TMT Cases 2026: Three Letters, Three Opposite Trends

Telecom, media and technology moved in three directions last quarter

The Census Bureau publishes quarterly revenue for all three inside one table, which makes the comparison checkable rather than assertable. All figures below are seasonally adjusted, in millions of dollars, for the second quarter of 2026 against the second quarter of 2025, checked 30 September 2026 at census.gov.

Every percentage above was recomputed by hand from the two revenue columns. Software rose 22,739 on a base of 154,219, which is 14.74 percent. Wired telecom fell 1,795 on 71,496, which is 2.51 percent. Motion picture fell 1,894 on 36,717, which is 5.16 percent.

Now read the headline against its members. The sector grew 9.9 percent, and software plus hosting are 301,038 of 704,388, which is 42.7 percent of the revenue. Those two contributed 22,739 plus 20,920, so 43,659 of the sector's 63,321 of growth, which is 68.9 percent. Meanwhile the two telecom lines together removed 2,701, which is minus 4.3 percent of the sector's growth.

So the 9.9 percent figure describes almost none of the companies inside it. A candidate who accepts a TMT growth rate as a market assumption has imported a number that is wrong for two thirds of the sector. The Census table does not publish a consolidated telecom line in its adjusted series, so the 151,459 above is the sum of the two published sub lines, and it should be quoted that way.

Which of telecom, media and technology am I actually in

Establishing the sub sector is the first analytical step rather than a stalling question, and each one has a single number that decides its cases.

1. Telecom is capital against take rate. Network cost is sunk before a single customer subscribes, so the whole business case turns on what fraction of the homes or sites reached actually buy.
2. Media is fixed content cost against hit rate. Each title costs the same whether it works or not, and the returns are concentrated in a few titles, so the portfolio arithmetic dominates.
3. Technology is near zero marginal cost against retention. The cost of serving one more customer is small and the cost of replacing a lost one is large.

Altman Solon is the firm that works only in this sector, and it says so plainly, describing itself as "the world's largest strategy consulting firm exclusively focused on telecommunications, media, and technology", checked 29 September 2026 at altmansolon.com. The same page states that the firm cares most about "your analytical chops, knowledge of our industries, and ability to work well with others."

That second quotation is unusually direct and worth taking literally. Sector knowledge sits in the top three, which means arriving with the right denominator for the right sub sector is part of the assessment rather than a bonus. It also means the interview will probe whether you know why a network business and a software business are underwritten differently.

Format matters here too, because it changes how the numbers arrive. The firm publishes that its American analyst process runs "two rounds", with the first being "two 1-hour interviews, each with an interviewer-led case and a brief behavioral portion", checked 29 September 2026 at altmansolon.com. An interviewer led case hands you data in a fixed order, so the skill being tested is arithmetic under direction rather than free structuring.

Telecom: the take rate decides the whole TMT case

Here is a network case with every input stated. An operator plans to pass 100,000 homes with fiber. Passing costs $900 a home, connecting a subscriber costs a further $600, monthly revenue per subscriber is $70, and gross margin on that revenue is 70 percent. Site operating cost is already inside the margin. The client underwrites capital on payback.

Passing capital first. 100,000 homes at $900 is $90 million, and that money is spent whether anyone subscribes or not.

At a 35 percent take rate the operator wins 35,000 subscribers, so 35,000 connections at $600 is $21 million of connection capital. Total capital is $90 million plus $21 million, which is $111 million.

Annual cash per subscriber is 70 percent of $70, which is $49 a month, so $588 a year. Then 35,000 subscribers at $588 is $20.58 million a year. Payback is $111 million divided by $20.58 million, which is 5.4 years.

Now drop the take rate to 25 percent. Subscribers fall to 25,000, connection capital falls to $15 million and total capital to $105 million. Annual cash is 25,000 subscribers at $588 is $14.7 million, so payback is $105 million divided by $14.7 million, which is 7.1 years. Ten points of take rate moved payback by 1.7 years.

