Q001 — ACH Matrix¶


Research	R0005 — AI Company Profitability Before 2030
Run	2026-03-17
Query	Q001 — Are any of the major AI companies expected to turn a profit before 2030?

Analysis of Competing Hypotheses¶

	H1: Yes, several	H2: No, none	H3: Depends on definition
Nvidia $120B net income FY2026	++	--	++
Alphabet profitable, $400B revenue	+	-	++
Microsoft profitable, AI revenue growing	+	-	++
Anthropic projecting FCF positive 2027-28	+	-	+
Anthropic revenue doubling in 2 months	+	-	+
OpenAI projecting $115B cumulative losses through 2029	-	+	+
OpenAI FCF positive 2029-2030 (own projection)	+	N/A	+
Big tech 2026 capex $650-700B	N/A	+	+
Alphabet FCF projected to drop 90% in 2026	-	+	+
Meta/Amazon projected negative FCF 2026	-	+	+
xAI burning $1B/month on $500M revenue	-	+	+
AI inference costs falling 99.7% since 2023	+	-	+
Consistency count	7+, 4-	4+, 7-	11+, 0-

Legend: ++ Strongly supports · + Supports · -- Strongly contradicts · - Contradicts · N/A Not applicable

Outcome¶

H3 is consistent with nearly all evidence. H1 is partially supported but overstates the case — it is only true if you count diversified tech giants as "AI companies" and accept corporate-level profitability as the standard. H2 is eliminated by the existence of already-profitable companies (Nvidia's $120B net income alone falsifies "no, none").

Diagnosticity Analysis¶

The most diagnostic evidence items — those that best differentiate between hypotheses — are:

Nvidia $120B net income FY2026 (SRC04-E01) — This single item eliminates H2 entirely. Any hypothesis claiming "none will be profitable" is falsified by Nvidia's SEC-filed earnings.
OpenAI projecting $115B cumulative losses through 2029 (SRC01-E03) — This contradicts H1's broad claim while supporting both H2 and H3. It shows that the largest pure-play AI lab by valuation is far from profitability.
Alphabet FCF projected to drop 90% in 2026 (SRC03-E01) and Meta/Amazon projected negative FCF 2026 (SRC03-E04, SRC03-E05) — These items expose the gap between corporate-level profitability and AI-specific profitability for diversified companies, strongly favoring H3's definitional framing.
Big tech 2026 capex $650-700B (SRC06-E02) — Not applicable to H1 (which focuses on revenue) but supports both H2 and H3, highlighting that massive investment is compressing margins even for profitable companies.

Evidence items that are least diagnostic (consistent with all hypotheses) are the Anthropic projections and declining inference costs, which support all three hypotheses to varying degrees.

Discriminating power: High between H2 and H1/H3 — Nvidia's profitability alone eliminates H2. Moderate between H1 and H3 — the distinction hinges on definitional scope (which companies count, which profitability metric), not on factual disagreement. H3 absorbs H1 as a special case.