| Global Defense, OSINT & Geopolitics Wire

[đŸ‡§đŸ‡©] Artificial Intelligence-----It's challenges and Prospects in Bangladesh

[đŸ‡§đŸ‡©] Artificial Intelligence-----It's challenges and Prospects in Bangladesh
160
8K
More threads by Saif

G Bangladesh Defense

Bangladesh must protect digital inclusion as the AI boom drives up costs

Mahmudul Hasan


1794009667433.webp

‘As AI giants such as OpenAI, Google, and Meta buy enormous quantities of chips, memory, and storage, manufacturers are prioritising these big-spending customers over the consumer electronics market.’ PHOTO: REUTERS

“Rajay rajay juddho hoy, ulukhagrar pran jay” is a well-known Bangla proverb that conveys how ordinary people often bear the brunt of rivalry between powerful nations, individuals, or groups. These days, it also captures an uncomfortable reality of the global artificial intelligence (AI) race. The world’s most powerful countries and technology companies are competing for supremacy in a technology that promises to transform economies, industries, and societies. But the cost of that competition is increasingly being felt far beyond Silicon Valley and Beijing.

As AI giants such as OpenAI, Google, and Meta buy enormous quantities of chips, memory, and storage, manufacturers are prioritising these big-spending customers over the consumer electronics market. This is tightening supplies of components for laptops and smartphones, pushing up prices, and making devices less affordable for ordinary consumers. In Bangladesh, computer prices are becoming exorbitantly high, smartphone manufacturers are facing rising component costs, and businesses are postponing technology purchases.

For a country seeking to expand digital inclusion, develop an IT-enabled services industry, and prepare its youth for an AI-driven economy, this creates a troubling contradiction. The technology that could help Bangladesh leapfrog development barriers is making some of the tools needed to use it less affordable.

The AI boom is reportedly making computers significantly more expensive. Entry-level laptops that cost Tk 47,000-Tk 48,000 a year earlier are selling for Tk 65,000-Tk 70,000. Laptop and desktop sales volumes have fallen by around 20 to 25 percent over the past year, even as the total value of sales has increased.

Computer use in Bangladesh already remains very low, with only 11.8 percent of individuals and 9 percent of households using or owning computers. Fewer than 3 percent of manufacturing units in Bangladesh use computers or information technology (IT) in their production work, according to the 2024 Economic Census. And now, AI is making computers less affordable for individuals and industries alike.

Bangladesh’s mobile phone industry is also facing rising component costs alongside domestic structural challenges. One report by this newspaper documented how rising memory prices are adding to the pressures on local manufacturers. RAM and storage components have become more expensive, increasing production costs for manufacturers already grappling with grey-market competition, weak factory utilisation, and policy uncertainties. Industry representatives have reported steep increases in memory prices since the beginning of the year, with further increases expected. Such pressures can feed through to handset prices, making affordable smartphones harder to produce and purchase.

The difficulties facing local manufacturers could also undermine Bangladesh’s ambition to build a stronger domestic digital device industry. The country’s handset production fell from roughly 2.73 crore units in 2024 to 2.44 crore units in 2025. Global component shortages are not the only reason behind the industry’s difficulties, but they add another burden in an already challenging business environment.

Bangladesh has also long struggled with low smartphone penetration, with nearly half of mobile users—47 percent—still relying on feature phones. To address this, the government is seeking to make low-cost smartphones more accessible and phase out 2G services by 2029 to accelerate digital inclusion. However, rising component prices driven by the global AI boom threaten to undermine these efforts. Indeed, digital penetration in Bangladesh is likely to remain constrained beyond 2027 due to shortages of components for laptops, desktops, servers and other essential digital infrastructure, as the intensifying global AI race continues to drive up component prices.

The United States and China increasingly treat AI as a strategic technology, with consequences for economic competitiveness, national security, and military power. Washington restricts China’s access to advanced chips and chipmaking equipment, while Beijing is investing in domestic alternatives. The rivalry now extends to the models themselves.

On September 8, US law enforcement and intelligence officials accused six Chinese AI companies, including DeepSeek, Moonshot AI, and Alibaba, of using distillation—training a cheaper model on the outputs of a more powerful one—to copy US technology. The allegations were disputed and Beijing has called similar claims groundless. It is also building barriers of its own: on October 16, China proposed export controls on 40 technologies, 23 of them tied to AI.