Solve it backwards and the case becomes a decision. Write the take rate as t. Capital is $90 million plus $600 times 100,000t, and six years of cash is 6 times $588 times 100,000t. Setting them equal gives 90,000,000 plus 60,000,000t equal to 352,800,000t, so 292,800,000t equals 90,000,000 and t is 30.7 percent. Everything else in the case is commentary on whether 30.7 percent is reachable in this footprint.

Oliver Wyman's published infrastructure case shows the convention being applied to a refusal, concluding that an opportunity "does not align with the client's investment criteria for a five-year payback period", checked 29 September 2026 at oliverwyman.com. At a five year bar this fiber build fails at both take rates, so ask for the bar before computing.

Media: the slate needs 1.62 hits

Media reverses the structure. There is almost no capital and a great deal of fixed cost per title, and the returns are concentrated.

A studio plans ten films. Production costs $60 million each, so 10 films at $60 million is $600 million. Marketing costs $30 million each, so 10 films at $30 million is $300 million. Total commitment is $900 million.

Assume a hit returns $400 million to the studio and a miss returns $30 million. With two hits the slate returns two at $400 million plus eight at $30 million, which is $800 million plus $240 million, so $1,040 million. Net is $140 million on $900 million of spend, a return of 15.6 percent.

Change the hit count by one in each direction and the business changes character. With one hit the slate returns $400 million plus nine at $30 million, which is $670 million, a loss of $230 million. With three hits it returns $1,200 million plus seven at $30 million, which is $1,410 million, a gain of $510 million.

That is a swing of $740 million between one hit and three, on $900 million of spend, which is 82.2 percent of the commitment riding on a count that moves by two. No amount of cost discipline touches it.

The breakeven hit count follows directly. With h hits, returns are 400h plus 30 times ten minus h, which has to equal 900. That gives 370h equal to 600, so h is 1.62 films. The slate needs one and two thirds hits to cover itself, which in practice means two, and the entire strategy conversation is about whether the studio can reliably produce two hits in ten attempts.

This is why media cases reward portfolio thinking and punish title level analysis. A candidate arguing about one film's marketing budget is arguing about $30 million inside a $740 million range.

Technology cases, and where the three businesses touch

The third letter has its own denominator, and it is retention. Near zero marginal cost means the profit on a retained customer is close to the revenue, so the length of the relationship carries the value. That arithmetic is worked end to end in software cases, including the ceiling a churn rate puts on the size of the business.

What belongs here instead is the place where all three sub sectors meet, because that is the TMT specific question an interviewer can only ask on this page. Convergence means an operator selling content, a studio selling subscriptions, and a software company selling network services. The arithmetic of those moves is a margin giveaway funded by a retention gain, and it can be priced.

Take the operator above. Broadband at $70 a month carries a 70 percent margin, which is $49. It bundles a streaming service that costs $12 a month wholesale and raises the package price to $75. Broadband cost is 30 percent of $70, which is $21, so total cost becomes $21 plus $12, which is $33, and margin becomes $75 minus $33, which is $42. The operator has given away $7 of monthly margin.

Price what that has to buy. At 1.5 percent monthly churn the customer life is 66.67 months, so lifetime margin without the bundle is 66.67 times $49, which is $3,267. For the bundle at $42 to match that, life has to reach $3,267 divided by $42, which is 77.78 months, and that means churn has to fall to one divided by 77.78, which is 1.286 percent.

So the bundle breaks even when churn falls from 1.50 to 1.286 percent, a reduction of 14.3 percent. That is the test to put to the client, and it is a far more useful statement than an argument about whether bundles improve loyalty. Below a 14.3 percent churn improvement the bundle destroys value, and above it the bundle pays.

How a TMT prompt is scored

The weights that follow are ours, pulled from the grading code in src/lib/grading, and nothing comparable is published by any consulting firm. The case for each weight is laid out on how an answer is assessed.