Chennai: Zoho founder Sridhar Vembu has warned that rapidly rising memory prices and increasing AI token costs are making it increasingly difficult for technology companies to do business without raising prices. Vembu said Zoho has so far held back from increasing prices, but the situation is becoming difficult as memory costs have surged sharply over the past year.
The comments come at a time when the technology industry is dealing with a major increase in demand for memory chips, largely driven by the rapid expansion of artificial intelligence infrastructure. AI data centres require enormous quantities of memory and computing hardware, putting pressure on global supply chains.
Vembu’s comments also highlight how the AI boom is affecting companies that are not directly involved in building AI data centres. Higher hardware costs can eventually filter through to software companies, cloud services and consumers.
Memory prices surge as AI demand grows
Vembu shared a post on X referring to a report that memory prices had risen by 500 per cent in 12 months. He said the combination of higher memory prices and rising AI token costs was creating a significant challenge for businesses.
“Memory prices, along with AI token prices, have made business very difficult,” Vembu said, adding that Zoho had held back from raising prices but was finding the situation increasingly difficult.
The scale of the memory-price increase is particularly striking.
According to data cited by India Today from Tom’s Hardware, average prices for certain DDR5 memory kits rose from around $90, or approximately Rs 8,600, in August 2025 to $425, or about Rs 40,000, in August 2026.
That represents a dramatic increase in just one year.
Higher-capacity memory has experienced even steeper increases. A 128GB DDR5-6400 kit was reported to be selling for $3,399, approximately Rs 3.25 lakh, compared with its lowest tracked price of $329, or around Rs 31,490.
Why AI is pushing up memory costs
The current memory crunch is closely connected to the rapid expansion of AI.
Modern AI systems require large quantities of high-performance memory and storage. Data centres operated by major technology companies are being expanded rapidly to support AI models, applications and services.
As these companies compete for computing capacity, demand for memory has increased.
This creates pressure throughout the semiconductor supply chain.
While memory manufacturers can increase production over time, expanding semiconductor manufacturing capacity requires substantial investment and cannot happen immediately.
The result is a market in which demand can rise much faster than available supply.
For companies such as Zoho, this creates an unusual problem. A software business may not directly purchase the same volume of memory as an AI data-centre operator, but its products still rely on servers, cloud infrastructure and computing resources.
Higher infrastructure costs can therefore eventually affect operating expenses.
Zoho has held back from raising prices
Vembu’s comments suggest that Zoho has attempted to protect customers from the increase in costs.
The company has so far avoided passing the entire impact through to customers in the form of higher prices.
However, Vembu indicated that maintaining that approach is becoming increasingly difficult.
For software companies operating on subscription models, pricing decisions can be particularly sensitive.
A sudden increase in subscription charges could make customers reconsider their plans, particularly in competitive software categories.
At the same time, absorbing higher infrastructure costs indefinitely could put pressure on profit margins.
Zoho therefore faces the same fundamental challenge as many other technology companies: determining how much of the increase in operating costs can be absorbed and how much eventually needs to be passed on.
AI token costs add another layer of pressure
Memory is not the only issue highlighted by Vembu.
He also pointed to rising AI token costs.
AI tokens are generally used as a unit for measuring the amount of text or other data processed by AI models. Companies using third-party AI models can therefore incur higher expenses as their usage increases.
As businesses integrate AI into more products and workflows, their dependence on AI infrastructure can increase.
This means a company could face rising expenses on two fronts: higher physical infrastructure costs and higher costs for AI services.
For software companies, this could make AI integration more complicated from a business perspective.
The technology may offer productivity and product benefits, but the cost of running AI-powered features must still be managed.
Vembu says programming languages need to evolve
Vembu believes the memory crisis could have implications beyond hardware prices.
He argued that programming languages were historically developed with an assumption that memory was relatively inexpensive and plentiful.
According to Vembu, that assumption may no longer hold.
“For a long time, programming languages were designed with the assumption that memory is ‘free.’ That era has now ended,” he said.
His argument is that developers may need to place greater emphasis on memory efficiency when designing software.
That could involve improvements to programming languages, compilers and software architecture.
The objective would be to reduce unnecessary memory consumption without compromising software safety or developer productivity.
Smarter compilers could become more important
Vembu specifically highlighted the need for more efficient compilers, including those used for AI-related workloads.
Compilers translate source code into instructions that computers can execute.
More efficient compilers can potentially help software use hardware resources more effectively.
If memory remains expensive, reducing unnecessary memory usage could become increasingly important for both software developers and companies operating large-scale computing infrastructure.
Vembu said his own research focus would include examining how programming languages can consume less memory.
He argued that software safety and developer productivity should not come at the cost of excessive memory usage.
Smartphone makers are already feeling the impact
The memory shortage is not limited to enterprise software companies.
Consumer electronics manufacturers are also facing pressure.
India Today reported that smartphone companies including Oppo, OnePlus, Vivo and Nothing have raised prices in recent months. There have also been reports that Apple could increase iPhone prices.
Memory is an important component in modern smartphones.
As devices increasingly include higher RAM capacities and larger storage options, memory costs can have a direct impact on manufacturing expenses.
Manufacturers have several options when component costs rise.