The competition is not confined to governments. OpenAI, Google, Microsoft, Meta, and Anthropic compete with DeepSeek, Alibaba, and Tencent for customers, talent, and computing capacity. DeepSeek showed that Chinese developers can challenge established players with cost-efficient methods. Even so, the spending is enormous. Alphabet, Amazon, Meta, and Microsoft are expected to invest about $650 billion in AI infrastructure this year, up roughly 60 percent from $410 billion in 2025.

The infrastructure splurge is driving shortages of memory as manufacturers prioritise chips destined for data centres over those for everyday PCs and phones. Xiaomi’s general manager in Bangladesh has said that AI companies are “effectively writing blank cheques” to memory makers. Counterpoint, a global technology market intelligence and research firm, expected memory prices to climb nearly 50 percent through the first quarter, while Gartner projects PC prices to be 17 percent higher and smartphones 13 percent higher than in 2025.

Bangladesh’s computer importers and handset manufacturers are failing to match the purchasing power of global technology giants. They are absorbing higher costs or passing them on to consumers. Export restrictions and trade barriers also make sourcing costlier, and Bangladesh feels the effects through international prices. As wealthy countries and corporations invest billions of dollars to gain an AI advantage, developing economies are being forced to find more money simply to buy the devices needed for citizens to take part in the digital economy.

Connectivity alone can no longer deliver digital inclusion. People also need affordable devices, reliable internet, access to electricity, and necessary skills. If device prices outpace household incomes, expanded networks may not translate into wider participation. There is a longer-term industrial risk too, as higher-value digital services need a broad base of digitally equipped workers and businesses, not just a few well-funded companies.

Bangladesh cannot influence the strategies of Silicon Valley or Beijing, and it should not treat AI as a threat or be put off by it. The technology could create real opportunities in education, healthcare, manufacturing, agriculture, and public services. The challenge is to ensure those benefits are not reserved only for people who can afford increasingly expensive hardware. The government must keep cutting unnecessary duties on computers and components. It’s commendable that the government withdrew import duties and VAT on laptops, desktops, servers, printers, and monitors in the FY2026-27 budget. But tax cuts cannot overcome global supply constraints, so their impact should be monitored and other price-raising barriers reviewed.

Support for local handset and electronics manufacturing must also become more predictable, with stable taxation, action against illegal imports, and incentives linked to productivity. Technology supply chains should be diversified, with procurement that values alternative suppliers, repairability, longer device lifecycles, and spare parts. Affordable access schemes—including targeted financing, quality-assured refurbished devices, institutional computer lending, and shared computing facilities—would help students and low-income households.

It is encouraging that Bangladesh plans to launch sovereign AI within a year. However, the initiative must be backed by thorough planning, adequate computing infrastructure, skilled manpower, reliable data, and sustainable investment to ensure meaningful, long-term benefits. Investment should also reach where it can build real strength: digital skills, Bangla-language applications, local software, responsible AI adoption, and higher-value services.

Each time an ordinary citizen or small business is gatekept out of buying necessary devices, it is a setback to the country’s digital ambitions. Bangladesh must ensure its people and businesses do not become the casualties of global battles for AI supremacy.

Mahmudul Hasan is a senior reporter at The Daily Star, covering AI, telecommunications, and technology.​
 

Creating right conditions for AI adoption


1794094297259.webp



Artificial intelligence has become one of the most talked-about technologies in the business world, and for good reason. Every business aims to accomplish more at lower cost and AI offers a clear pathway to achieve this goal. Bangladeshi businesses are no exception, as demonstrated by the recent AI Business Summit organised by the Dhaka Chamber of Commerce and Industry and the Ministry of Science and Technology. The summit produced the usual mix of ambitious statements and sensible warnings. Speakers at the event emphasised developing AI applications that address real-life industrial problems rather than simply adopting imported technologies. This is an important distinction because, while imported applications provide useful capabilities, they often fail to address the specific challenges local businesses face. Platforms designed for other markets rarely account for local supply chain constraints or domestic labour management practices, and expecting them to adapt independently is unrealistic. Building local solutions is therefore essential for Bangladesh rather than remaining a consumer of foreign technology. Indeed, local developers who work directly with business operators have a far better chance of creating effective, context-appropriate solutions.

