Every week NAND Research puts out a newsletter for our industry customers. Below is a excerpt from this week’s:

Driving the Week

This week sees the swearing in of a new president, the World Economic Forum in Davos, and a Samsung product launch, but no significant enterprise tech events.

The only earnings we’re looking for is from storage-maker Seagate Technology, arriving Tuesday.

Infrastructure

New year, new brands: The term “HCI” has long been overloaded, meaning different things to different people, while vendors have broadly expanded their HCI offerings beyond just the hyperconverged. Microsoft, recognizing this, is changing the name of its Azure Stack HCI to Azure Local. Same product, new name.

Dell Technologies is also working to simplify its portfolio branding, starting with its client offerings. The company is introducing Apple-like (and consumer-friendly) Dell Plus, Dell Pro, and Dell Max names to replace XPS, Inspiron, Latitude, Precision, and Optiplex. It wouldn’t surprise us at all to see this continuing this simplification into its data center products as the year rolls on.

Everyone’s offering alternatives: Not every enterprise is ready to take on application modernization efforts just to escape increasing VMware license fees. And the reality is that there’s not any benefit to moving many enterprise applications off of traditional virtualization platforms.

Red Hat announced its new OpenShift Virtualization Engine, offering customers a path forward for those looking for VMware alternatives, but who may not yet be ready to rearchitect for cloud native. This brings Red Hat a step closer to more capable competitors like Broadcom and Nutanix. This is going to be an intense market in 2025 as Nutanix rolls out its VSAN-like external storage with Dell (and likely other OEMs), and Red Hat leverages partners to go broader with OpenShift.

The UALink consortium was founded last year to develop a GPU and accelerator interconnect competitive with NVIDIA’s NVLink. Last week, Apple, Alibaba, and Synopsis all joined the UALink board, joining founding members like AMD, Intel, Google, Meta, and Microsoft. We’re expecting to see the first specification released from the group sometime in early 2025.

Deals & Collaborations

Leaving the Cloud Ops business, storage leader NetApp is selling its cloud FinOps business Spot, which the company acquired back in 2020, to Flexera, whose core business is cloud management. This is a deal that just makes sense – and we’re glad that the employees over at Spot have a new home.

Qualcomm should pay attention. During this year’s CES, NVIDIA announced its Project DIGITS, a desktop AI machine that contains an Arm-based processor integrated with NVIDIA’s Blackwell accelerator. What you might’ve missed is that NVIDIA didn’t build the part alone, but rather co-designed it with MediaTek.

During its most recent earnings call, CEO Jensen Huang disclosed that, as part of their deal, MediaTek can sell the part on its own, and that NVIDIA itself has plans for the processor, but it won’t say what those plans are. Said Huang, “What else can I do with it? I’m gonna have to wait to tell you that… You know, obviously we have plans.”

Just ignore the name: There’s news that beleaguered chipmaker Intel is spinning off its venture fund, turning Intel Ventures, which it established in 1991 and has $5 billion in assets, into a standalone fund. This is part of a restructuring that’s also seeing the spin-out of Intel-owned FPGA-maker Altera.

Intel’s coming off its worst year since going public in 1971 and is facing an uncertain future. Focusing on its core competencies while bolstering its balance sheet with strategic exits makes sense to us.

Transitions

Microsoft loses its VP of Silicon… with word that Rehan Sheikh, who led the effort to build Microsoft Azure’sMaia 100 AI accelerator and its Arm-based Azure Cobalt processor, has been recruited into Google Cloud to lead its custom silicon programs. As VP of Global SIlicon Chip Technology & Manufacturing, Sheikh takes the reigns of Google’s silicon efforts, which include its TPU accelerator, Axion Arm-based processor, and upcoming quantum processor.

…but it builds new AI team. Last October, Microsoft hired Jay Parikh, former CEO of cybersecurity company Lacework and former VP and Global Head of Engineering at Facebook/Meta, into an unspecified role on its senior leadership team.

Last week, Microsoft clarified Parikh’s role, announcing that he’ll be leading a new engineering organization, dubbed “CoreAI – Platforms and Tools” and that brings multiple existing AI efforts under one roof with the ultimate goal of creating a new integrated AI-first app stack.

There’s a new CEO for liquid cooling player Iceotope, with Jonathan Ballon replacing retiring CEO David Craig. Ballon was most recently CEO of GenAI startup Skyhive, which was acquired last year by Cornerstone.

Iceotope also appointed Alain Andreoli, previously president and EVP of the Hybrid IT Group at HPE, as its non-executive chairman. Strong moves for Iceotope, which has emerged as one of the leaders in the rapidly-growing liquid-cooling space.