They can increase retail prices, reduce discounts, change specifications, absorb the additional cost or combine several of these approaches.
For consumers, the most visible consequence can be higher smartphone prices.
AI boom is reshaping the technology supply chain
The current situation illustrates the broader impact of the AI boom.
AI companies are investing heavily in data centres to support increasingly sophisticated models.
OpenAI, for example, has signed a deal with SoftBank’s SB Energy for a 10-gigawatt data centre in Ohio, according to India Today. The project is partly backed by a commitment from Nvidia.
Large investments of this scale create enormous demand for computing equipment.
The consequences are felt beyond the companies directly involved in AI.
Chip manufacturers, cloud providers, software companies, device makers and ultimately consumers can all be affected by changes in the cost of computing infrastructure.
India is also expanding data-centre capacity
The pressure on computing infrastructure is not limited to the US.
Indian technology companies are also planning large investments in data centres.
HCLTech has announced plans to invest Rs 3,500 crore in data centres, while TCS is planning to spend as much as $7 billion on a 1-gigawatt data-centre unit, according to the report.
Such investments reflect the growing importance of data-centre infrastructure as businesses adopt cloud computing and AI.
However, they could also contribute to continued demand for computing components.
As more data centres come online, demand for processors, memory, networking equipment, storage and power infrastructure is likely to remain strong.
Software companies could face margin pressure
The situation could have a significant impact on software businesses.
Software companies traditionally benefit from relatively low marginal costs when they scale digital products.
However, modern cloud-based applications depend on physical computing infrastructure.
If the cost of memory, processing and AI services rises substantially, the marginal cost of delivering software can also increase.
This is particularly relevant for companies offering AI-powered features.
An AI feature may attract customers, but each interaction can carry a measurable infrastructure cost.
Businesses therefore need to balance the benefits of AI adoption with the expense of operating AI systems at scale.
Could higher costs reach consumers?
Vembu’s comments raise the possibility that consumers could eventually see higher technology prices.
Companies cannot indefinitely absorb major increases in operating costs.
If memory and AI costs remain elevated, technology companies may eventually have to reconsider their pricing structures.
That could mean higher subscription prices for software services, more expensive smartphones and computers, or changes to the amount of AI usage included in different plans.
However, companies may also find ways to offset some of these costs through improved efficiency.
Better software optimisation, more efficient hardware and smarter AI models could reduce the amount of computing resources required for individual tasks.
Efficiency could become the next major priority
The memory crisis could therefore encourage a renewed focus on efficiency.
For years, falling hardware costs allowed software developers to prioritise speed of development and functionality over resource consumption in many applications.
When memory was inexpensive, using additional memory was often a relatively minor concern.
A sustained increase in memory costs could change that approach.
Developers may increasingly need to optimise applications, reduce memory overhead and design systems that can achieve more with fewer resources.
This could also encourage innovation in programming languages, compilers and operating systems.
AI companies face a similar challenge
The irony is that AI is simultaneously driving demand for computing resources and creating tools that could potentially make software development more efficient.
AI coding assistants can help developers write code faster, but AI workloads themselves require significant computing resources.
This creates a complicated economic equation.
Companies are adopting AI because they expect productivity gains, but those gains must be weighed against the cost of running AI systems.
If memory and token costs remain high, companies may increasingly favour smaller and more efficient models for tasks that do not require the most powerful systems.
The memory crisis may last longer
There is no indication that the underlying demand pressure will disappear quickly.
AI infrastructure investment continues to grow, while major technology companies are expanding their computing capacity.
That does not necessarily mean memory prices will continue rising at the same rate.
Semiconductor manufacturers can increase supply, and market conditions can change.
However, the current episode demonstrates how quickly an increase in demand from one part of the technology industry can affect businesses across the entire ecosystem.
For companies like Zoho, this makes cost management increasingly important.
What this means for the technology industry
Vembu’s warning is significant because Zoho is primarily known as a software company rather than a semiconductor or hardware manufacturer.
His comments demonstrate that the memory crisis has consequences far beyond PC builders and smartphone makers.
Software companies depend on computing infrastructure, while AI-powered applications add another layer of infrastructure costs.
If memory prices remain high, companies may have to rethink product pricing, infrastructure strategies and software architecture.
That could accelerate investment in more efficient systems across the technology industry.
Conclusion
Sridhar Vembu’s warning highlights a growing challenge for the technology industry: the AI boom is creating enormous demand for computing infrastructure, while rising memory prices and AI token costs are increasing the expense of running digital businesses.
According to data cited by India Today, certain DDR5 memory kits have risen from around $90 to $425 in a year, while a 128GB kit has reached approximately Rs 3.25 lakh.
For Zoho, Vembu says the company has so far avoided raising prices, but maintaining that position is becoming difficult.
His concerns also extend to the way software is designed. He believes the era in which developers could effectively treat memory as inexpensive is ending and that programming languages and compilers will need to become more memory-efficient.
The developments could have implications for consumers as well. Smartphone manufacturers are already raising prices, while businesses face higher infrastructure and AI costs.
As AI investment continues, efficiency may become just as important as raw computing power. The companies that can deliver software and AI services while using less memory and computing capacity could have a significant advantage in an increasingly expensive technology environment.