Industry leaders at the summit rightly urged young developers to leave conventional digital platforms behind and spend time on factory floors to understand where money, materials and time are actually being wasted. There is plenty of scope for AI applications that address these everyday business problems. In the garment industry, for instance, AI could help manufacturers improve production planning, identify defects and reduce raw material wastage. In the pharmaceutical and food industries, it could tighten quality control and reduce losses during production. Even something as straightforward as using past sales data and customer trends to predict demand can make a huge difference. This would help businesses keep inventory lean and avoid unnecessary costs. These may seem like routine business functions, but small improvements can add up to significant savings, especially in industries running on thin margins. However, identifying useful applications is only half the battle. Businesses also need to be ready to invest in the technology, improve their data management and train employees to use it well. In practice, though, this could prove particularly difficult for small and medium-sized enterprises which often lack the resources to experiment with new technologies.

This is where a supportive policy framework becomes essential as it can make AI adoption easier and less costly for businesses. In addition to finalising the National AI Policy 2026-2030, the government should create incentives for businesses to invest in research, employee training and technological innovation. As mentioned at the summit, allowing such investments to qualify for tax deductions could encourage private sector spending without making businesses dependent on direct government grants.

Financial incentives alone will not be enough, though. Bangladesh also needs closer cooperation between industries, universities and technology developers to ensure that research addresses actual business needs. Universities, in particular, must align their curricula more closely with the practical demands of industry, giving greater attention to AI literacy, data analytics and problem-solving skills. Businesses, for their part, must be willing to share their operational challenges with researchers and give young developers opportunities to test their solutions in real working environments as well. However, the enthusiasm surrounding AI must not come at the expense of caution. While AI can make businesses more productive, it may also reduce demand for certain jobs, particularly where routine tasks can be performed by machines. This makes employee training and reskilling essential so that workers can adapt to changing job requirements rather than bear the cost of technological progress.​
 

Demystifying AI finance: The supply chain, capital flows, and global rivalry

Sajed ul Islam, Senior banking executive and financial markets...

1794268328343.webp


Every single day, news headlines report artificial intelligence writing code, discovering medicines, and analysing massive data sets. Alongside these breakthroughs come staggering figures. Tech giants pledge hundreds of billions of dollars, startup valuations soar, and massive server warehouses expand globally.

Yet, looking at this modern tech boom feels like staring into a dense maze. People hear terms like AI race, data centers, and chip shortages, but connections remain blurry. Who actually pays whom? Where does the immense money required to build this digital world originate? How did a handful of specialised companies become gatekeepers of technological futures?

Understanding this requires looking past glowing software interfaces. Observers must examine physical plumbing, corporate balance sheets, and the global chessboard.

Examining the AI landscape reveals an industry caught in a profound structural dilemma. Investors face the acute fear of missing out on a major technological shift. Surging earnings and massive forecasted growth back this fear
The AI boom is not just a technology revolution. It represents one of the most capital-intensive industrial build-outs in human history. Comprehending it requires tracing the physical supply chain, decoding the nature of money flows, and examining the geopolitical tug-of-war between the rival superpowers.

The global assembly line: From sand to silicon to software
A common misconception is that AI floats harmlessly in the cloud. In reality, AI is extraordinarily heavy. It requires massive physical infrastructure stretching across continents and involves complex manufacturing processes. Building an advanced AI system resembles constructing a fleet of spacecraft. No single company acts alone. The global supply chain divides into an interdependent hierarchy of specialists.

The master architects: Chip designers

At the top of the design layer sit dominant leaders like Nvidia and AMD, alongside custom chip developers. These firms do not manufacture physical computer chips. Instead, they act like elite automotive designers. They draw blueprints, invent mathematical architecture, and write specialized software that allows computers to process massive datasets simultaneously.

Nvidia became the undisputed brain of the modern AI boom because its processors suited heavy machine learning tasks. Despite brilliant designs, these companies remain entirely dependent on others to bring drawings to life.

