The Next Platform
@TheNextPlatform
Followers
9K
Following
1K
Media
53
Statuses
3K
For all that matters at scale. HPC, hyperscale, large enterprise and now out to the edge.
Joined October 2015
Slowdown in GenAI spending? Not a chance, says Jensen Huang. $NVDA $AMD.
nextplatform.com
The expectations for GenAI are unreasonably high and the pressure on Nvidia is tectonic. As we listened to the call with Wall Street going over the
0
0
3
IBM needs FPGAs, CPUs, and GPUs to build quantum computers. AMD doesn’t want to build QPUs. Neither wants Nvidia to own quantum. It’s that simple. $AMD $IBM.
nextplatform.com
As we talked about a decade ago in the wake of launching The Next Platform, quantum computers – at least the fault tolerant ones being built by IBM,
1
0
7
If you want to see the future of networking in AI datacenters, Google has already invented it – as usual. $GOOG.
nextplatform.com
Doubling the transistor count every two years and therefore cutting the price of a transistor in half because you can cram twice as many on a given area
1
2
11
Lenovo shows just how hard it is for anyone other than Nvidia to reap profits from AI hardware. $NVDA $AMD $LNVGY.
nextplatform.com
There are four businesses that are very difficult for an equipment manufacturer in the IT sector to make any money on: smartphones, PCs, datacenter
0
0
1
Broadcom rounds out its AI networking for 2026 with Jericho 4, and a cool new HyperPort to make 3.2 Tb/sec fat Ethernet pipes. $AVGO.
nextplatform.com
Speaking metaphorically, the blast area of a datacenter should only be so large, for mechanical as well as technical reasons, in order to mitigate the
0
0
3
Enfabrica says that its CXL memory godbox can cut the cost of AI inference in half on Nvidia GPU systems. $NVDA #Enfabrica.
nextplatform.com
Generative AI is arguably the most complex application that humankind has ever created, and the math behind it is incredibly complex even if the results
4
2
7
IBM has generated $3.5 billion in annualized cost savings from using AI, but it probably has not sold $3.5 billion in AI stuff as yet. But that will change. $IBM.
nextplatform.com
While the hyperscalers and clouds and their AI model builder customers are setting the pace in compute, networking, and storage during the GenAI
1
1
1
Broadcom was aiming at InfiniBand with Tomahawk Ultra, but it may take out NVSwitch and UALink, too. $AVGO $NVDA $AMD.
nextplatform.com
InfiniBand was always supposed to be a mainstream fabric to be used across PCs, servers, storage, and networks, but the effort collapsed and the remains
0
1
11
If you need a behemoth of a server to run big jobs with lots of threads and memory, you really only have one choice: A sixteen-socket Power E1180 based on IBM’s new Power11 chip. $IBM $INTC $AMD.
nextplatform.com
If you need a big, badass box that can support tens of terabytes of memory, dozens of PCI-Express peripheral slots, thousands of directly attached storage
0
0
6
Reversing course, the Trump administration is going to let Nvidia sell H20 GPUs in China. Why? $NVDA $AMD.
nextplatform.com
Perhaps the most interesting conversation that has happened so far in the White House in 2025, at least from the point of view of the IT sector, is when
0
2
3
Moore’s Law is only really alive in the guise of shrinking precision, not so much in shrinking transistors. FP2 anyone? $AMZN $NVDA.
nextplatform.com
This week, Amazon Web Services announced the availability of its first UltraServer pre-configured supercomputers based on Nvidia’s “Grace” CG100 CPUs and
0
1
4
CoreWeave had no choice but to buy Core Scientific if it hoped to take on its OpenAI commitments. $CRWV $CORZ $NVDA $ORCL.
nextplatform.com
There are definitely easier businesses to be in than operating a neocloud. For one thing, the makers of AI accelerators, which are driving around half of
0
0
1
HPE has completed its Juniper buy, and now redrawing the switch and router roadmaps begins. $HPE $JNPR $NVDA $CSCO.
nextplatform.com
It has taken the better part of a year and a half and some wrangling with the US Department of Justice to get it done, but Hewlett Packard Enterprise has
0
0
1
If Google, AWS, Microsoft, and Meta can’t get the HBM they need for their XPUs, they will sue the crap out of Nvidia or AMD or the HBM makers. $MU $NVDA $AMD $GOOG $MSFT $AMZN $META.
nextplatform.com
Micron Technology has not just filled in a capacity shortfall for more high bandwidth stacked DRAM to feed GPU and XPU accelerators for AI and HPC. It has
0
0
5
The Doudna supercomputer at Lawrence Berkeley National Laboratory shows how HPC centers are going to have to adapt simulations and models for mixed precision and cannot rely on affordable FP64 compute. $NVDA $DELL $AMD.
nextplatform.com
Right or wrong, we still believe that we live in a world where traditional HPC simulation and modeling at high precision matters more than mashing up the
0
0
3
Thanks to AI, Nvidia is now a contender to be the largest datacenter Ethernet switch maker in the world. $NVDA $ANET $CSCO $HPE.
nextplatform.com
There is not one Ethernet business, but several, and now, with the evolution of Ethernet switches for back-end AI cluster networks, there is a new one
0
1
13
Intel has added top-notch engineers from Cadence, Apple, and Google to its AI efforts, and has an engineer leading sales. $INTC $AAPL $GOOG $CDNS.
nextplatform.com
It has been two and a half months since new chief executive officer Lip-Bu Tan gave the keynote at Intel’s Vision 2025 event, and the company has been
0
0
1
AMD finally gave out a full forecast for AI accelerator spending in the datacenter. The numbers are surprising in a number of ways. $AMD $NVDA.
nextplatform.com
Given two endpoints and a compound annual growth rate between those two points over a specific amount of time is not as useful as it seems. Not when you
0
0
5
A first look at the future GPUs and rackscale systems coming from AMD and its partners. Watch out, Big Green. Here comes Big Red. $AMD $NVDA.
nextplatform.com
To a certain extent, Nvidia and AMD are not really selling GPU compute capacity as much as they are reselling just enough HBM memory capacity and
1
4
6
If Germany wants its Jupiter supercomputer to break 1 exaflops on the LINPACK benchmark, it will need nearly 3,000 more nodes. $NVDA #Eviden #ParTec.
nextplatform.com
The newest of the exascale-class supercomputer to be profiled in the Top500 rankings in the June list is the long-awaited “Jupiter” system at
0
0
1