Every week NAND Research puts out a newsletter for our industry customers. Below is a excerpt from this week’s:
Driving the Week
For as long as we can remember, the first tech event of the year (and one of the largest, with 138,000 expected attendees) has been the Consumer Electronics Show. Our own Michael McDowell details what he’s looking forward to in his latest blog, which includes lots of AI and 5G.
We’re already on the ground in Las Vegas, so stay alert for on-going coverage.
By the Numbers
Expect a big year for cloud. According to Gartner, IaaS spending will be about $170 billion in 2025, growing year-over-year at 21%. This is still a nice market, but slightly lower than Garter’s mid-year projections of $180 billion.
Factor in SaaS, PaaS, and DaaS (Desktop-as-a-Service), and Gartner predicts that the overall cloud market will land at about $723.4 billion, up 21.4% over 2024.
Spending big to meet that demand, number two cloud provider Microsoft says that it’s investing a whopping $80 billion on AI enabled data centers in its fiscal 2025. This comes directly from a blog by the company’s president, Brad Smith, who went on to say that “more than half of this total investment will be in the United States, reflecting [Microsoft’s] commitment to this country and our confidence in the American economy.”
Microsoft’s not the only company spending big on AI, with TechCrunch rolling up VC investment in the sector to report that, in 2024, generative AI-focused startups raised $56 billion across 885 deals, up 192% from 2023.
This includes some big funding deals, like Databricks’ $10 billion Series J, xAI’s $6 billion Series C, and multi-billion-dollar rounds for both Anthropic and OpenAI (in an earlier piece, TechCrunch details the 49 US AI-based startups that have raised $100 million or more in 2024).
Semiconductors
It’s been an active couple of years in the semiconductor space, continuing into 2025. There are reports that both NVIDIA and Qualcomm may be following Apple’s lead and moving their respective upcoming 2nm projects away from TSMC to Samsung due to “high costs and limited production capacity.”
Is NVIDIA getting into the custom ASIC business? The market for custom silicon is hot, with Broadcom and Marvell each showing record growth from CSPs looking to build their own processors and accelerators. NVIDIA, the current king of AI technology, may be looking to get into that business, with the Commercial Times reporting that the company is actively recruiting talent in Taiwan to staff the new division.
If this one’s true, there will be significant implications for Broadcom, Marvell, and Mediatek. We’ll keep watching.
It’s all about that HBM: word is that SK hynix, expected to show off the first samples of its HBM4 memory, has landed Broadcom as key customer for its burgeoning HBM business. This would be a big win for the memory supplier, who already supplies HBM to NVIDIA.
Broadcom competitor Marvell already has a deal in place with SK hynix, Micron, and Samsung for its recently unveiled custom HBM architecture.
Speaking of Broadcom, CEO Hock Tan has officially denied rumors that its buying Intel.
Deals & Alliances
Just before the holiday break, Cisco announced its intent to acquire SnapAttack, a privately held company focused on cybersecurity threat detection. Cisco says that the acquisition will be help enterprises migrate to Splunk Enterprise Security from competing solutions. It’s a solid acquisition.
NVIDIA has closed its acquisition of Run:ai. In a surprising move, NVIDIA announced that it will open-source the Run:ai software, which provides orchestration capabilities for GPU clouds, and which NVIDIA paid a reported $700 million to acquire. NVIDIA is likely making the move to head-off anti-trust efforts, but, regardless the reason, it benefits the entire segment.
Under the wire: Tech services giant World Wide Technology announced on December 31 that its acquiring Canadian IT provider Softchoice for about $1.25 billion. WWT CEO Jim Kavanaugh said that the deal is motivated by Softchoice’s capabilities in cloud, cybersecurity, and AI. Softchoice, if you’re not familiar, is one of the largest tech solutions and service providers in the North America.
What We’re Reading
Will AI replace musicians? It’s a concern detailed by the Financial Times in its piece “Abba’s Björn Ulvaeus warns of AI threat to musicians’ revenues,” which cites a recent study published by the International Confederation of Societies of Authors and Composers (CISAC, of which Abba’s Björn Ulvaeus is president) to note that musicians face a “significant fall in income as AI-generated music is trained on, and then replaces, the work that they do,” going on to say:
“Music and audiovisual creators risk losing 24 per cent and 21 per cent of their revenues to the technology, respectively, by 2028… This would amount to a cumulative loss of €22bn over a five-year period.”
AI or not, technology transitions are hard. Just ask any 1970’s rock band struggling to understand digital music, programable drum machines, and something called MTV that were converging at the end of the decade. The New York Times takes a look in its long-read “’You Can Never Look Back’: How ’70s Rockers Rebooted for the ’80s.”
Looking at prog-rock band Yes’s hit “Owner of a Lonely Heart,” the piece tells of its producer insisting that the band program beats on a drum machine instead of using the band’s drummer:
“One day, [the producer] said, the band told him they would not move forward with ‘Owner of a Lonely Heart’; the drum programming was just too much. ‘I threw myself on the floor and I pulled at their trouser legs,’ he recalled. ‘You’ve got to do it. Let me program it. Please, please, please! You’ll ruin my career!’”
The band relented and the song hit number one. ZZ Top and Don Henley have similar stories, but each also eased up on the technology and found a balance between 80’s tech-driven pop sounds and more traditional instrument-driven approaches.
It’s hard not to read this and think about what’s happening with various iterations of Co-Pilot and other AI-assisted tools. The industry’s swinging hard towards AI but will ultimately find the right balance.
