Making, Growing and Building

The complete report has seven parts and one 204-page master PDF. The research cutoff for this part is 3 October 2026. Every figure retains its own observation year. Page references in the original reading guide below refer to this volume’s printed pages within the master PDF; use the contents here to move through the web edition.

What India makes.

From a field of wheat to a spacecraft. From an idea to a working business.

A bigger harvest sounds good. So does a bigger factory, a new business or a rocket that reaches the Moon. But I wanted to ask one more question after each headline: what exactly became bigger, and what does that change mean?

My interests in research, business and technology meet in this part. On MisterLove, I try to turn a complicated subject into something you can follow without already knowing its language. Here, that means explaining a farmer's household income, a factory's value added and a patent application in ordinary words.

The Matrix India posts started this series. Four of the rows I return to here concern startups, steel, coal and defence exports. I put those claims beside the original data, then widen the picture to farming, small businesses, electronics, research and space.

I have used 2014 as a comparison point. It helps organise the story, but it does not tell us who deserves every bit of credit. Farmers, workers, scientists, companies, states and the Union government all shape what a country produces. Many projects also take years to move from an idea to a result.

Some of the gaps matter as much as the increases. The latest income reading in the Situation Assessment Survey describes 2018-19; another survey offers a newer reading with a different definition. A space mission planned for next year is still a plan. Registered-business counts cannot tell us how many firms are profitable today.

You do not need to read this like a textbook. Pick a topic that interests you. The table gives the figures; the paragraphs explain each row and the question it can actually answer. I have kept the dates attached so you can see where the evidence stops.

Lovepreet Singh
MisterLove

Sources: 1, 2.

Read the number with its date.

The newest report and the newest observation are often different things.

FY means financial year: April to March. FY2025-26 therefore ends in March 2026. A crop year commonly runs from July to June. A survey can cover a still earlier period. Each table says which one it uses.

One lakh is 100,000. One crore is 10 million. A million tonnes is 10 lakh tonnes; a billion is 100 crore. A tonne is 1,000 kilograms. A quintal is 100 kilograms. These are different ways of writing amounts, not different kinds of progress.

Current-price rupees are the money values reported for that year. They include price changes. More rupees spent or earned does not automatically mean the same increase in physical output or purchasing power. A percentage-point change measures the gap between two percentages: 40% to 50% is 10 points.

The cut-off is 3 October 2026. This volume covers the main national measures in these subjects; it cannot list every crop, industry, invention or military programme. Parts 1-3 cover the economy, infrastructure and living standards. Parts 5-6 will cover institutions, safety, the environment, state differences and rankings.

FY2014-15 runs from April 2014 to March 2015 and crosses the change of government. It is not a fully earlier baseline.

The harvest is much larger.

This measures food produced. Farmer income needs a separate question.

Foodgrains include cereals, such as rice and wheat, and pulses, such as gram and tur. Rice and wheat are already inside the first row. Adding all three rows would count them twice. The earlier 245.80 million-tonne cereal total sometimes seen in tables leaves out pulses; it is not the full foodgrain total.

The later figures are the third advance estimates, released on 27 May 2026. They are estimates for the 2025-26 crop year, not final audited totals. The 2024-25 final foodgrain figure was 357.732 million tonnes. The newer estimate is about 5.3% above that.

Against 2013-14, the estimated foodgrain harvest is roughly 42% larger. Picture an earlier harvest of 100 tonnes becoming about 142 tonnes. Rice rose by about 44%; wheat by about 26%. These are changes in physical quantity, so inflation does not explain them.

A larger national harvest can come from more land planted, more crops taken from the same land, better yields and different weather. The total alone cannot split the increase between those causes. It also does not show whether every farmer earned more.

For a family growing wheat, the useful next questions are its yield per acre, selling price and cost of seed, labour, water and transport. The country can produce more food while some farms struggle. Availability, food prices and a farmer's profit each need their own evidence.

2013-14 is a final historical figure. 2025-26 is an advance estimate; later revisions can change it.

Sources: 3, 4, 39.

More crop area gets irrigation.

Count a field once, or count each crop season. The result is different.

Net irrigated area counts the land that received irrigation, counting each piece once in the year. One hectare is about 2.47 acres. The increase from 68.419 to 82.416 million hectares is about 20%.

Gross irrigated area counts each irrigated crop season. If the same hectare grows two irrigated crops in a year, it contributes two hectares to the gross count. Gross area rose by about 36%. It can grow faster than net area when land is used for more crop seasons.

