence.

Much of Apple Intelligence is on the come, and you’ll need a newer iPhone to get the features, but as usual Apple managed to make an emerging technology approachable and show how it will be useful. We’ll see if it delivers, but investors already gave it credit, boosting its stock more than 7% in recent days.

Deeper into the AI weeds, Databricks held its Data + AI Summit, following Snowflake’s Data Cloud Summit last week in the same Moscone Center in San Francisco. The upshot: The privately held company, on the list of top candidates to go public this year, sought to negate the knotty data format wars and expand its appeal beyond data scientists, but as CEO Ali Ghodsi (pictured) admitted, such battles will continue “until the sun burns up.”

Meanwhile, the money vacuum in AI keeps sucking in billions, as Mistral and AlphaSense each raised about $650 million this week. But revenue is coming in too: $2.4 billion in the first half for Databricks, $3.4 billion since last year for OpenAI, and AI drove upside earnings this week at Oracle, Broadcom, Rubrik and Adobe.

Pat Gelsinger’s comeback plan for Intel hit a rough patch as it delayed a $25 billion fab in Israel. Meantime, Samsung, the No. 2 foundry Intel aims to catch, just outlined its new two-nanometer chipmaking process.

Is the deep freeze for initial public offerings thawing? The little computer that could, Raspberry Pi, had a successful IPO in London as shares soared at the open. And there was another IPO this week, as health diagnostics firm Tempus AI raised $411 million and its stock popped as much as 15% Friday.

Microsoft admitted its security problems and promised to do better, and it started by putting off the introduction of its much-criticized Recall online activity tracking feature. Elsewhere on the cybersecurity front this week, at its re:Inforce conference Amazon Web Services outlined a bunch of new AI-driven security features. Meantime, consolidation chugs on as Fortinet bought Lacework apparently for a song.

Next week is HPE Discover, and what do you think CEO Antonio Neri will be talking about? 🙂

This and other news are discussed in depth on John Furrier’s and Vellante’s weekly podcast theCUBE Pod, out this afternoon on YouTube. Also, don’t miss Vellante’s weekly deep dive, Breaking Analysis, this week looking at the prospects for AI servers versus AI in the cloud, ahead of HPE Discover.

Here’s the biggest news and analysis from this week’s stories on SiliconANGLE:

AI and Data Week 2: It’s Databricks’ turn

Apple unveils generative AI ‘Apple Intelligence’ coming to iPhone and Mac There’s a lot on the come, rather than available now, and you’ll need a new iPhone for most of the new features. Plus, despite commendable work on the privacy front, Apple will face many of the same challenges of copyright and hallucinations, given that its training data (“open web,” mainly) looks similar to what other AI model providers are using. Investors nonetheless were thrilled, or at least relieved, that Apple’s finally on the gen AI train, as its market cap again topped Microsoft’s (barely). 

Commentary from theCUBE Research analyst David Linthicum: On the AI Insights and Innovation podcast: Where are the use cases to justify generative AI investment?

Elon Musk drops lawsuit that claimed OpenAI abandoned its founding mission

A sobering study from LucidWorks: Delays, implementation Issues and unrealized benefits challenge generative AI initiatives in 2024. But as earnings (below) showed this week, the AI picks-and-shovels suppliers are still getting the benefits of gen AI spending, so this could be the usual technology chasm that needs to be crossed.  And this study also showed that almost two-thirds of enterprises still plan to increase AI spending, while almost none intend to lower it.

Surprise: Google tops Forrester’s first ranking of AI foundation models

News and analysis from Databricks’ Data + AI Summit:

• Databricks broadly boosts support for AI model training
• Databricks debuts new data pipeline development and business intelligence tools
• Integrating gen AI and interoperability in data management: theCUBE’s day 1 analysis from Data + AI Summit
• Databricks’ strategic shifts in data engineering and interoperability: theCUBE keynote analysis from Data + AI Summit
• Databricks leads the way in open-source and no-code technologies: key analysis
• Informatica rolls out new integrations for Databricks’ cloud data platform
• Databricks told analysts it expects annualized sales growth to top 60%, to $2.4 billion in the first half of this year, per CNBC
• How data transformation is shaping the future of AI: analyst insights from Howie Xu

