The last time that Dell had a blowout quarter like the one that it has turned in was probably back when Michael Dell was selling custom PCs from his dorm room at the University of Texas. More than four decades later, the iron driving the business to new heights is, of course, AI systems, and is much bigger and more expensive but only delivering mid single digits of operating income. Which means Dell has to sell a lot more of it to get a return.
Luckily, Dell has enough orders for AI systems that bending metal around GPUs to increase its profits, which is better than not having those operating income dollars at all – even if its other businesses, which are smaller, are more profitable. Dell’s other option was to let Hewlett Packard Enterprise, Supermicro, Lenovo, and the ODMs make all of the money and sit out the GenAI boom.
It was smarter for Michael Dell to wait for Nvidia co-founder and chief executive officer Jensen Huang to get to the point that he needed the OEMs to reach enterprises, neoclouds, governments, and sovereigns to keep the Nvidia on its explosive growth trajectory. And as you can see from Dell’s financial results for the first quarter of its fiscal 2027 year, Dell’s server business is now going exponential and is fully benefitting from the GenAI boom. And as far as operating income goes, Dell is content to make it up in volume.
In the quarter, product revenues rocketed up by 2.2X to $38.11 billion – there’s some rounding up in those last digits of both of those numbers – thanks by and large to the huge costs of the AI systems that Dell has finally been granted the GPU allocations to build on behalf of customers.
Yes, it is like that. Nvidia is no doubt a kingmaker, as we pointed out way back in February 2024 in He Who Can Pay Top Dollar For HBM Memory Controls AI Training. Dell could have been selling AI systems at a higher volume years and years ago if all of the hyperscalers and cloud builders were not allocated most of the GPUs from Nvidia. What applied to AI training two years ago applies to AI inference equally well now. Because Nvidia can pay whatever it wants for HBM because it can charge whatever it wants for GPUs, it controls the flow of AI compute every bit as much as Standard Oil under John David Rockefeller controlled the refining and distribution of oil products in the United States more than a century ago.
We wonder what Jen-Hsun Huang’s middle name is, and when people might start using it. . . . Jen translates to “benevolence” and Hsun translates to “achievement” in Chinese, and Huang is of course one of the many variations of the name that means “yellow,” which was the symbol of imperial power in the Middle Kingdom, named after the mythical founder of Chinese civilization, the Yellow Emperor, known as Huangdi. This is a powerful name.
Anyway, back to Nvidia anointing Dell as the key AI systems supplier here in the United States and maybe in Europe, too, and benefitting from that benevolence and achievement.
Services revenues, which trundle along at their own pace and don’t seem to be bolstered by the GenAI tsunami, were off seven-tenths of a point to $5.74 billion. Ho hum.
Add it all up and Dell raked in $43.84 billion in the first quarter of fiscal 2027 ended in April, an increase of 87.5 percent year on year and 31.3 percent sequentially. Operating income rose by 3.1X to $3.66 billion and net income was up a little faster at 3.6X to $3.44 billion. This is the moment that Dell, the man, has been waiting for. The top line is growing thanks to GPU, CPU, main memory, and flash memory shortages, with Dell passing along those increases to customers, and the bottom line is being bolstered even moreso as enterprises, sovereigns, neoclouds, and model builders are getting into the GenAI game and are turning to Dell for political as much as strategic reasons. The latter group pays a lot more per unit for the AI systems and the traditional systems that feed into them, and this is where Dell gets its profits and always has. A big deal with xAI, as Dell did with the Colossus supercomputer in Memphis, is a loss leader that demonstrates cred and scale.
Loss leading, as you all know, is what the biggest supercomputer deals have always been about. You make it up by selling smaller systems to smaller customers at much higher margins.
Let’s drill down into Dell’s datacenter business, which is what we care about here at The Next Platform. And let’s posit two questions and try to answer them:
One: How much of the revenue increase being driven by higher CPU, memory, and flash prices?
Two: How much operating income did Dell make just on the increase in sales of traditional, non-AI servers?
Dell’s datacenter business hit a new high in the history of the company, with revenues up by a factor of 2.8X year on year and up 48 percent sequentially to just a tad bit over $29 billion. I remember when all of Dell was a $30 billion company (fiscal 2002) and a year or so later it declared its ambitious goal of being a $60 billion company by fiscal 2006 – a goal it hit a year early in fiscal 2005.
Now, here we are in fiscal 2027, and the company believes it will have $60 billion in AI server sales, which is a 20 percent increase over the $50 billion it was projecting only a quarter ago. Clearly, Nvidia has enough HBM memory and enough GPU chips to ramp Dell up. There may be a little wiggle room to push that even higher, but as far as we know, all of Nvidia’s GPUs have been allocated between now and the end of calendar 2027. The memory makers can only make HBM so fast, and at the prices and profit margins they are commanding, there is very little incentive for them to try to build HBM capacity any faster than they are doing. They are getting so filthy stinking rich from the GenAI boom that employees not involved in the memory business at SK Hynix and then Samsung were protesting the massive six-figure bonuses of employees who did. Micron employees are probably watching this very carefully.
In the first quarter, Dell sold $4.33 billion in storage, up 8.5 percent but down 9.7 percent from its fourth quarter of last year – pretty good, historically speaking. Servers and networking, however, accounted for $24.68 billion, up by a factor of 3.9X year on year and up 66.7 percent sequentially. All told, the ISG business had $3.06 billion in operating income, which represented 10.5 percent of revenues.
Let’s take a stab at the two questions that I raised. In our model, storage operating income is running at about 12 percent at Dell, nearly twice that of the 6.2 percent of revenue I had for the year ago period and driving $520 million of that $3.06 billion in overall ISG operating income. The increase in costs of CPUs, memory, and flash as well as richer configurations of PowerEdge 17G and 18G servers is what nearly doubled sales of traditional servers by 89.3 percent. Of this, I reckon that maybe 20 percent to 25 percent of that increase was driven by cost of materials passthrough, and the rest was driven by higher volumes and richer configurations.
Those richer configurations as well as higher volumes drove operating profits from $639 million for traditional servers in the year ago quarter to $1.54 billion in the quarter just ended. That is a 2.4X increase, and it shows that customers are indeed buying new servers to consolidate servers that are four, five, or six years old to make way for AI systems, which need the budget, power, and cooling being wasted by this ancient iron. The operating income with traditional servers was 14.2 percent of revenue a year ago and has risen to 18 percent in the current quarter in my model.
AI systems drive revenue, but even Dell concedes that they only deliver “mid single digits” of operating income. I think it was 5.5 percent a year ago, meaning only $99 million against $1.81 billion in AI system sales, but has grown to 6.2 percent in the current quarter against $16.13 billion, which is just a tad under $1 billion. That is a factor of 8.9X increase in revenue driving a 10X increase of operating income. This is about as well as any OEM or ODM could do, given the situation.
Dell has a $51.3 billion backlog for AI systems as the quarter ended, up by a factor of 3.6X from a year ago and up 19.3 percent sequentially. And it absolutely knows when it will get what GPUs and who will get what PowerEdge systems using them for that backlog.
We wonder what Dell is thinking about Nvidia’s entry into the PC market today at the Computex conference in Taiwan. Dell, the man and the company, will be enthusiastic about the idea if it knows what’s good for it.