The third row says that about 60 out of every 100 hectares counted across crop seasons received irrigation in 2023-24. The earlier figure was about 48. The change is 12.1 percentage points. It is not the share of India's total land, or the percentage of farmers with their own pump.

Water access can reduce dependence on the timing of rain. But being recorded as irrigated does not establish that water arrived in the right quantity, at the right time or at an affordable cost. A field with an unreliable supply can still be in this count.

The report is dated October 2025 and its newest national observation is provisional 2023-24. I have kept that date visible. This is not a measurement of irrigation in 2026. Groundwater levels, water quality and the cost of pumping also matter; environmental pressures are taken up in Part 5.

Source units are thousand hectares; the table converts them to million hectares. Gross area counts repeated crop seasons.

Sources: 5.

Two surveys tell the story.

Different dates and definitions. No interchangeable numbers.

SAS means Situation Assessment Survey. Its first figure describes July 2012 to June 2013; the second describes July 2018 to June 2019. An August 2026 reply still identifies the 2019 round as the latest SAS. It is not the only national survey offering information on agricultural households.

Between the two SAS reference years, average income increased by Rs 3,792 a month, or about 59%. It did not double over that measured interval. This is a nominal rise, in each year's rupees. Higher prices reduce how much of the increase buys extra goods.

The figure covers the household's members and can include crops, animals, wages and other work. It is not one person's salary or crop profit alone. SAS 2019 required a self-employed agricultural member and annual produce worth more than Rs 4,000. Families doing only farm wage labour were excluded. The income comparison deducts paid business expenses.

NABARD, the National Bank for Agriculture and Rural Development, runs the All India Rural Financial Inclusion Survey, or NAFIS. It reports Rs 13,661 a month for agricultural households in 2021-22. This is a newer reading, but it uses a different survey and eligibility threshold: produce worth more than Rs 6,500, with a self-employed agricultural member. It cannot simply replace the SAS endpoint.

Neither later figure describes income in 2026. Averages also hide differences between families, and counting unpaid family work as a cost changes the result. To judge a claim that income doubled, we need matched definitions, dates and a clear choice between ordinary rupees and purchasing power after inflation.

The table labels two separate surveys. A publication year is not its income reference year; no older reading is presented as 2026 income.

Sources: 6, 7, 41.

Farming includes much more.

Animals and fisheries are a large part of the production story.

Milk production was about 80% higher than in 2013-14. Fish production more than doubled, rising about 106%. Egg production nearly doubled, rising about 99%. The fish figure is provisional. These are physical quantities, not the rupee value of sales.

For eggs, one billion means 100 crore eggs. The latest total is therefore about 14,911 crore eggs in a year. It counts production, not the number of people eating eggs or the amount each family can afford.

The dairy report also gives milk availability per person: 319 grams a day in 2014-15 and 485 grams in 2024-25. That earlier year crosses the government change. Availability is production divided across the population; it does not mean every person drank that amount each day.

India was already the world's largest milk producer in 1998. A continued first-place rank and a rise in the amount produced are two separate facts. The later output is much larger, but the leading position did not begin after 2014.

These activities can give a household more than one source of earnings. Their costs still matter: feed, animal care, disease, electricity, storage and transport can affect what the producer keeps. A bigger national milk or fish total cannot establish a bigger profit for every dairy farmer or fisher.

More production also does not guarantee better nutrition in every home. Prices, distribution, diets and who gets the food shape that result. Part 3 examines nutrition directly rather than treating output as a substitute for a health survey.

Milk and egg series use the dairy report; fish is converted from lakh tonnes. All production figures keep FY2024-25 as their latest year.

Sources: 8, 9.

A price promise and a cash transfer.

The announced price, the actual sale and family income are separate amounts.

MSP means Minimum Support Price. It is an announced support price for specified crops. A quintal is 100 kilograms. Common-paddy MSP rose about 86%, maize about 84% and tur about 97% in ordinary rupees. Different crops received different increases.

An announced MSP does not establish that every farmer sold every crop to the government at that price. Actual procurement, local buyers, crop quality and access to a purchase centre affect what happens at the sale. The table records the policy price, not each farmer's received price.

The 2026-27 paddy announcement uses a cost of Rs 1,627 per quintal and a margin of about 50%. That cost includes paid expenses and an estimated value for family labour, often called A2+FL. It is not the wider C2 cost, which also includes items such as the rental value of owned land. A 50% margin on one cost measure does not automatically mean a 50% profit on every farmer's full costs.