There’s a lot of good analysis in there, with more on theCUBE Pod, but a few other things struck me from the event:

• Databricks is looking to shift the war with archrival Snowflake, whose annual Data Cloud Summit was also at Moscone Center in San Francisco, beyond data formats. CEO Ghodsi promised it would find a way for its Delta Lake format to be compatible with Apache Iceberg, the open-source format championed by Snowflake users. Indeed, that was why Databricks last week bought Tabular Technologies, whose founders created Iceberg, just before Snowflake CEO Sridhar Ramaswamy took the stage at Moscone for his keynote. “The differences are nonsensical,” Ghodsi said during a press conference, after promising at his keynote earlier Wednesday, “We are all in on open.”
• Databricks also sought to negate any disadvantage prospective users might see in its Unity Catalog for organizing data assets, making it open source, as Snowflake did last week with its Polaris Catalog. “Their aspiration is to govern your entire data estate,” said analyst George Gilbert. “By open-sourcing it, Databricks is saying, ‘Forget the data catalog now. We’ve got that covered. You won’t have to care. The new platform is the catalog.’”
• How important those format wars are remains to be seen. I talked with Jon Pruitt, IT director for innovation R&D for the Hartsfield-Jackson Atlanta International Airport, which has settled on Databricks to help it bring its disparate data together to drive new revenue streams through analytics. He hadn’t even heard of Iceberg until the day before.
• With its new AI/BI product, Databricks is looking to move beyond its roots as the place for data scientists to more business analysts and even mere mortals. “We could end up with millions of users if we do it right,” he said. “Our mission since day one is to democratize data and AI.”
• Ghodsi is betting strongly that AI models will be developed for specific industries — a belief that was apparent on the exhibit hall floor, where Databricks had an “industry solutions hub” featuring seven industries, and during the keynote, when Shutterstock was trotted out as an example of using a specialized model, ImageAI. “We’re looking to have models for each industry,” he said. “The future will be more and more specialization.” Indeed, he added, “when we made [Databricks’ own large language model] DBRX and tried to make it good for some things, it got worse on other things. Like it was really good at English, and when we tried to get it better at coding, it got worse at English. You can’t have your cake and eat it too.”
• At the same time, he sees more customers using “compound systems,” which is to say bringing together multiple AI models for various use cases. “People are combining many different models,” he said. “More and more you need a complicated system to get what you want.”
• A huge concern for generative AI is how much power it takes to train and run models — it frankly seems unsustainable. But Nvidia CEO Jensen Huang, whose appearance is mandatory at a conference if a company wants to be taken seriously in AI, said during the keynote that he thinks there are ways around that — by locating training data centers in areas where energy isn’t as expensive, for example, and by implementing more efficient models that can run on devices, so they don’t have to constantly send and receive data over the internet. “The computer model’s going to transform completely,” he said. “It’s going to save a ton of energy.” We’ll see.
• The bottom line for Databricks is to make it easy for customers to store their data from Salesforce, Google Analytics, databases and anywhere else in Databricks. “It’s the top thing everybody wants,” Ghodsi said. “Data engineering is maybe the biggest financial opportunity for Databricks.”

From last week’s Qlik Connect, Shelly Kramer’s wrap-up: Qlik Connect recap: Where there’s data, there’s opportunity

Money matters

AlphaSense raises $650M at $4B valuation, agrees to acquire rival Tegus in $930M deal

French open-source AI model startup Mistral AI raises $640M at $6B valuation

Healthcare technology provider Tempus AI raises $410M in IPO

Generative AI and cleantech show growth despite overall VC downturn

OpenAI’s annualized revenue doubled to $3.4B since 2023, according to The Information

Amazon to splurge $230M on free cloud credits for generative AI startups

Contact center automation startup Cognigy closes $100M funding round

Dublin-based AccountsIQ bags $65M to bring more AI-powered products to accountants

Enveda raises $55M to accelerate drug discovery with AI

Restate raises $7M to make building fault-tolerant applications easier

AI-powered voice clarity startup Tomato.ai launches with $2.1M in seed funding

BlinqIO raises $5M to automate software testing using generative AI

New services

Startup claims it can automate 80% of software development with generative AI

Zeta Labs unveils JACE, an action-oriented generative AI assistant quite unlike any other

Automation Anywhere levels up business process automation with generative AI

Stability AI releases SD3 Medium, its ‘most advanced’ text-to-image generating AI model yet

Luma AI’s Dream Machine expands access to generative AI video creation

Researchers develop new LiveBench benchmark for measuring AI models’ response accuracy

Check out the rest of our AI and data stories.