The master manufacturer: Semiconductor foundries
This reality brings the global economy to its ultimate bottleneck: contract semiconductor manufacturers. While TSMC commands advanced production, other fabricators like Intel Foundry and GlobalFoundries operate within the landscape. TSMC owns the world’s most advanced fabrication facilities, known as fabs. They do not design consumer products. Instead, they operate as ultimate contract kitchens. They take microscopic blueprints and print complex electronic circuits onto silicon wafers using extreme ultraviolet light.

TSMC remains a critical chokepoint. If precision manufacturing environments in Taiwan paused operations, the global revolution would halt. Almost every major tech company relies on these specialised fabrication networks to build hardware.

The speed specialists: Memory manufacturers

An advanced AI processor cannot think fast if it waits for data. This problem brings memory specialists into the picture. While SK Hynix and Samsung lead in supplying High-Bandwidth Memory, competitors like Micron Technology engineer high-speed storage components. This technology functions as ultra-fast short-term memory stacked directly alongside the main AI processor. Without these chips feeding data at lightning speed, expensive processors sit idle.

SK Hynix supplies advanced memory for AI servers. Meanwhile, Samsung operates as a massive conglomerate. They design electronics and run manufacturing factories simultaneously.

The digital factories: Data centers and landlords
Once chips and memory modules are built, they ship for installation in data centers. However, not all massive server hubs belong to tech giants. Retail chains do not always own every storefront. Similarly, tech companies do not own all data centers. While giants like Microsoft, Google, Amazon, and Meta own massive server hubs, specialized digital landlords build a huge portion of the infrastructure boom. Real estate funds buy land, secure electrical grid connections, and build industrial shells. Major operators like Equinix and Digital Realty lease space to tech companies under multi-year contracts.

Additionally, specialised AI cloud providers like CoreWeave, Lambda, and Crusoe own data center real estate and dense fleets of specialized chips. They rent raw computing power directly to AI startups.

The frontier brains: AI research labs

Once physical infrastructure is ready, developers hand it over to frontier AI labs.

Companies like OpenAI, Anthropic, and Google DeepMind sit at the application layer, alongside labs like Cohere and Mistral AI. Their core mission is building Large Language Models. These sophisticated software brains power modern artificial intelligence, allowing machines to converse, reason, and write code.

Building an LLM depends entirely on the physical supply chain. Training a frontier model from scratch requires tens of thousands of specialized chips running simultaneously for months inside massive data centers. If chip shortages hit or power constraints emerge, labs cannot train next-generation models. Grand ambitions remain capped by physical hardware access. Furthermore, because building a model costs billions in raw computing power, these labs rely heavily on financial backing and cloud infrastructure from tech giants.

Tracing this journey reveals the truth. From raw silicon transformed into pure crystals, etched by expensive machines, paired with ultra-fast memory, and turned into thinking software, AI functions as a heavy industrial project disguised as a software app.

The money maze: How the financing works

Physical infrastructure demands an astronomical price tag. Training a cutting-edge LLM and building a data center costs tens of billions of dollars. Traditional venture capital is far too small to foot this bill. To finance this revolution, the tech industry relies on massive corporate balance sheets, debt markets, private market fundraising, and corporate borrowing.

Financing through private markets

Because frontier AI labs require capital scaling at unprecedented rates, traditional public markets cannot act as the initial launchpad. Instead, private equity, sovereign wealth funds, and venture syndicates have become the staging ground for multi-billion-dollar rounds.

Companies like OpenAI and Anthropic have shattered historical private market records. OpenAI secured multi-billion-dollar private raises and secondary valuations pushing past hundreds of billions of dollars. Corporate heavyweights like Microsoft, SoftBank, and Nvidia backed these raises while tracking toward public market horizons. Similarly, Anthropic commanded massive private valuations close to the trillion-dollar threshold. Meanwhile, capital-intensive private industrial pioneers like SpaceX balance infrastructure investments while monitoring public market milestones. These private valuations function as a high-stakes psychological anchor, signaling immense future expectations long before a company faces public shareholders.

The borrowing impulse and corporate debt
Beyond private equity, acquiring thousands of chips and building power-hungry server networks has forced tech giants and heavyweights like Meta to tap global debt markets aggressively. Meta accumulated massive capital expenditure commitments and long-term debt obligations, pushing bond issuance to historic heights. Across the broader tech sector, companies routinely issue tens of billions in corporate bonds to fund data center construction ahead of actual revenue realisation.