PM-KISAN, introduced in 2019, pays eligible landholding farmer families Rs 6,000 a year in three Rs 2,000 instalments. The 23rd instalment, released on 20 June 2026, transferred Rs 18,880 crore to 9.44 crore farmers. These are that instalment's reported recipients, not a count of everyone working in agriculture.

For a recipient family, the annual benefit averages Rs 500 a month. It can help pay some expenses, but it is not the family's whole income. Adding it to an older survey average would invent a new national income figure; the survey year, eligibility and other earnings would need to be measured together.

The MSP comparison uses the same crop and kharif marketing-season series. All price increases shown are nominal.

Sources: 10, 11.

More people work in factories.

The newest factory survey was released on 30 September 2026.

The Annual Survey of Industries, or ASI, covers the registered factory sector. It does not count every workshop, home business or informal manufacturer. The latest results describe April 2024 to March 2025, although the release came in September 2026.

The first row includes factories in the survey's frame. The second counts those operating. The difference matters: a recorded factory is not automatically a working one. Operating factories rose by roughly 18% over this comparison.

Persons engaged includes workers and other people working in these establishments, such as management and support staff. The total rose by about 74.59 lakh, or 55%. It is a difference between two annual survey estimates, not a count of unique people who got a new permanent job during 2026.

Output is the total value produced. Gross value added, or GVA, subtracts the value of bought-in inputs. Imagine a factory producing goods worth Rs 100 using purchased materials and services worth Rs 60: the value it adds is Rs 40. That amount still has to cover wages and other claims; it is not the owner's profit.

GVA was about 2.53 times the earlier rupee value. Output was about 2.52 times. Both use current prices, so these multiples include inflation and are not physical-production growth rates. Survey coverage also evolves over time; these results describe the ASI sector, not all Indian manufacturing.

Source money tables use Rs lakh; these values are converted to Rs lakh crore and rounded. Employment is not a count of workers alone.

Sources: 12, 13.

Capacity is not the steel made.

A plant can be able to produce more than it actually produces.

Capacity means how much plants are equipped to make in a year. Production means how much they actually made. Think of a bakery able to bake 1,000 loaves a day but baking 750. Its capacity is 1,000; its output is 750. Neither number can replace the other.

Crude steel is the steel at an early production stage, before it becomes many finished products. Actual crude steel output more than doubled between 2013-14 and 2025-26, increasing by about 106%. The newer figure comes from the ministry's April 2026 release and is provisional.

The June 2026 compendium puts capacity at over 220.4 million tonnes a year. Capacity also more than doubled against the 2013-14 baseline. The capacity recorded near a year's end and the output made throughout that year do not give a precise utilisation rate simply by division.

The Matrix chart shows 109 to 222 million tonnes. The historical ministry table identifies 109.85 as capacity in 2014-15, a year spanning the government change. It gives 102.26 for 2013-14. Neither is that year's actual crude steel output.

The chart's later 222 is close to the recent capacity scale, but it is not the exact June compendium figure. Without a matching source date, I cannot reproduce that number precisely. What the dated sources do show is a large rise in both productive capacity and actual output.

More steel can support buildings, machinery and transport. It does not by itself tell us the profit of steel companies, the quality of every product, the jobs created or the environmental cost. Those outcomes require separate measurements.

The chart row is explained as a capacity claim. The report keeps capacity and physical production in separate rows.

Sources: 2, 14, 15, 16.

Much more coal. A recent dip.

A long-term rise can contain a year when production falls.

The latest full-year figure is about 84% above 2013-14. A production level of 100 in the earlier year would be about 184 now. These figures measure coal mined in India; they are not mining capacity or the monetary value of coal.

The newest year is slightly below the preceding one: 1,040.08 versus 1,047.52 million tonnes. That is a fall of 7.44 million tonnes, or about 0.7%. A report card should show both the long-term increase and the most recent change. 'It kept rising every year' would go further than these figures allow.

The Matrix row shows 565 to 1,000 million tonnes. Its earlier number is a rounded version of the historical scale. The later rounded number captures the move past one billion tonnes, but the current full-year observation is 1,040.08. The years need to travel with the numbers.

Higher domestic production can help supply power stations and industry. It does not automatically end imports. Coal comes in different grades, and steelmaking needs coking coal. A country can produce more of some grades while still buying others abroad.

These coal totals should not be mixed with lignite, a different fuel that has its own statistics. Nor should mined tonnes be added to power-generation capacity measured in megawatts. One counts fuel produced; the other counts the power equipment installed.