AI drives enterprise earnings

Earnings

• Big cloud AI deals with Google and OpenAI send Oracle’s stock higher, despite earnings miss
• Broadcom’s stock rises on strong AI chip sales and improved annual sales forecast
• Rubrik delivers earnings beat in first post-IPO quarterly results
• Adobe’s stock jumps on AI-powered earnings and revenue beats
• Autodesk posts higher Q1 revenue after accounting-probe delay

More money matters

Construction of Intel’s new $25B chip fab in Israel grinds to a halt

Samsung details upcoming two-nanometer chip manufacturing process

Amazon Web Services will launch a new region in Taiwan by early 2025, with a plan to invest billions of dollars over a long period.

Black Semiconductor secures €254.4M to develop graphene-based chip interconnects

Miniature computer maker Raspberry Pi raises £166M in London IPO

Startup Flow raises $4M to iterate on parallel processing tech that massively accelerates any CPU

Nexus Labs closes $25 million Series A round to scale verifiable computation

And in other news

Fastly releases global cloud AI accelerator to help developers reduce costs and boost performance

Synopsys debuts ‘first complete’ chip blueprint bundle for implementing PCIe 7.0

Zeus Kerravala’s wrap on last week’s big Cisco event: Five thoughts from Cisco Live 2024

Check out more cloud and infrastructure news.

Cyber beat: Consolidation continues

Money matters

Fortinet acquires cloud security startup Lacework

Incode acquires MetaMap to enhance identity verification services

Cyberhaven reels in $88M for its data security platform

Industrial cybersecurity firm XONA raises $18M for zero-trust user access enhancement

Attacks and apologies

Microsoft’s Brad Smith acknowledges past security failures, outlines new initiatives

Microsoft delays release of Windows 11’s Recall feature

Mandiant finds 165 Snowflake customers were targeted in hacking campaign And now, Bloomberg reported, Snowflake plans to close an internal probe into recent cyberattacks after not detecting any unauthorized access into customer accounts since early last week

Tile’s parent company Life360 discloses data breach and extortion threat

Black Basta suspected of using patched Windows flaw in recent cyberattacks

Black Basta ransomware group also suspected in Ascension data theft incident

Forescout report finds network infrastructure attacks have overtaken endpoint security risks

New services

AWS reaffirms security commitment with new AI-powered features and other measures

AuthenticID debuts new algorithms to identify when deepfakes are being used to trick ID verification tools

Check out the rest of the week’s cybersecurity news.

Elsewhere in tech

At WWDC, Apple announces macOS Sequoia, iOS 18 and VisionOS 2

GM gives troubled Cruise self-driving car unit $850M amid strategic review

​​Waymo issues another recall after driverless taxi crash

Terraform Labs and co-founder Do Kwon agree to $4.47B settlement with SEC

Near Foundation launches Nuffle Labs with $13M to advance modular blockchain infrastructure

Comings and goings

Former NSA Director Paul Nakasone joins OpenAI board Also, former NextDoor CEO Sarah Friar is OpenAI’s chief financial officer, a role previously vacant for two years, per Bloomberg. Kevin Weil, formerly at Instagram and Twitter, is OpenAI’s new chief product officer.

Longtime Googler Dave Burke is shifting from his job as Android’s VP of engineering to an advisory role while “exploring AI/bio projects,” part of a reorganization of Google’s Platforms & Devices team.

Former IBM and Xerox PARC engineer Lynn Conway, whose seminal work in the VLSI approach to making chips made possible the systems-on-a-chip that dominate today, died at age 86. She was also a pioneer for her gender transition.

What’s next: events

June 17-20: HPE Discover TheCUBE will be onsite with interviews and analysis, and we’ll have all the news on SiliconANGLE.

June 17-20: Collision Conference, Toronto

June 18-20: Augmented World Expo, Long Beach, California

Photo: Robert Hof/SiliconANGLE