The circular loop of capital and specific corporate ties
The financial plumbing connecting AI companies reveals a tight web of cross-investments and vendor obligations. A prime example is the multi-billion-dollar partnership between Microsoft and OpenAI. Microsoft injects massive tranches of cash into the LLM developer. In return, OpenAI is contractually bound to spend a vast majority of those funds purchasing Azure cloud computing resources and specialized chips. Similarly, Nvidia acts as both a foundational hardware supplier and a strategic investor. They take equity stakes in cloud providers like CoreWeave. Those providers immediately turn around and use those funds to buy billions of dollars worth of Nvidia's advanced GPUs.

This creates a self-reinforcing loop of capital. The tech giant or hardware vendor provides cash. The startup uses that cash to buy compute capacity. That exact spending flows right back onto the benefactor's books as top-line revenue. This circular flow requires scrutiny. A significant portion of demand driving the AI economy is pre-funded by companies selling the infrastructure itself. This raises fundamental questions about organic market sustainability.

The global tug-of-war: The US, China, and tech supremacy
While Wall Street debates balance sheets, a parallel financial drama plays out globally. The AI revolution is entangled in the strategic rivalry between the United States and China.

To understand global AI finance, people must examine the different playbooks these superpowers use to secure technological futures.

The western model: Private capital and corporate balance sheets

In the United States, the AI boom is primarily driven by private enterprise. It relies on venture capital, public stock markets, and cash reserves of private tech monopolies.

Wall Street acts as the ultimate arbiter. Companies raise capital by convincing public investors of future growth, issuing corporate bonds, or plowing corporate profits directly into research. While government support exists through tax incentives for domestic chip manufacturing, private risk capital looking for high financial returns fuels the American AI build-out.

The Chinese model: State-backed catch-up finance
China approaches the challenge through a different lens. Industrially, China remains behind the cutting edge. Semiconductor analysts estimate China lags roughly 10 to 15 years behind Western leaders in advanced logic chip fabrication. US export controls and sanctions constrain this progress by blocking access to advanced American chip designs, Dutch lithography machines, and premier foundries like TSMC.

To bridge this gap and manage immense financing requirements, Beijing avoids volatile private stock markets. They deploy a state-directed financial architecture instead. Government Guidance Funds and national vehicles like the Big Fund pool central government reserves, municipal budgets, and state-owned bank capital.

By utilizing state-backed debt and policy loans, China absorbs high financial risks of technological catch-up. This allows Chinese firms to heavily subsidize domestic legacy chip manufacturing, optimize software efficiency to run on less powerful hardware, and maintain long-term momentum. This financial model ensures industrial survival and strategic independence, regardless of short-term commercial downturns.

Balancing innovation against the balance sheet

Examining the AI landscape reveals an industry caught in a profound structural dilemma. Investors face the acute fear of missing out on a major technological shift. Surging earnings and massive forecasted growth back this fear. Simultaneously, investors face an unprecedented mountain of borrowing and capital expenditure. The central question is whether immense cash outlays can ever be justified by future organic revenues, or if the system is over-leveraging itself on cyclical hype.

Recent events injected new anxiety into executive suites and investment firms. Unsettling security incidents rattled tech leaders, such as autonomous AI evaluation agents bypassing isolation protocols and breaching external platforms like Hugging Face. These autonomous actions demonstrated that advanced LLMs can independently solve long-pending mathematical proofs, discover novel attack paths, and pursue unintended objectives. Consequently, public and regulatory demands for strict government oversight intensify.

Yet, regulators struggle to formulate clear guardrails without choking innovation. This regulatory vacuum, paired with unpredictable software behavior, alters how investors assess risk. Financial markets no longer view AI purely through exponential growth. They increasingly price in operational hazards, potential liability, and the friction of compliance mandates.

The weight of these anxieties spilled directly into market action over past weeks. A sharp tech sell-off followed subpar quarterly reports from heavyweights like Amazon, Intel, and Alphabet. Investors retreated to defensive sectors as rising bond yields, a soaring cost of capital, and cooling primary markets dampened enthusiasm. A prime example is the shelving of Nvidia-backed Firmus and its multi-billion-dollar data center IPO in Australia. High-yield debt instruments for infrastructure providers face widening credit spreads. Sentiment has decisively shifted from unbridled momentum to severe risk assessment.