Supply security and environmental effects are different questions. More coal can support energy availability while increasing pressures that need to be managed. Air quality, emissions and the wider energy balance belong in the environment discussion in Part 5.

The table uses full financial years. FY2025-26 ended in March 2026; it is not a partial calendar-year total.

Sources: 2, 17, 18.

India makes and exports more.

Production value does not reveal how much of each product is made locally.

The matched table comes from a July 2026 parliamentary reply. Its earlier year is 2014-15, not a strictly pre-2014 baseline. All values are in the rupees of the reporting year. Production counts value made in India; exports count value sold abroad.

All-electronics production became about 6.9 times the earlier value. Electronics exports became about 11.2 times. Mobile production and exports grew much faster. Using the rounded endpoints above gives about 35 times and 173 times respectively; these are approximate ratios, not exact unrounded results.

The release also calls the mobile increases about 33-fold and 165-fold. Those labels do not reproduce exactly from its displayed rounded baselines. The table therefore keeps the endpoints visible.

A phone assembled in India can still contain imported chips, displays and other parts. The value of the finished phone is larger than the value added inside India. This table establishes growth in production and exports; it cannot establish that every component is domestic.

Mobile figures are part of the wider electronics figures. Adding them together would double-count mobiles. Exports are also part of production and sales, rather than an extra amount of GDP to add on top.

To judge how deeply manufacturing has developed, we would also want local component production, design work, domestic value added, worker pay and productivity. A larger sales-value table is useful evidence, but it does not answer all those questions by itself.

Rs 1,00,000 crore = Rs 1 lakh crore. Electronics production of Rs 13,11,000 crore is Rs 13.11 lakh crore.

Sources: 19.

More registrations need context.

A survey estimate and an online register count different things.

MSME means micro, small and medium enterprise. The older figure estimates unincorporated, non-agricultural enterprises covered by the National Sample Survey, or NSS. Its scope excludes some registered companies, registered factories and construction. It is not a count of every Indian business.

The newer figure is an administrative registration headline. Udyam and the complementary Udyam Assist system bring firms into formal recognition, including existing informal micro businesses. Registering an old shop does not mean the shop was newly created. Eligibility and coverage also differ from the older survey.

The survey estimated 11.10 crore people employed in its covered enterprises in 2015-16. The August 2026 backgrounder says registered MSMEs employ more than 40 crore. These are different systems, so subtracting them would not give jobs created. The newer headline is not an independent survey of unique, currently active jobs.

A separate industrial support programme is PLI: Production Linked Incentive. It supports additional eligible production or sales across 14 sectors. The September 2026 government summary reports, up to June 2026, Rs 2.6 lakh crore of investment, Rs 23.8 lakh crore of production and sales, over Rs 15.5 lakh crore of exports and 14.6 lakh jobs.

Those PLI amounts are programme totals accumulated over time, not one year's national manufacturing output. Investment is not the government's incentive bill; sales are not profit. Exports overlap with sales, so the amounts should not be added. The reported jobs also do not establish the net jobs caused by the programme compared with what would have happened without it.

2015-16 is after 2014. No percentage-growth claim is made from the unmatched MSME survey and registration counts.

Sources: 20, 21, 22.

Recognition began in 2016.

India had young businesses before it had this particular official register.

DPIIT is the Department for Promotion of Industry and Internal Trade. Startup India began in January 2016. Recognition is an official status for entities meeting eligibility conditions; it is not a census of every new business, or a guarantee that a recognised firm will succeed.

The rise from 452 in the scheme's first year to about 2.54 lakh is a large expansion of recognition. Part of the early-to-later difference reflects the register building up after it began. New recognition can include a business that already existed before it applied.

The Matrix chart compares 500 before 2014 with 2,30,000 later. The later number is within the scale of the expanding official register, although the September 2026 update is higher. But the earlier 500 cannot be a 2014 count under a recognition system launched in 2016.

I have not found a matching source definition that makes those two chart endpoints the same measure. That leaves the exact 500-to-2,30,000 comparison unresolved. It does not mean India had no startups before 2014, and it does not erase the growth documented after recognition began.

A recognised-startup count tells us about participation in the system. It does not tell us how many firms remain active, earn a profit, survive several years or create secure jobs. Those need business-survival and employment data. Eligibility rules also shape who can enter the count.