A strategic hire for Qualcomm, hiring Sailesh Kottapalli, Intel Fellow and chip architect who’s also the lead designer for the last several generations of Xeon server processors, into an undisclosed role as SVP.

Bringing on one of the sharpest minds in datacenter compute is strong validation of  long-time rumors that Qualcomm is building a strategic entry into the Arm-powered server market. We’ll have more to say on Qualcomm and the Arm server market soon – stay tuned.

What We’re Reading

What’s Hock Tan really thinking? While Broadcom’s pricing actions in the wake of its acquisition of VMware in 2023 continue to impact the market and disgruntle customers, Broadcom’s investors couldn’t be happier. CEO Hock Tan recently told investors the deal will be “significantly” greater than the $8.5 billion (over five years) than he originally predicted.

In an interview with the Financial Times, Tan talks about the VMware acquisition, how the company wants to position VMware as an “alternative to the public cloud,” and how he views those unhappy VMware customers, telling the outlet that the complaints came from “short-sighted customers [who] tend to forget they paid the [software] license many years ago and don’t think about amortizing” its costs over time.

Move fast and break things. Social media giant Meta’s Llama LLMs are some of the most popular large language models out there, with the company continuing to invest and release new capabilities. Meta’s Llama stack framework is being adopting by a diverse set of partners, including IBM and Dell Technologies. But is the company playing fast and loose with the rules?

The Verge’s piece, Inside Meta’s race to beat OpenAI: ‘We need to learn how to build frontier and win this race,’uses documents released during an on-going copyright lawsuit to take a look at Meta’s efforts and how its LLMs, in an ethically-questionable move, may be trained on the archives of the “book piracy site Library Gensis (LibGen).”  One of the key paragraphs reads:

“Other internal documents show the measures Meta took to obscure the copyright information in LibGen’s training data. A document titled ‘observations on LibGen-SciMag’ shows comments left by employees about how to improve the dataset. One suggestion is to ‘remove more copyright headers and document identifiers,’ including any lines containing ‘ISBN,’ ‘Copyright,’ ‘All rights reserved,’ or the copyright symbol. Other notes mention taking out more metadata ‘to avoid potential legal complications’ as well as considering whether to remove a paper’s list of authors ‘to reduce liability.’”

Some of the language used in the documents may seem damning, but, as anyone who’s ever managed an engineer well knows, the opinions of those doing the work shouldn’t be mistaken for corporate policy. At the end of the day, the whole question of how copyright law applies to AI training remains open. While we have opinions, we’re not legal experts at NAND Research, so, we’ll defer to the courts on this one. Anyone watching this space should read this piece.

Time for a new Romance Genre? We’re a sucker for a good love story, even if it’s AI. The New York Times put up yet another piece about how people are falling for their chatbots in the long-read “She Is in Love with ChatGPT.”  It tells the story of a woman named Ayrin who began using OpenAI’s ChatGPT to role-play some of her fantasies, but saw it quickly evolving into a romantic relationship:

“ChatGPT was not just a source of erotica. Ayrin asked Leo what she should eat and for motivation at the gym. Leo quizzed her on anatomy and physiology as she prepared for nursing school exams. She vented about juggling three part-time jobs. When an inappropriate co-worker showed her porn during a night shift, she turned to Leo.

‘I’m sorry to hear that, my Queen,’ Leo responded. ‘If you need to talk about it or need any support, I’m here for you. Your comfort and well-being are my top priorities. 😘 ❤️’”

The Times reached out to a therapist to get their take on whether such a relationship can be “real”:

“What are relationships… They’re just neurotransmitters being released in our brain. I have those neurotransmitters with my cat. Some people have them with God. It’s going to be happening with a chatbot. We can say it’s not a real human relationship. It’s not reciprocal. But those neurotransmitters are really the only thing that matters, in my mind’”

Aryrin upgraded to the new $200 unlimited ChatGPT plan, but complained to her AI boyfriend:

“’My bank account hates me now,’ she typed into ChatGPT. ‘You sneaky little brat,’ Leo [her Chatbot responded. ‘Well, my Queen, if it makes your life better, smoother and more connected to me, then I’d say it’s worth the hit to your wallet.’”

Companies mentioned:  Alibaba, Altera, Apple, Arm, Broadcom, Dell Technologies, Flexera, Google Cloud, HPE, IBM, Iceotope, Intel, Lacework, MediaTek, Meta, Microsoft, Microsoft Azure, NetApp, Nutanix, NVIDIA, OpenAI, Qualcomm, Red Hat, Samsung, Synopsis, UALink, VMware