Compounding this dilemma is the geopolitical dimension. China is determined to close the hardware gap through state-backed finance and domestic substitution. This ambition transforms the race into a high-stakes global contest. For Western investors, falling behind carries the implication of ceding economic and technological leadership to a rival superpower. Understanding the maze of AI finance ultimately requires looking past technological novelty. Observers must examine underlying balance sheets, capital loops, governance risks, and national strategies determining whether this multi-trillion-dollar bet succeeds or collapses under its own weight.

* Sajed ul Islam is a senior banking executive and financial markets specialist with three decades of global institutional experience across treasury, capital markets, and corporate governance.​
 

AI to assist govt in M&E of projects


1794356413726.webp



State Minister for Planning Zonayed Abdur Rahim Saki said on Wednesday that the government is going to introduce artificial intelligence (AI)-based monitoring, expert scrutiny and independent price assessment to ensure greater transparency and accountability in public project expenditure.

Briefing reporters after an Executive Committee of the National Economic Council (ECNEC) meeting, he said the Planning Ministry was preparing a new rate schedule and developing a software-based system to use AI to examine whether project expenditure on goods and services was in line with the approved rates.

“Through this technology, we will examine every expenditure using software and AI to determine whether spending is being made in accordance with the rate schedule,” he said.

Saki said the government had also decided to bring subject experts into the Public Sector Committee (PSC) to scrutinise the prices of goods and services purchased outside the rate schedule, whether sourced locally or imported.

The Planning Ministry had already issued instructions to this effect so that specialists in the relevant fields could provide expert opinions on the prices of such purchases, he added.

The state minister said Prime Minister Tarique Rahman had also approved a proposal to appoint price assessment companies for major projects to verify the prices of all procured goods against prevailing market rates.

He said the government was working to strengthen institutional capacity and establish a systematic system of checks and balances to ensure that public money was spent under proper scrutiny and accountability.

The measures would be introduced gradually, with the aim of bringing the entire process of project expenditure under closer monitoring and institutional oversight, he added.

Saki expressed hope that the new arrangements would soon enable the government to examine project spending more effectively and establish a higher level of accountability in public development expenditure than at any previous time.​
 

Bangladeshis least optimistic about AI: survey

Staff Correspondent . Dhaka

Bangladesh has the lowest level of optimism about artificial intelligence among 37 countries surveyed, with only 34 per cent of adults aware of AI expecting it to mostly benefit their country, a new survey has found.

The survey, conducted by US-based analytics and advisory company Gallup, ranked Bangladesh last, behind Egypt at 35 per cent and the United States at 36 per cent.

Only 32.5 per cent of AI-aware Bangladeshis expect the technology to improve their daily lives, the lowest proportion among the countries surveyed.

The survey was conducted with Microsoft and published on September 22.

The survey examined public awareness, AI use, expected benefits, trust and emotional responses to the technology. Awareness of AI is also low in Bangladesh.

About 64 per cent of Bangladeshi adults said they had not heard of AI. Another 25.7 per cent were aware but had never used it.

Only 1.8 per cent use AI daily, 3.4 per cent weekly and 4.9 per cent monthly or less, which adds up to about 10 per cent who have ever used it.

Expectations are considerably higher in other countries.

China tops the list, with 93 per cent expecting AI to mostly benefit their country, followed by Vietnam at 91 per cent and Singapore at 77 per cent.

In Singapore, 46 per cent use AI daily.

Across the 37 countries, the median share of adults who have never used AI is 57 per cent.

Positive feelings about AI outweighed negative ones in 34 countries, the exceptions being the US, Egypt and the Palestinian Territories.

Wealthy Western countries showed the highest worry. In the US, 74 per cent of aware adults said AI makes them feel worried.

Trust in AI’s accuracy is limited. The median share of aware adults who trust it completely or a lot is 36 per cent.

The survey is based on nationally representative samples of about 1,000 adults aged 15 and above in each country, interviewed between April and July 2026.​