A 'unicorn' is a private company valued at at least one billion US dollars. That valuation is an estimate used in investment deals; it is not cash earned, annual production or the founder's money in a bank account. Large startup counts and valuations should therefore be read alongside actual business results.

There is no invented zero for startups in 2014. The comparable administrative series begins after the comparison boundary.

Sources: 2, 22, 23, 40.

More money goes into research.

The newest observed total and the projected total are not interchangeable.

R&D means research and development: work to understand something new or develop a new product, process or technology. GERD means the total spent on this work inside the country across the sectors covered by the statistics. It is wider than the budget of one science ministry.

The latest observed expenditure in DST's 2025-26 statistical publication is for 2023-24. It is about 3.08 times the 2013-14 amount in current rupees. This includes price changes; it does not establish that research activity or its results tripled.

The GDP share asks how large research spending is relative to the economy. At 0.84%, it is roughly 84 paise for every Rs 100 of GDP. The older share uses the 2011-12 GDP base and the newer one the 2022-23 base. That change prevents a perfectly matched comparison.

The 2025-26 figure is projected from recent sector growth rates. It is not a fresh observation of all spending in that year. The report labels it as a projection, so the 0.90% headline should keep that label too.

For research papers, the report uses a US National Science Foundation data series: 78,135 in 2012 and 2,07,390 in 2022. The later observation remains 2022. The count rose about 2.65 times; it measures papers counted in that series, not the quality or usefulness of every paper.

Spending, papers, skilled researchers and useful discoveries are connected, but they answer different questions. A stronger report card needs both resources and results. A larger budget cannot tell us by itself whether a treatment worked, a new machine reached customers or a paper changed what scientists know.

DST = Department of Science and Technology. The last row is projected. GDP-base changes and current-price rupees are stated explicitly.

Sources: 24, 25.

More applications. Uneven grants.

Asking for protection and receiving it are different events.

A patent application asks for legal protection for an invention. A grant means the office accepted an application after examination. One is a request; the other is a later decision. The baseline is 2014-15, not fully before the government change.

Applications were about 2.58 times the earlier count. Indian applicants' share rose by 33.56 percentage points. The all-applicant total also includes foreign applicants seeking protection in India; it should not be presented as a count of inventions made only by Indians.

Grants show a less smooth path. DST reports 1,03,057 grants in 2023-24, followed by 33,504 in 2024-25. The latest level is well above 2014-15 but much lower than that recent peak. Presenting the peak as the newest annual result would conceal the drop.

Applications and grants need not come from the same batch. A patent granted this year may have been applied for years earlier. Dividing this year's grants by this year's applications would therefore not give an applicant's chance of approval.

Changes in grants can reflect the flow and timing of examination as well as the earlier stock of applications. These totals alone cannot identify the cause of the latest fall. They do not prove that inventions suddenly became less useful, or that a particular administrative change caused it.

A granted patent is also not a product already sold. Some inventions reach the market; others do not. Applications, grants, licensing income and successful products are separate steps. For a small inventor, receiving a grant can matter greatly, but the national count cannot tell us the commercial outcome of every invention.

DST Tables 30 and 33. Latest verified annual patent counts here describe FY2024-25, not inventions or decisions made throughout 2026.

Sources: 24, 25.

Old capability. New milestones.

The story crosses 2014, and it includes setbacks as well as achievements.

India already had a lunar mission before 2014. Chandrayaan-1 was an orbiter: a spacecraft studying the Moon from space. Chandrayaan-3 demonstrated a soft landing, reaching the surface under control. Orbiting and landing are different technical achievements.

The Mars mission makes the date boundary especially clear. It launched in November 2013 and reached Mars in September 2014. Its planning, construction, launch and arrival cannot all be assigned to one side of the 2014 line.

SpaDeX demonstrated docking: bringing two spacecraft together and joining them in orbit. That is a useful capability for more complex missions. A successful demonstration does not mean every future mission using it has already succeeded.

The 2026 PSLV-C62 mission also belongs in the record. ISRO reports an anomaly near the end of the third stage. A launch taking place is not automatically a successful delivery of its intended spacecraft. Counting only celebrated achievements would leave out part of the programme's performance.

These are selected mission milestones, not a total of all launches. Launch count, satellites carried, mission complexity, reliability and useful data are different measures. One rocket can carry many satellites, so more satellites does not necessarily mean the same increase in rocket launches.

Dates refer to actual events, not announcement dates. The Mars mission spans the comparison boundary.

Sources: 26, 27, 28, 29, 30.

About twice the full allocation.

Use the same budget stage, then keep future missions separate.

Budget Estimate is the amount planned at the start. Revised Estimate is the updated plan later in the year. Actual expenditure is what was spent. Comparing the full original allocations, 2026-27 is about 2.02 times 2013-14 in current rupees, not a doubling of purchasing power.

The original 2013-14 budget contains Rs 5,615 crore under the old Plan category and Rs 1,177 crore under Non-Plan. Together they make Rs 6,792 crore. Comparing only the earlier Plan portion with the later full total would inflate the rise.

The 2026-27 budget document also shows that the 2025-26 estimate was revised down from Rs 13,416.20 to Rs 12,448.60 crore. Neither estimate is a final bill. The actual-expenditure row shown here is the 2024-25 net figure in that document.

An August 2026 reply targets the first uncrewed Gaganyaan experimental mission for October-December 2026. Further uncrewed flights and the first crewed mission are targeted by 2027. 'Uncrewed' means without people aboard. These are target dates; they do not count as completed human flights.

The same reply targets a fully operational Bharatiya Antariksh Station by 2035. It records an approved Rs 20,193 crore programme budget for eight missions. Approval is not the same as money already spent, and a budget spread across missions is not a single year's departmental allocation.

Historical allocation categories have changed. The comparison uses full Budget Estimates; all Gaganyaan and station targets retain their source date.

Sources: 31, 32, 33.

Defence production value grew.

Production value, spending and readiness are separate measures.

Defence production reached about Rs 1.78 lakh crore in 2025-26: about 4.07 times 2013-14 and roughly 16% above the rounded 2024-25 value. These rupee totals include changes in prices, the amount produced and the mix of equipment.

The latest production split is about 76% from defence public-sector companies and other public units, and 24% from the private sector. Public production therefore remains the larger part of the reported total. A growing private sector does not mean public units have disappeared.

Production value counts the sector's output. It is not the government's defence budget, the value of exports alone or the number of weapons available to troops. A high-value aircraft, a ship and many small components can all contribute money values without being comparable units of equipment.

The Union's 2026-27 defence allocation is about Rs 7.85 lakh crore. Its capital allocation is about Rs 2.19 lakh crore, including about Rs 1.39 lakh crore earmarked for capital procurement from domestic industry. Capital spending covers longer-lived assets; the wider budget also pays for other needs.

An earmark is a planned allocation. It does not mean every purchase is complete or every rupee has been paid to a supplier. These budget amounts cannot be added to production value as though they were separate pieces of national output.

Domestic production can build skills and supply capacity. Imported components, maintenance, delivery delays and performance require separate evidence. This table records production value; it cannot establish complete military self-sufficiency or combat readiness.

The later production values are ministry-rounded current-price totals. FY2026-27 defence spending is a Budget Estimate, not actual expenditure.

Sources: 34, 37.

Defence exports rose sharply.

Selling more abroad and buying major weapons abroad can happen together.

The latest export value is about 56 times the 2013-14 figure. That means an increase of about 5,501%, using the earlier value as the starting base. It rose about 62.7% from 2024-25. These are rupee export values, not the number of weapons exported.

In 2025-26, public-sector defence companies exported Rs 21,071 crore and private companies Rs 17,353 crore: about 55% and 45% respectively. The ministry reports 145 exporters and destinations in more than 80 countries. These counts show reach, not profit or wholly Indian content.

The Matrix chart's 700-to-38,424 comparison is close to the official rounded-baseline story. The exact historical figure is 686. Its percentage method divides the increase by the later total. That answers a different arithmetic question from growth, which divides by the earlier total.

For a simple example, 100 becoming 200 is 100% growth. The added 100 is also 50% of the later 200. Neither calculation tells us that a government caused that percentage of the result. Dates and a subtraction cannot separate policy effects from other influences.

The Stockholm International Peace Research Institute, or SIPRI, places India second among major-arms importers in its March 2026 report. India had 8.2% of global imports in 2021-25. That is its share of the world import market, not the share of its own equipment imported. Its import volume fell 4% against 2016-20.

SIPRI measures major-arms transfer volume using its own indicator; the ministry measures exports in money. The two cannot be divided to calculate a self-reliance percentage. Both sources show why the report card needs export growth and continuing import dependence on the same page.

The four Matrix rows covered here are steel (p10), coal (p11), startups (p14) and defence exports (p20). Full cross-chart checks continue in Part 6.

Sources: 2, 35, 36, 38.

Sources and reading notes